CN1809166B - Image coding and decoding method and apparatus - Google Patents
Image coding and decoding method and apparatus Download PDFInfo
- Publication number
- CN1809166B CN1809166B CN2006100067204A CN200610006720A CN1809166B CN 1809166 B CN1809166 B CN 1809166B CN 2006100067204 A CN2006100067204 A CN 2006100067204A CN 200610006720 A CN200610006720 A CN 200610006720A CN 1809166 B CN1809166 B CN 1809166B
- Authority
- CN
- China
- Prior art keywords
- image
- section
- unit
- coding
- picture
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Lifetime
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/154—Measured or subjectively estimated visual quality after decoding, e.g. measurement of distortion
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/124—Quantisation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/13—Adaptive entropy coding, e.g. adaptive variable length coding [AVLC] or context adaptive binary arithmetic coding [CABAC]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
- H04N19/176—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a block, e.g. a macroblock
Abstract
The invention discloses an image coding method and a device which consider the human visual characteristics. The image coding device comprises a section modeling unit which segments the image into at least one separated rectangle section, an image header coding unit which uses the position and the size information of the section segmented by the section modeling unit, together with other information to code an image header and a section coding unit which codes the image in each section unit through referring to the header information of the image.
Description
The present invention is to be on January 5th, 2003 applying date, and application number is 03107502.9, and denomination of invention is divided an application for the application for a patent for invention of image coding and decoding method and the device of Human Visual System " consider ".
Technical field
The present invention relates to a kind of Image Coding, and more particularly, relate to a kind of method and apparatus of considering the image coding and decoding of Human Visual System.
Background technology
Generally speaking, for the storage/transmission purpose, image is compressed.Fig. 1 shows the flow chart of normal image compression method, and wherein, in order to generate the bit stream of compression, order is carried out space/time prediction coding (100), and transition coding (110) quantizes (120) and entropy coding (130).In this case, produce many losses in the process of quantization operation 120.This lossy compression method method comprises the lossy compression method method of rest image and the lossy compression method method of live image.JPEG is the lossy compression method method of representational rest image, and MPEG-1,2,4 and be H.261 and H.263 the lossy compression method method of representational live image.
Simultaneously, when being encoded, carries out image discrete cosine transform (DCT).In this case, because for whole image is carried out DCT, the quantity of calculating is too large, thus image segmentation is become the piece of pre-sizing, namely 8 * 8, and then encode.And, when carrying out quantification, if use the quantization parameter of each cell block image to be encoded the quantity increase of information so that the quantization parameter difference of each cell block.Therefore, use identical quantization parameter in whole image.At MPEG-4 with H.263 in system, for being adjusted quantization parameter by ± 2, each piece of 16 * 16 provides information.Then, use this information to reach accurate target bit rate.
When using such encoder, at the upper image with similar image quality that shows of whole scene (scene).Yet when the user watched image, it is more important than the picture quality of background area that the user thinks that (region of interest) picture quality (ROI) is distinguished in evaluation.Here it is, and why the user can see the reason that there are differences between the zone simultaneously.The user thinks more carefully to watch other detail section of estimating district (ROI) and skipping over background area.Especially, this phenomenon is more remarkable in live image.Therefore, when using less position that image is encoded, the raising of estimating district's (ROI) picture quality must be by for estimating district's (ROI) distribution ratio background area more position, rather than by whole image uniform distribution position is reached.
At MPEG-4 with H.263 in system, a part of image segmentation is become the zone and encodes.In the MPEG-4 system, the user can limit zone in pixel cell with the shape coding that surpasses median plane.Said method mainly is used in MPEG-4, because operation can be carried out in each object unit that consists of scene.Each object is encoded with different bit streams, and uses said structure can carry out user's reciprocation in MPEG-4.Make in this way, with the ROI of each object with background area is separated from one another makes it that different picture quality be arranged in order to image is encoded.Yet the separating treatment process of object is very complicated.Even use coarse shape with the mutual simple separation of object, also need in addition information when each object shapes of expression, therefore reduced compression ratio.
And, in system H.263, use annex K place piece cutting structure pattern a part of image segmentation to be become zone in macro block in groups (MB) unit of regional or a certain rectangular shape in groups continuous macro block (MB) unit, and can encode by image.This method that is used in system H.263 has very strong effect to error.Pith in using the environment of multi transmission channel by the transmission channel transmission in environment preferably in order to improve efficiency of transmission and stop during the error that occurs in a zone passes to another zone.In this case, ROI can encode with the piece cutting structure of rectangular shape.Yet, for the display background district, a part of image segmentation must be become several rectangles, therefore, H.263 the structure of system becomes complicated.
Be called in name in the United States Patent (USP) 5764803 of " Motion-adaptive Modeling Scene Content for very low Bit RateModel assisted Coding of Video Sequences (the movable adaptive model scene content of the auxiliaring coding of the ultralow bit rate model of video sequence) ", a part of image segmentation is become to estimate district (ROI) and background area, then image is encoded.Yet, being limited in scope of the quantization parameter that can change in each zone.Therefore, due in the difference of estimating the picture quality between district (ROI) and background area, can see the border of estimating between district (ROI) and background area.
And, be called in name in the United States Patent (USP) 6263022 of " System and Method for fine grannular scalable (FGS) Video with selective Quality Enhancement (system and method with fine granularity retractility (FGS) video that the selectivity quality strengthens) " and disclose a kind of compression method that is used in the multi transmission channel environment that comprises a basic unit and a reinforced layer.The method is fit to a kind of environment of transmission channel, but in carrying out, prediction is difficult, has therefore reduced code efficiency.And, although improved the picture quality of estimating district (ROI), but reduced the efficient of binary encoding.Therefore, greatly reduce the picture quality of background area.That is, the difference of estimating the picture quality between district and background area has increased, so the border occurred significantly between evaluation district and background area.
And, be called in name in the United States Patent (USP) 6256423 of " Intra-frame quantizer Selection for Video Compression (in the frame of video compression quantizer select) " and disclose a kind of compression method, wherein definition is estimated district (ROI) and background area, is being estimated the transition region between district (ROI) and background area and determining each quantization parameter between distinguishing.In the method, because transition region can stop the phenomenon that occurs on the border of estimating between district (ROI) and background area a little.Yet, being limited in scope of the quantization parameter that each is regional, and also need n transition region in a part of zone when existing n to estimate district (ROI), so the coding method complexity that will become.In addition, become smooth in order to make the border between the zone, another transition region that also need to be between transition region and another zone.As a result, determine that each regional quantization parameter is difficult.In order to address this problem, also used iteration to select the method for quantization parameter, but the method cause the increase of amount of calculation.
Summary of the invention
The present invention proposes a kind of method and apparatus of image coding and decoding, its stoped estimate that border between district (ROI) and background area forms in image and the method by the less amount of calculation of consideration Human Visual System needs.
The present invention further provides a kind of method and apparatus of image coding and decoding, by the method and device, when image is carried out Code And Decode, effectively carried out estimating the Code And Decode in district (ROI) by using a plurality of rectangular areas.
The present invention further provides a kind of recording medium, on this recording medium, the method for image coding and decoding is recorded as by the executable program code of computer.
According to an aspect of the present invention, provide a kind of method for encoding images.This method for encoding images comprises that (a) distributes to the picture quality of input picture in the scene unit and carry out modeling so that the picture quality of inputting reduces to background area gradually from estimating the district in the scene unit, (b) modeling result that distributes according to picture quality determines to consist of each regional quantization parameter of scene, (c) according to quantization parameter quantized image data, and (d) entropy of the view data that quantizes is encoded.
According to a further aspect in the invention, provide a kind of picture coding device.This picture coding device comprises: a picture quality modeling unit, be used for the picture quality at scene unit input picture is distributed and carry out modeling so that the picture quality of inputting reduces to background area gradually from estimating the district in the scene unit, and according to the modeling result that picture quality distributes, determine to consist of each regional quantization parameter of scene; One adaptive quantizing unit is used for the quantization parameter quantized image data of determining according to by the picture quality modeling unit; With an entropy coding unit, be used for the entropy of the view data that quantizes by the adaptive quantizing unit is encoded.
According to a further aspect in the invention, provide a kind of picture decoding method.this picture decoding method comprises (a) decode image data, this view data comprises the value of position that each is regional and size information and the quantization parameter in receives bit stream, (b) use the information of each regional position and size and the value of the quantization parameter that recovers is determined the value of the quantization parameter that each is regional and reduced gradually to background area in order to picture quality is distinguished from evaluation in (a), (c) use the value of quantization parameter to carry out re-quantization to the view data of decoding, the relevant position that (d) conforms to according to the value with each regional position of recovering in (a), for each zone increases the image that recovers and builds a scene.
According to a further aspect in the invention, provide a kind of picture decoding apparatus.This picture decoding apparatus comprises: an entropy decoding unit, be used for the entropy of view data is decoded, and this view data comprises position that each is regional and the value of size information and the quantization parameter in receiving bit stream; One picture quality modeling unit is used each regional position and size information and the value of the quantization parameter that recovers by the entropy decoding unit is determined the value of the quantization parameter that each is regional, in order to picture quality is reduced to background area gradually from estimating the district; One adaptive inversion quantifying unit according to the value of each regional quantization parameter of being determined by the picture quality modeling unit, is carried out re-quantization to the view data that provides by the entropy decoding unit; With a picture construction unit, the relevant position that conforms to according to the value with each the regional position that provides by the entropy decoding unit is for each zone increases Recovery image and builds a scene.
According to a further aspect in the invention, provide a kind of picture coding device.This picture coding device comprises: a section modeling unit becomes at least one independently to cut into slices each image segmentation; One picture headers coding unit will be encoded to a picture headers together with other information by position and the size information of the section cut apart of section modeling unit; With all slice encodes unit, by the reference picture header message, the image in the section unit is encoded.
According to a further aspect in the invention, provide a kind of picture decoding apparatus.This picture decoding apparatus comprises: a picture headers decoding unit is used for the picture headers at bit stream is decoded; One section construction unit uses the position of the section in picture headers information and size information to build section; One section decoding unit by the reference picture stem, is decoded to the image of the unit of cutting into slices; With a picture construction unit, by with reference to the position of the section that is obtained by the section construction unit and the information of size, the picture construction in the section unit that will recover by the slice decoder unit is the part of image.
According to a further aspect in the invention, provide a kind of method for encoding images.This method for encoding images comprises that (a) arranges position and the size of section in a part of image, position and the size information of the section that (b) will arrange in (a) are encoded to a picture headers together with other information, (c) by the picture headers information with reference to coding in (b), in the unit of cutting into slices, image is encoded.
According to a further aspect in the invention, provide a kind of picture decoding method.This picture decoding method comprises that (a) decodes to the picture headers in bit stream, (b) use the relevant position and the size information that are included in the section of the picture headers of decoding in (a) to build section, (c) by the picture headers with reference to decoding in (a), in the unit of cutting into slices, image is decoded, the information of the relevant position of (d) cutting into slices by reference and size, the picture construction in the section unit that will decode in (c) is a part of image.
According to a further aspect in the invention, provide a kind of method for encoding images, wherein image segmentation is become have the section unit of pre-sizing and it is encoded.This method for encoding images comprises: limit the relevant position of a plurality of rectangular areas and the information of size, wherein will estimate to distinguish and be included in image and larger rectangle comprises less rectangle; All sections that are included in interior rectangular area minimum in a plurality of rectangular areas are encoded; Do not encoded with the overlapping section of less interior rectangle in the outer rectangular area of getting rid of interior rectangular area; Be defined as background area with the zone that will not be included in the outer rectangular area of ragged edge and the section that is included in background area is encoded.
According to a further aspect in the invention, a kind of picture decoding method is provided, wherein image segmentation is become have the section unit of pre-sizing and it is encoded, image segmentation is become background area and estimates the district, and image is decoded in bit stream, this bit stream limits by a plurality of rectangular areas, wherein will estimate the district and be included in image and larger rectangle comprises less rectangle.This picture decoding method comprises: each position and the size information to a plurality of rectangular areas of comprising the evaluation district in bit stream extracted; All sections in the interior rectangular area that is included in the minimum in a plurality of rectangular areas are decoded; Only decode in section get rid of in the outer rectangle of rectangle less than the section overlapping with less rectangle; With all sections that are not included in the background area in a plurality of rectangular areas are decoded.
Description of drawings
By reference to the accompanying drawings, by detailed description of the preferred embodiment, above-mentioned and further feature of the present invention and advantage will become more obvious, wherein:
Fig. 1 shows the flow chart of conventional method for compressing image;
Fig. 2 has schematically shown the block diagram of considering according to an embodiment of the invention the picture coding device of Human Visual System;
Fig. 3 shows according to one embodiment of the invention the flow chart of carries out image coding method in the picture coding device in Fig. 2;
Fig. 4 A and 4B show and determine that in a scene one estimates the step of district's (ROI), an insert district and background area;
Fig. 5 A shows to Fig. 5 C the curve chart that the distribution with the distribution of the picture quality of prior art and picture quality of the present invention compares;
Fig. 6 A shows the feature of determining quantization parameter by the picture quality modeling unit in Fig. 2 to 6C;
The variation that Fig. 7 A and 7B show in linearity or non-linear quantization according to quantization parameter represents the quantity of position and the relation of picture quality;
Fig. 8 A shows the comparison that method for encoding images according to the present invention is applied to the situation in real image and conventional coding method is applied to the situation in real image to 8C;
Fig. 9 has schematically shown the block diagram of considering according to an embodiment of the invention the picture decoding apparatus of Human Visual System;
Figure 10 shows according to the flow chart that should invent carries out image coding/decoding method in the picture decoding apparatus of an embodiment in Fig. 8;
Figure 11 shows the block diagram of picture coding device according to an embodiment of the invention;
Figure 12 shows the flow chart of carries out image coding method in picture coding device in Figure 11;
Figure 13 has schematically shown the block diagram of picture decoding apparatus in accordance with another embodiment of the present invention;
Figure 14 shows the flow chart of carries out image coding/decoding method in picture decoding apparatus in Figure 13;
Figure 15 A and 15B show the method for processing section according to the present invention;
Figure 16 shows the comparison that coding method according to the present invention is applied to the situation in real image and conventional coding method is applied to the subjective picture quality of the situation in real image; And
Figure 17 shows the comparison that coding method according to the present invention is applied to the situation in real image and conventional coding method is applied to the objective image quality of the situation in real image.
Embodiment
Hereinafter, describe with reference to the accompanying drawings image coding and decoding method and device thereof according to consideration Human Visual System of the present invention in detail.
Fig. 2 has schematically shown the block diagram of considering according to an embodiment of the invention the picture coding device of Human Visual System.With reference to Fig. 2, picture coding device comprises 150, one adaptive quantizing unit 155, picture quality modeling unit 140, one predict coding unit 145, one a transition coding unit, and an entropy coding unit 160.
Picture quality modeling unit 140 is set up the picture quality distributed model of input picture so that the picture quality of the image of inputting reduces to background area gradually from estimating district (ROI) in the scene unit in the scene unit.And the modeling result that picture quality modeling unit 140 distributes according to picture quality determines to consist of each regional quantization parameter of a scene.
The image of 145 pairs of inputs of predict coding unit carries out predictive coding, and the input picture of the 150 pairs of predictive codings in transition coding unit carries out transition coding.In this case, calculate in order to simplify coding, coding is carried out in predict coding unit 145 and transition coding unit 150 in having the module unit of pre-sizing.
Adaptive quantizing unit 155 quantizes according to the view data of the quantization parameter of being determined by picture quality modeling unit 140 to each MB.
160 pairs of entropys by the view data of adaptive quantizing unit 155 quantifications of entropy coding unit are encoded.
Fig. 3 shows according to one embodiment of the invention the flow chart of carries out image coding method in the picture coding device in Fig. 2.
The operation of the picture coding device in Fig. 2 is described with reference to Fig. 2 and 3 now.
At first, in step 200,140 pairs of the picture quality modeling unit input picture in the scene unit carries out modeling that picture quality distributes so that the picture quality of the image of inputting reduces to background area gradually from estimating district (ROI) in the scene unit.After step 200, in step 210, the modeling result that picture quality modeling unit 140 distributes according to picture quality, namely the quantization parameter of 8 (pixel) * 8 or each piece of 16 * 16 and the quantization parameter that will determine offer adaptive quantizing unit 155 to determine to consist of the pre-sizing that has of a scene.
The picture quality of input picture reduces to background area gradually from the central area of estimating district (ROI) because 140 pairs of picture qualities of picture quality modeling unit are carried out modeling, so quantization parameter is little in estimating district (ROI) and increases towards background area gradually.Quantization parameter is marked (bit target) by the position and is determined.Here, picture quality modeling unit 140 can be carried out modeling so that the picture quality distribution has Gaussian Profile to picture quality.In addition, picture quality modeling unit 140 can arrange a predetermined insert district between the evaluation district of each scene (ROI) and background area, and can distribute to picture quality and carry out modeling in order to obtain best picture quality at the center of estimating district (ROI) and the picture quality that each is regional will reduce at background area, but the difference in background area is inapparent in the insert district.For example, picture quality modeling unit 140 can determine to estimate district (ROI) 250 and background area 260 in a scene, as shown in Fig. 4 A, and insert district 300 can be set remove the piece phenomenon, this piece phenomenon is that the difference by the picture quality between the zone in the zone that connects each other causes, as shown in Figure 4 B.Picture quality modeling unit 140 is determined quantization parameter in order to picture quality is connected naturally by picture quality being carried out modeling in the insert district.Quantization parameter in the insert district can linearity or nonlinear change.The modeling of picture quality is described in detail with reference to Fig. 5 and 6.
After step 210, in step 220, adaptive quantizing unit 155 each regional predictive coding of input and transition coding view data, and picture quality modeling unit 140 quantizes each regional view data according to the quantization parameter of being determined by picture quality modeling unit 140.Simultaneously, the quantization parameter that provides by picture quality modeling unit 140 increases to background area from the evaluation district (ROI) of scene, and the extent of damage that therefore causes by quantification is different.That is, minimum loss is at the center of estimating district (ROI), estimates the most attracting eyes of Qu Zhongshi at this, and loss increases towards background area, in this background area, and attracting eyes relatively less.
After step 220, in step 230, the entropys of the view data that 160 pairs of entropy coding units quantize by adaptive quantizing unit 155 encode and the entropy of the coding of output image data as bit stream.
As mentioned above, by considering that Human Visual System carries out modeling to the predetermined image quality, quantization parameter increases to background area gradually from estimating district (ROI), therefore effectively removes in interregional piece phenomenon, has improved simultaneously code efficiency.
Fig. 5 A shows to 5C the curve chart that the distribution with the distribution of the picture quality of prior art and picture quality of the present invention compares.In 5C, X-axis and Y-axis represent respectively the spatial domain of each image at Fig. 5 A, and Z axis representative image quality.And the evaluation district (ROI) of supposition image is the mid portion of scene.
Fig. 5 A shows in 140 pairs of scenes of picture quality modeling unit picture quality and distributes and carry out modeling, makes the situation of the picture quality distribution that has to estimate the Gaussian Profile centered by district (ROI).With reference to Fig. 5 A, the center that picture quality is distributed in image on average has the bidimensional Gaussian Profile, and wherein best picture quality is to obtain in estimating district (ROI) and picture quality will reduction gradually in background area.In this case, the slope that changes in picture quality can be adjusted according to discrete (dispersion) value of X and Y-axis.Can form different picture quality according to the quantity of estimating district (ROI) distributes.In addition, in this case, the quantization parameter that causes being distributed by picture quality in a scene distributes and carries out modeling and transmit its model, rather than the quantization parameter of each cell block is encoded and transmitted this quantization parameter.For example, be transmitted if be equivalent to estimate position and the centrifugal pump of cell block at the center in district (ROI), the quantization parameter of all cell blocks can calculate in receiver automatically.
Fig. 5 B shows and places the insert district in order to reduce the situation of quick difference in picture quality between evaluation district (ROI) and background area.In this case, a quantization parameter is distributed to each zone.The interval that quantization parameter changes arranges according to the size in zone.Do not need by identical rule is set in encoder, additional information to be inserted in the interval.For example, in the situation that rectangular area, if will be equivalent to the region division of each width and length 20% is an interval, in this interval, the parameter that quantizes changes, the quantization parameter of variation can be applied to be equivalent to from the zone of 20% piece of the outside in each zone and it is encoded.Even when decoding during quantization parameter, the quantization parameter that changes is applied to is equivalent to 20% part, can correctly encode to the quantization parameter that changes like this.And by use identical quantization method in encoder, in the situation that there is no to insert the relevant additional information that changes the method for quantization parameter, quantization parameter can change according to default rule.
Fig. 5 C shows the distribution that scene in prior art is divided into the picture quality of the situation of estimating district (ROI) and background area.In this case, due to the difference of the picture quality between two zones, occur in two borders between the zone and the piece phenomenon occurs, therefore destroyed picture quality.
At last, with reference to Fig. 5 A and 5B, picture quality reduces to background area gradually from estimating district (ROI), and therefore the user can not feel in the difference of estimating the picture quality between district (ROI) and background area.Yet as shown in Fig. 5 C, due in the quick difference of estimating the picture quality between district (ROI) and background area, the user feels the piece phenomenon.
Fig. 6 A shows the feature of determining quantization parameter by the picture quality modeling unit 140 in Fig. 2 to 6C.For simplicity, suppose that the 0-2 of spatial domain and 8-10 are set to background area, the 2-8 of spatial domain is set to estimate district (ROI), and quantization parameter changes from 0-31.
Fig. 6 A shows the situation of quantization parameter linear change in the insertion zone of Fig. 4 B or 5B, and Fig. 6 B shows the situation of quantization parameter nonlinear change in the insertion zone of Fig. 4 B or 5B.With reference to Fig. 6 A and 6B, quantization parameter is determined in order to be not to change fast but gradually change in estimating insert district between district (ROI) and background area in the difference of estimating the picture quality between district (ROI) and background area.Fig. 6 C shows in the situation that there is no conventional insert district when the feature that scene is divided into quantization parameter when estimating district (ROI) and background area, and in Fig. 6 C, is fast estimating the difference of distinguishing the quantization parameter between (ROI) and background area.Therefore, the piece phenomenon may appear at the boundary of estimating between district (ROI) and background area.
Simultaneously, the linear/non-linear feature of quantization parameter can be determined by the feature of quantization method, as shown in Fig. 6 A and 6B.That is, determine that according to the variable in quantization parameter quantity and the picture quality of position are linear change or nonlinear change.
Fig. 7 A and 7B show in linearity or non-linear quantization the variation according to quantization parameter, the quantity and the relation of picture quality of expression position, and in Fig. 7 A and 7B use MPEG-4 and encoder H.26L.
Fig. 7 A shows the variation of the picture quality that changes according to quantization parameter, and show the MPEG-4 encoder have nonlinear characteristic and H.26L encoder have linear character.In addition, Fig. 7 B shows the variation in the bit quantity that changes according to quantization parameter, and show the MPEG-4 encoder have nonlinear characteristic and H.26L encoder have linear character.That is, the method for linear or non-linear change quantization parameter makes by quantization method to be determined, does not therefore need additional information transmission.And the quantity of position reduces in estimating district (ROI), and the quantity of position increases towards background area.Therefore, all quantity of position do not have large variation and do not need additional calculations.
Fig. 8 A shows the comparison that coding method according to the present invention is applied to the situation in real image and conventional coding method is applied to the situation in real image to 8C.Fig. 8 A shows situation about using according to adaptive quantizing method of the present invention, and Fig. 8 B shows at the mid portion of image and forms rectangle evaluation district (ROI), the Region Segmentation of scene is become two zones, and the situation of in each zone, different quantization parameters being encoded.And Fig. 8 C shows in the situation that there is no segmentation evaluation district (ROI), the situation of in all pieces of scene, identical quantization parameter being encoded.
To 8C, the subjective picture quality of image is improved by the coding in in-service evaluation district (ROI) with reference to Fig. 8 A, as shown in Fig. 8 B, rather than uses identical quantization parameter in all pieces, as shown in Fig. 8 C.Yet, formed the border between ROI and background area.Due to the border phenomenon, limited estimate district (ROI) thereby and the generation difference of the picture quality between background area limited the efficient of coding ROI.Yet, in the situation of the self-adaptive encoding method in Fig. 8 A, between ROI and background area, the insert district being set, picture quality gradually changes in the insert district so that marked change does not appear in the difference of the picture quality between two insert districts.
Fig. 9 has schematically shown the block diagram of considering according to an embodiment of the invention the picture decoding apparatus of Human Visual System.With reference to Fig. 9, picture decoding apparatus comprises picture quality modeling unit 300, one entropy decoding unit 310, one adaptive inversion quantifying unit 320, one image restoration unit 330 and a picture construction unit 340.
Figure 10 shows according to one embodiment of the invention the flow chart of carries out image coding/decoding method in the picture decoding apparatus in Fig. 8.With reference to Fig. 9 and 10, entropy decoding unit 310 receives a bit stream and the bit stream that belongs to each regional data is decoded.In step 410, in bit stream, the view data that comprises each zone position information, the value of quantization parameter and the size in each zone are decoded.
In step 420, each regional position that picture quality modeling unit 300 use are decoded by entropy decoding unit 310 and the quantization parameter value of size information and data are determined the value of the quantization parameter of respective regions.Picture quality modeling unit 300 offers adaptive quantizing unit 320 with the value of quantization parameter.In this case, the value of quantization parameter can be set to gradually change at the predetermined space place so that the value of the quantization parameter between each zone does not change fast.As above described to 6C to 5C and 6A with reference to Fig. 5 A, picture quality distributes and can be modeled as the Gaussian Profile that has to estimate centered by district (ROI), or can put into an insert district between ROI and background area in case minimizing in the quick variation of picture quality.Like this, carry out in above-mentioned coding method an interval is set, wherein the value by each definite regional quantization parameter of picture quality modeling unit 300 changes, or carries out the change to a quantization parameter.
In step 430, use the value of the quantization parameter of being determined by picture quality modeling unit 300, adaptive inversion quantifying unit 320 re-quantizations are by data or the data f of each piece of entropy decoding unit 310 transmission.
In step 440,330 pairs of image restoration unit have each piece of pre-sizing and carry out inverse transformation, compensation prediction information and Recovery image.
In step 450, picture construction unit 340 is increased to by the Recovery image that each is regional and builds a scene in an image part according to the relevant position that conforms to the information of the relevant position in each zone that is provided by entropy decoding unit 310.
Above-described structure according to adapting to image decoding device of the present invention and operation are used for decoding at the image of encoding by the adapting to image code device of Fig. 2.This operation is the operation in reverse execution adapting to image code device, and as the execution setting in coding method wherein the value of the quantization parameter of each piece be an interval that changes, or change the quantization parameter of carrying out in the adapting to image decoding device.Like this, for simplicity, these operations will no longer here repeat.
Figure 11 shows the block diagram of picture coding device according to an embodiment of the invention.With reference to Figure 11, picture coding device according to the present invention comprises section modeling unit 1100, one picture headers coding units 1200 and all slice encodes unit 1300.Section coding unit 1300 comprises one space/time prediction coded portion 1310, one transform and quantization parts 1320, and an entropy coded portion 1330.
1200 pairs of communications that the required public information of all sections of decoding is encoded and will be encoded in image of picture headers coding unit are in section coding unit 1300.In this case, the quantity of section, shape, position and size all are included in transmission information.
By the picture headers information with reference to input from image header coding unit 1200, section coding unit 1300 is encoded to image in the unit of cutting into slices.For this reason, space/time prediction coded portion 1310 removes the room and time overlay information.For example DCT of predetermined conversion is carried out in the output of 1320 pairs of spaces of transform and quantization part/time prediction coded portion 1310, and quantizes a transformation parameter.The entropy of the output of 1330 pairs of these transform and quantization parts 1320 of entropy coded portion is encoded and and is produced a compression position flow.
When by network, image being encoded and transmitting, section coding unit 1300 is with image segmentation one-tenth section unit.Especially, with image segmentation orthogonal section unit and it is encoded, like this, reduced prediction loss in section.And, use to be divided in rectangle and the piece cutting structure of exterior domain, carry out ROI coding and picture-in-picture (PIP) coding.Here, in the ROI coding, image segmentation is become to estimate district (ROI) and background area, the picture quality of ROI has improved, and the picture quality of background area has reduced in order to use strict bit rate to improve subjective picture quality.In the PIP coding, can carry out the independence decoding in order to utilize this part as other images to the part that comprises the rectangle section.
According to the present invention, when the rectangular area that comprises a plurality of sections and another rectangular area when overlapping, be set to the insert district so that in the ROI encoding operation in the presumptive area between overlapping region and background area, quick variation due to the picture quality between ROI and background area has stoped the reduction of subjective picture quality.In addition, in PIP encoding operation process, can use the PIP with each size.
Figure 12 shows the flow chart of the method for encoding images of carrying out in picture coding device in Figure 11.With reference to Figure 12, if position and the size of the rectangular area of independent process are set, in step 1400, section modeling unit 1100 is divided at least one individual slices with corresponding rectangular area.In step 1500,1200 pairs of picture headers of picture headers coding unit are encoded, and in step 1600, the coding of section are carried out by section coding unit 1300.
Here, section coding (step 1600) comprises space/time prediction coding, to remove overlapping information (step 1610) on the room and time that is present in the image in the unit of cutting into slices, carry out predetermined conversion, DCT for example, and the data of removing overlay information are quantized (step 1620), and the entropy of quantized data is encoded and generated a bit stream (step 1630) that compresses.
Figure 13 has schematically shown the block diagram of picture decoding apparatus in accordance with another embodiment of the present invention.With reference to Figure 13, picture decoding apparatus according to the present invention comprises picture headers decoding unit 2000, section construction units 2100, one section decoding units 2200 and a picture construction unit 2300.This slice decoder unit 2200 comprises entropy decoded portion 2210, one re-quantizations and an inverse transformation part 2220, and an image recovered part 2230.
Picture headers information in the bit stream that 2000 pairs of picture headers decoding units receive by network is decoded.Quantity, shape, position and size information about the section of the rectangular area of each decoding are transferred in section construction unit 2100, and with other communication in slice decoder unit 2200.
Response is by quantity, shape, position and the size information of the section of the rectangular area of each decoding of picture headers decoding unit 2000 transmission, and section construction unit 2100 is selected the position of section, processes the lap of rectangular area and builds section.The processing of the lap of rectangular area will be described with reference to Figure 15 below.
By the picture headers information of reference from 2000 inputs of picture headers decoding unit, the image in the unit of cutting into slices, 2200 pairs of unit of slice decoder is decoded.For this reason, the entropy of 2210 pairs of bit streams of entropy decoded portion is decoded, and the bit stream of re-quantization and 2220 pairs of entropy decodings of inverse transformation part is carried out re-quantization and inverse transformation.And the space of the output data of image recovered part 2230 compensation re-quantizations and inverse conversion part 2220/time prediction coded message is also recovered this image.In this case, response is by the information of section construction unit 2100 input, and the image that will recover in the section unit is added in a part of image 2400 that builds by picture construction unit 2300.
The picture decoding apparatus that will have according to the present invention said structure is used for decoding by the image of the coding of the picture coding device in Figure 11.The operation of picture decoding apparatus and the operation of picture coding device are oppositely carried out.Yet the essential characteristic of the slicing treatment operation of picture decoding apparatus is identical with picture coding device.Therefore, for for purpose of brevity, these operations will no longer here repeat.
Figure 14 shows the flow chart of the picture decoding method of carrying out in the picture decoding apparatus of Figure 13.With reference to Figure 14, in step 2500, picture headers information is decoded in the bit stream that receives.In this case, in step 2600, about the information exchange of quantity, shape, position and the size of the section of the rectangular area of each decoding is crossed 2100 transmission of section construction unit, and the information of response input, section construction unit 2100 is selected the position of section, processes the lap of rectangular area and builds section.In step 2700, carry out decoding in the section that builds, and in step 2800, the position of cutting into slices by reference and size and design of graphics picture.
Here, slice decoder (step 2700) comprises the entropy of bit stream decode (step 2710), data to entropy decoding are carried out re-quantization and inverse conversion (step 2720), and to through the compensation data space of inverse transformation/time encoding information (step 2730).
Figure 15 A and 15B show the method for processing section according to the present invention.With reference to Figure 15 A, the part of image 500 comprises two rectangular areas 502 and 503 and one background area 504.These two rectangular areas 502 and 503 overlap each other.And, two rectangular areas 502 and 503 and background area 504 formed by a plurality of independently sections.
Less rectangular area 502 in two rectangular areas 502 and 503 is included in larger rectangular area 503 fully.In this case, larger rectangular area 503 expressions are a part overlapping with less rectangular area (hypographous part) 502 not, gets rid of less rectangular area 502 (zone of getting rid of shaded portion from larger rectangular area 503) from this part.In this case, less rectangular area 502 comprises that the user thinks and namely estimates district (ROI) in more important zone in the picture of image 500, and rectangular area 503 that will be larger is as the insert district that is placed between ROI 501 and background area 504.Here, comprise the less rectangular area of ROI 501 and as the larger rectangular area 503 of insert district, with top-left position information 520 and 510 and bottom right positional information 521 and 511 represent position and the big or small information of each rectangular area.As mentioned above, these two rectangular areas 502 and 503 and background area 504 comprise a plurality of independently sections and it carried out Code And Decode in section in the unit.And, when when being illustrated in image in Figure 15 A and decoding and encode, preferably, according to less rectangular area 502, than the order of large rectangle zone 503 and background area 504, image is encoded, and can come identified region by the order-assigned regional number according to the rectangular area of coding.For example, as shown in Figure 15 A, can be with the rectangular area 502 of coding at first of number distributing to of regional id=0, and with regional id=1 number distribute to the rectangular area 503 of next coding, and id=2 is distributed to background area 504.
Similarly, between ROI 501 and background area 504, insert district 503 is set, in order to effectively reduce the appearance on the border between ROI and background area.
Here, can independently encode to a plurality of sections that consist of less rectangular area 502.Therefore, when not needing whole image is decoded, only formation is comprised that the section of the less rectangular area 502 of ROI encodes.In this case, the image that recovers from less rectangular area 502 becomes picture-in-picture (PIP).The image that will recover from larger rectangular area 503 is increased in the image that recovers from less rectangular area 502, and consists of another larger PIP.Therefore PIP can build step by step as the rectangular area in a part of image number, can represent to larger image from less image.
In Figure 15 B, the part of image 600 comprises three rectangular areas 603,604 and 605 and background area 606.Two rectangular areas 603 and 604 do not overlap each other and overlap in another rectangular area 605.And, these three rectangular areas 603,604 and 605 and background area 606 comprise a plurality of independently the section.
With reference to Figure 15 B, each in two rectangular areas 603 and 604 comprises different ROI 601 and 602, and these two different ROI are that the user thinks more important zone in the section of image 600.Here, comprise ROI 601 and 602 less rectangular area 603 and 604 and as larger rectangular area 605 use the top-left position information 630,620 of insert district and 610 and bottom right positional information 631,621 and 611 represent position and big or small information about each rectangular area.As mentioned above, these three rectangular areas 603,604 and 605 and background area 606 form and it carried out Code And Decode in section in the unit by a plurality of independently sections.And, when when being illustrated in image in Figure 15 B and decoding and encode, preferably, according to less rectangular area 603 and 604, than the order of large rectangle zone 605 and background area 606 to Image Coding, and come identified region by the order-assigned regional number according to the rectangular area of coding.For example, as shown in Figure 15 B, can be with the rectangular area 603 of coding at first of number distributing to of regional id=0, and the rectangular area 604 of number distributing to next coding with regional id=1, with the larger rectangular area 605 of number distributing to next coding of regional id=3, and id=4 is distributed to background area 606.
Similarly, ROI 601 and 602 and background area 606 between insert district 605 is set, in order to effectively reduce the appearance on the border between ROI and background area.
Like this, ROI coding or PIP coding are by using piece cutting structure effectively to carry out, and wherein the inside of rectangular area is ROI, and the outside of rectangular area is the background area.Especially, as mentioned above, several rectangular areas are overlapped each other, in order to reduce the border between the zone and the PIP that can support to have various sizes.In addition, the section that comprises ROI is encoded to make error is had more pretend with in order to can obtain subjective picture quality preferably in having the transmission environment of error.
Figure 16 shows the comparison that coding method according to the present invention is applied to the situation in real image and conventional coding method is applied to the subjective picture quality of the situation in real image.With reference to Figure 16, left column represents to use the subjective picture quality of conventional coding method, and the subjective picture quality that uses according to coding method of the present invention is shown in right list, shows the result that 20% grouping has loss rate.In Figure 16, attention be improved estimate district (ROI) thus picture quality improved whole subjective picture quality.And ROI more is protected and avoids error in order to improve whole subjective picture quality.
Figure 17 shows the comparison that coding method according to the present invention is applied to the situation in real image and conventional coding method is applied to the objective image quality of the situation in real image.With reference to Figure 17, left column is illustrated in the PSNR in whole image, and right list is shown in the PSNR that estimates in district (ROI).In Figure 17, it should be noted use according in the whole image of coding method of the present invention and the PSNR in ROI than using conventional coding method to improve.
In addition, the present invention's code of can be used as the embodied on computer readable on the recording medium of embodied on computer readable is realized.The recording medium of embodied on computer readable comprises various tape decks, can store the data of embodied on computer readable on these tape decks.The recording medium of embodied on computer readable comprises ROM, RAM, CD-ROM, tape, floppy disc and optical data storage device, and further comprises carrier wave (namely transmitting on the internet).The storage medium of embodied on computer readable is arranged on computer system connected to the network, therefore can stores the code that can read with object computer in the mode of disperseing.
As mentioned above, method for encoding images and device according to consideration Human Visual System of the present invention, scene is divided into estimates district (ROI) and background area, and carry out modeling that picture quality is distributed so that the difference of the picture quality between ROI and background area is not fast, improved thus picture quality.In addition, the ROI coding can be carried out effectively by using the rectangular area, and Code And Decode is independently carried out in each zone in the unit of cutting into slices.Especially, the rectangular area is overlapping in order to effectively prevented the appearance on the border between ROI and background area each other, has improved thus subjective picture quality.Further, do not need each regional quantization parameter of iteration adjustment, in order to meet the determined number of giving of position, therefore do not need larger amount of calculation.
Although illustrated and illustrated the present invention with reference to the preferred embodiment of the present invention, but be understandable that, in the situation that do not deviate from the spirit and scope of the invention by the appended claims, those of ordinary skills can make various changes to the present invention in form and details.
Claims (22)
1. picture coding device comprises:
One section modeling unit, the part of each image is assigned at least one independently rectangle section, each image comprises at least two overlapped rectangular areas, and wherein larger rectangular area comprises less rectangular area, and non-overlapping copies between less rectangular area;
One picture headers coding unit will be about the information coding of the position of the section of appointment and size in picture headers; With
By the reference picture header message, encode to the image in each section unit in all slice encodes unit,
Wherein, described section modeling unit is configured to each rectangular area is configured at least one independently rectangle section, and the remaining area that is not included in each image in described at least two rectangular areas is assigned at least one independently rectangle section.
2. according to claim 1 device, wherein said section coding unit is configured to the section in the part of the larger rectangle of getting rid of less rectangular area is encoded.
3. according to claim 1 device, wherein this section coding unit comprises:
One space/time prediction coded portion is carried out space/time prediction coding to the image in the unit of cutting into slices;
One transform and quantization part is to the transform and quantization of being scheduled to by the information and executing of space/time prediction coded portion predictive coding; With
One entropy coded portion is encoded to the entropy of the information that partly obtains by transform and quantization.
4. the regional number that according to claim 1 device, wherein said section coding unit distribute based on the coding order according to section is encoded to the image in the section unit.
5. picture decoding apparatus comprises:
One picture headers decoding unit is decoded to the picture headers in a bit stream;
One section construction unit uses in picture headers information about the position of section and the Information generation section of size; With
One section decoding unit by the reference picture stem, is decoded to the image of the unit of cutting into slices,
Wherein, described section construction unit is configured to utilize described picture headers information to build at least two overlapped rectangular areas in image, wherein larger rectangular area comprises less rectangular area, and non-overlapping copies between less rectangular area, each rectangular area is configured at least one independently rectangle section, and the remaining area that is not included in image in described at least two rectangular areas is assigned at least one independently rectangle section.
6. according to claim 5 device, wherein said slice decoder unit are configured to the section in the part of the larger rectangle of getting rid of less rectangular area is decoded.
7. according to claim 5 device, wherein this slice decoder unit comprises:
One entropy decoded portion is used for the entropy of bit stream is decoded;
One re-quantization and inverse transformation part are to information and executing re-quantization and the inverse transformation by the decoding of entropy decoded portion; With
One image recovered part is carried out space/time prediction to the information that partly obtains by re-quantization and inverse transformation and is compensated and recover at the image of section in the unit.
8. the regional number that according to claim 5 device, wherein said slice decoder unit distribute based on the decoding order according to section is decoded to the image in the section unit.
9. the method for an Image Coding comprises:
(a) position and the size of cutting into slices in a part of image is set, this image comprises at least two overlapped rectangular areas, and wherein larger rectangular area comprises less rectangular area, and non-overlapping copies between less rectangular area;
(b) will arrive picture headers about the position of section and the information coding of size; With
(c) by the reference picture header message, the image in the unit of cutting into slices is encoded,
Wherein, arranging of slice position and size, each rectangular area is built at least one independently in rectangle section, and the remaining area that is not included in described image in described at least two rectangular areas is built as at least one independently rectangle section.
10. according to claim 9 method, wherein said coding to the image in the section unit comprise encodes to the section in the part of the larger rectangle of getting rid of less rectangular area.
11. method according to claim 9, wherein (c) comprising:
(c1) image in the unit of cutting into slices is carried out space/time prediction coding;
(c2) to information and executing predetermined map and the quantification of predictive coding in (c1);
(c3) entropy of the information of acquisition in (c2) is encoded.
12. method according to claim 9, wherein the coding of image comprises based on according to the regional number of the coding order distribution of cutting into slices, the image in the unit of cutting into slices being encoded.
13. a picture decoding method comprises:
(a) picture headers in a bit stream is decoded;
(b) specify section with position and the big or small information about the section that comprise picture headers; With
The picture headers of (c) decoding by reference is decoded to the image in the unit of cutting into slices,
Wherein, carry out described appointment section in order to utilize described picture headers information to build at least two overlapped rectangular areas in image, wherein larger rectangular area comprises less rectangular area, and non-overlapping copies between less rectangular area, each rectangular area is configured at least one independently rectangle section, and the remaining area that is not included in described image in described at least two rectangular areas is configured at least one independently rectangle section.
14. method according to claim 13, wherein said decoding to the image in the unit of cutting into slices comprise, the section in the part of the larger rectangle of getting rid of less rectangular area is decoded.
15. method according to claim 13, wherein (c) comprising:
(c1) entropy in bit stream is decoded;
(c2) information and executing re-quantization and the inverse transformation to decoding in (c1); With
(c3) information that obtains in (c2) being carried out space/time prediction compensate and recovers image in the unit of cutting into slices.
16. method according to claim 13, wherein the decoding of image comprises based on according to the regional number of the decoding order distribution of cutting into slices, the image in the unit of cutting into slices sequentially being decoded.
17. a method for encoding images wherein becomes have the section unit of pre-sizing and it is encoded with an image segmentation, the method comprises:
Restriction wherein estimates the district and be included in this image, and larger rectangle comprises less rectangle about the position of a plurality of rectangular areas and the information of size;
All sections that are included in rectangle less in a plurality of rectangular areas are encoded;
All sections in the part that is included in the larger rectangle of getting rid of this less rectangular area are encoded; With
The zone that will not be included in described a plurality of rectangular area is defined as a background area and the section that is included in background area is encoded.
18. method according to claim 17 represents comprising each position and left edge and the bottom right edge position of size by each rectangle of a plurality of rectangular areas of estimating the district.
19. method according to claim 17 wherein will have the intrinsic digital distribution of predefined procedure to comprising a plurality of rectangular areas of estimating district and background area.
20. picture decoding method, wherein an image segmentation is become have the section unit of pre-sizing and it is encoded, become a background area and to estimate the district this image segmentation, and this image is decoded in bit stream, this bit stream limits by a plurality of rectangular areas, in these a plurality of rectangular areas, will estimate the district and be included in this image, and larger rectangle comprises less rectangle, and the method comprises:
Extract from bit stream comprise a plurality of rectangular areas of estimating the district each the position and the information of size;
All sections that are included in rectangle less in a plurality of rectangular areas are decoded;
Only the section in the part of the larger rectangle that is included in eliminating less rectangular area is decoded; With
All sections that are not included in the background area in a plurality of rectangular areas are decoded.
21. method according to claim 20, wherein extract from bit stream be located at a plurality of rectangular areas each the upper left and the information of the section of bottom right edge in order to limit each position and size of a plurality of rectangular areas.
22. method according to claim 20 wherein limits this bit stream in order to will have the intrinsic digital distribution of predefined procedure and gives these a plurality of rectangular areas comprise this evaluation district and this background area.
Applications Claiming Priority (4)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| KR1020020000602A KR100624404B1 (en) | 2002-01-05 | 2002-01-05 | Adaptive coding method and apparatus considering human visual characteristics |
| KR602/02 | 2002-01-05 | ||
| KR51883/02 | 2002-08-30 | ||
| KR10-2002-0051883A KR100528324B1 (en) | 2002-08-30 | 2002-08-30 | Apparatus and method for coding and decoding image using overlapped rectangular slices |
Related Parent Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB031075029A Division CN1250008C (en) | 2002-01-05 | 2003-01-05 | Method and apparatus for coding & decoding picture considering human vision characteristic |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1809166A CN1809166A (en) | 2006-07-26 |
| CN1809166B true CN1809166B (en) | 2013-05-15 |
Family
ID=36840816
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2006100067204A Expired - Lifetime CN1809166B (en) | 2002-01-05 | 2003-01-05 | Image coding and decoding method and apparatus |
Country Status (2)
| Country | Link |
|---|---|
| KR (1) | KR100624404B1 (en) |
| CN (1) | CN1809166B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100528324B1 (en) * | 2002-08-30 | 2005-11-15 | 삼성전자주식회사 | Apparatus and method for coding and decoding image using overlapped rectangular slices |
| KR100750138B1 (en) * | 2005-11-16 | 2007-08-21 | 삼성전자주식회사 | Method and apparatus for image encoding and decoding considering the characteristic of human visual system |
| KR100792247B1 (en) * | 2006-02-28 | 2008-01-07 | 주식회사 팬택앤큐리텔 | System for processing image data and method therefor |
| KR100786413B1 (en) * | 2006-06-13 | 2007-12-17 | 주식회사 팬택앤큐리텔 | System for processing image data |
| CN101102495B (en) * | 2007-07-26 | 2010-04-07 | 武汉大学 | A video image decoding and encoding method and device based on area |
| US20120170648A1 (en) * | 2011-01-05 | 2012-07-05 | Qualcomm Incorporated | Frame splitting in video coding |
| KR101662739B1 (en) * | 2014-10-29 | 2016-10-05 | 삼성전자주식회사 | Method and apparatus for encoding and decoding coding unit of picture boundary |
| KR102379185B1 (en) * | 2015-08-31 | 2022-03-25 | 엘지디스플레이 주식회사 | Display device |
| KR20230141952A (en) * | 2018-03-29 | 2023-10-10 | 닛폰 호소 교카이 | Image encoding device, image decoding device, and program |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5614952A (en) * | 1994-10-11 | 1997-03-25 | Hitachi America, Ltd. | Digital video decoder for decoding digital high definition and/or digital standard definition television signals |
| CN1255022A (en) * | 1998-11-24 | 2000-05-31 | 通用仪器公司 | Rate control system of MPEG code converter under condition that picture type is not understood in advance |
| CN1280744A (en) * | 1997-10-28 | 2001-01-17 | 太阳微系统公司 | Encoding system and method for scrolling encoded MPEG stills in an interactive television application |
| US6256423B1 (en) * | 1998-09-18 | 2001-07-03 | Sarnoff Corporation | Intra-frame quantizer selection for video compression |
| CN1328748A (en) * | 1998-10-02 | 2001-12-26 | 通用仪器公司 | Method and apparatus for providing rate control in video encoder |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100493854B1 (en) * | 1995-10-26 | 2005-08-04 | 주식회사 팬택앤큐리텔 | Quantization Method in Object-Centered Encoding |
| JPH09200764A (en) * | 1996-01-16 | 1997-07-31 | Canon Inc | Encoder |
| KR100273459B1 (en) * | 1998-12-21 | 2000-12-15 | 구자홍 | Bit string control method of image data |
| KR100643453B1 (en) * | 2001-11-17 | 2006-11-10 | 엘지전자 주식회사 | Bit rate control based on object |
-
2002
- 2002-01-05 KR KR1020020000602A patent/KR100624404B1/en active IP Right Grant
-
2003
- 2003-01-05 CN CN2006100067204A patent/CN1809166B/en not_active Expired - Lifetime
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5614952A (en) * | 1994-10-11 | 1997-03-25 | Hitachi America, Ltd. | Digital video decoder for decoding digital high definition and/or digital standard definition television signals |
| CN1280744A (en) * | 1997-10-28 | 2001-01-17 | 太阳微系统公司 | Encoding system and method for scrolling encoded MPEG stills in an interactive television application |
| US6256423B1 (en) * | 1998-09-18 | 2001-07-03 | Sarnoff Corporation | Intra-frame quantizer selection for video compression |
| CN1328748A (en) * | 1998-10-02 | 2001-12-26 | 通用仪器公司 | Method and apparatus for providing rate control in video encoder |
| CN1255022A (en) * | 1998-11-24 | 2000-05-31 | 通用仪器公司 | Rate control system of MPEG code converter under condition that picture type is not understood in advance |
Also Published As
| Publication number | Publication date |
|---|---|
| KR100624404B1 (en) | 2006-09-18 |
| KR20030060009A (en) | 2003-07-12 |
| CN1809166A (en) | 2006-07-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1250008C (en) | Method and apparatus for coding & decoding picture considering human vision characteristic | |
| US9716891B2 (en) | Quantization method and apparatus in encoding/decoding | |
| CN108293125B (en) | Digital image processing system and method | |
| CN101969563B (en) | Image processing device, image processing method | |
| CN101233697B (en) | Coding device, coding method, decoding device, decoding method, and programs of same | |
| EP2168382B1 (en) | Method for processing images and the corresponding electronic device | |
| CN103563374A (en) | Image-processing device and method | |
| CN102792695A (en) | Method and apparatus for encoding and decoding image by using large transform unit | |
| EP0859517A2 (en) | Video coder employing pixel transposition | |
| KR20040018243A (en) | Video processing | |
| JP2006304329A (en) | Encoding method, decoding method, encoding device, and decoding device | |
| CN1809166B (en) | Image coding and decoding method and apparatus | |
| EP0932306A2 (en) | Method and apparatus for encoding video shape and texture information | |
| KR100697516B1 (en) | Moving picture coding method based on 3D wavelet transformation | |
| CN114501012A (en) | Image filtering, coding and decoding method and related equipment | |
| Chung et al. | Quantization control for improvement of image quality compatible with MPEG2 | |
| KR100319916B1 (en) | Hierarchical Scalable Digital Image Compression / Restoration Methods and Apparatus | |
| Strukov et al. | Evaluation of Video Compression Methods for Network Transmission on Diverse Data: A Case Study | |
| KR20060103409A (en) | Region of interest (roi) coding method and apparatus with slice structure | |
| Tiwari et al. | A comparative study on image and video compression techniques | |
| KR100720005B1 (en) | Image coding method | |
| JPH0970038A (en) | Image data processing method | |
| Aminlou et al. | Pattern-Based Error Recovery of Low Resolution Subbands in JPEG2000 | |
| KR20040020306A (en) | Apparatus and method for coding and decoding image using overlapped rectangular slices | |
| JPH03177165A (en) | Picture data signal processing system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CX01 | Expiry of patent term | ||
| CX01 | Expiry of patent term |
Granted publication date: 20130515 |