CN100424717C - Method and apparatus for improving video quality of low bit-rate video - Google Patents
Method and apparatus for improving video quality of low bit-rate video Download PDFInfo
- Publication number
- CN100424717C CN100424717C CNB2004800102531A CN200480010253A CN100424717C CN 100424717 C CN100424717 C CN 100424717C CN B2004800102531 A CNB2004800102531 A CN B2004800102531A CN 200480010253 A CN200480010253 A CN 200480010253A CN 100424717 C CN100424717 C CN 100424717C
- Authority
- CN
- China
- Prior art keywords
- adjacent blocks
- determine whether
- block
- adjacent
- edge pixels
- 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 - Fee Related
Links
Images
Abstract
Embodiments describe a method, apparatus and system for processing images using block based compression. In one embodiment, a method comprises determining whether two blocks are neighboring blocks, determining whether the two neighboring blocks are both subdivided, if the two blocks are neighboring blocks; performing deblocking filtering on one or more edge pixels of the two neighboring blocks, if it is determined that both of the two neighboring blocks are not subdivided.
Description
The cross reference of related application
Present patent application is advocated the provisional application 60/456 of " Method andApparatus for Improving Video Quality of Low Bit-Rate Video (being used to improve the method and apparatus of the video quality of low-bit rate video) " by name of submitting on March 17th, 2003,030 right of priority, it is conveyed this assignee, and is incorporated herein by reference.
Technical field
The present invention relates generally to data compression, more specifically, relates to block-based compressibility.
Background technology
Transition coding is a kind of Image Compression commonly used, and it is subimage or the piece that is used to handle with image division.Because piece is encoded independently, block-based being compressed in introduced pseudo-shadow (artifact) between the block boundary.Therefore, the correlativity between the block boundary is not considered in described conversion.So this technology can produce the low bit rate image of severe compression usually, still, but comprises serious compression artefacts in the image, such as blocking (blocking), ring, motion blur.
Thereby, algorithm after the processing of several be used to deblock (deblocking) has been proposed, to reduce and/or to remove compression artefacts.Yet, many calculating that all relate to complexity, and may produce total blur effect to output image.Other blocking wave filter can not be preserved marginal information effectively, and hardware is realized complicated usually.So, need more simply and/or effectively to deblock processing.
Summary of the invention
The method and apparatus that is used for the image that the block-based compression of processing and utilizing compresses can comprise: determine whether that two pieces are adjacent block; Determine whether that described two adjacent blocks are all divided again; If described two adjacent blocks are all divided again, on one or more edge pixels of these two adjacent blocks, carry out de-blocking filter.Determine whether that two adjacent blocks are all divided and can comprise: the variance yields that obtains in described two adjacent blocks each; Described variance yields and first threshold are compared; And, determine whether that described two adjacent blocks are all divided again based on the described comparison of described variance yields and described first threshold.Perhaps, determining whether that two adjacent blocks are all divided again may further include: obtain block size assignment; And utilize described block size assignment to determine whether that described two adjacent blocks are divided again.
Described method and apparatus may further include: if described two adjacent blocks are all divided again, determine whether that a quilt in described two adjacent blocks divides again; If a quilt in described two adjacent blocks divides again, on one or more edge pixels of described two adjacent blocks, use first deblocking filter; And,, on one or more edge pixels of described two adjacent blocks, use second deblocking filter if described two adjacent blocks are not all divided again.
Described method and apparatus can further include: if described two adjacent blocks are not all divided again, obtain one or more differences of one or more edge pixels of described two adjacent blocks; The described one or more differences and second threshold value are compared; And, relatively select described second deblocking filter based on described one or more differences and described second threshold value described.
Obtaining one or more differences can comprise: obtain the difference between three edge pixels of described two adjacent blocks; And, if in the described difference at least two greater than described second threshold value, select described second deblocking filter can comprise the use Gaussian filter.
Description of drawings
With reference to the following drawings various embodiment are described in detail below, wherein, similarly reference number is indicated similar elements, wherein:
Fig. 1 is an example of image compressor;
Fig. 2 is an example of image decompressor;
Fig. 3 shows and is used to the instantiation procedure that determines whether that piece is divided again;
Fig. 4 A shows the example that piece divides again to 4D;
Fig. 5 shows the example of two adjacent blocks in the image;
Fig. 6 shows the instantiation procedure that is used to determine whether to use deblocking filter;
Fig. 7 shows another instantiation procedure that is used to determine whether to use deblocking filter;
Fig. 8 A shows the ordering of the ABSDCT that is used for the 16x16 piece to 8D;
Fig. 9 A shows the example of block size assignment data to 9B;
Figure 10 A shows more many cases of block size assignment data to 10D;
Figure 11 shows the instantiation procedure that produces the block size assignment data that is used for ABSDCT;
Figure 12 is a chart, shows different variance threshold values;
Figure 13 shows some variable-definitions; And
Figure 14 shows the instantiation procedure that is used for determining whether to use in the system that uses ABSDCT deblocking filter.
Embodiment
In the compressibility that has used block-based discrete cosine transform (DCT), data stream is divided into the discrete cosine of block of pixels and process conversion.Because being correlated with between the block boundary do not considered in described conversion, and because each piece absolute coding, this block-based processing makes and produces the pseudo-shadow of blocking between the block boundary.
Usually, in the compressibility that uses DCT, the size of each data block is fixed.Yet the dynamic image compression technique that has is used the piece and the sub-piece of the self-adaptation size of encoding D CT coefficient data, can provide good compression when guaranteeing quality of image signals.This technology is called as variable block size DCT.An example of variable block size DCT is the United States Patent (USP) 5 at " Adaptive Block SizeImage Compression Method And System (adaptive block sized images compression method and system) " by name, disclosed self-adaptation block size discrete cosine transform (ABSDCT) in 021,891.The DCT technology is also at the United States Patent (USP) 5 that is called " Adaptive Block Size ImageCompression Method And System (adaptive block sized images compression method and system) ", 107, obtain open in 345, and, United States Patent (USP) 5 at " Adaptive Block SizeImage Compression Method And System (adaptive block sized images compression method and system) " by name, the use that the ABSDCT technology combines with discrete quadtree conversion (Discrete Quadtree Transform) technology has been discussed in 452,104.Described adaptive block size is selected to utilize the redundancy that exists for the information in the image data frame.Below ABSDCT will be described in more detail.
The process of deblocking reduces pseudo-shadow to embodiment described below by allowing simply and effectively, and this process can easily realize in the compressibility that has used block-based DCT.Described embodiment is especially effective in variable block size DCT.
In the following description, detail will be provided so that embodiment can thoroughly be understood.Yet it will be understood by those skilled in the art that does not need these details, can implement this embodiment.Embodiment for example, circuit has been shown in block diagram, so that can not become hard to understand because of unnecessary details.In other example, circuit, structure and the technology known can at length be shown, so that described embodiment is more distinct understandable.
It may be noted that described embodiment can be described to a process, it obtains describing by process flow diagram, flow chart, structural drawing or block diagram.Although process flow diagram can be described as sequential process with operation, many operations can walk abreast or carry out simultaneously.In addition, the order of described operation can be rearranged.When its operation was finished, process stopped.Process can be corresponding to method, function, program, subroutine, subroutine etc.When process during corresponding to function, it stops in response to described function returning to call function or principal function.
Fig. 1 shows the example of image compressor 100, and Fig. 2 shows the example with the symmetrical image decompressor 200 of image compressor 100.Image compressor 100 comprises variable block size DCT (VBSDCT) module 110, quantization modules 120, and variable-length encoding (VLC) module 130.Image decompressor 200 comprises variable length decoding (VLD) module 210, inverse quantization module 220, and anti-VBSDCT module 230.Image decompressor 200 also comprises de-blocking filter module 240, when needed block edge being carried out filtering, and processor 250, with control de-blocking filter module 240.
Usually, the data stream that is input to image compressor 100 is made up of picture frame.Picture frame generally is divided into sheet, and sheet can be divided into data block, and data block can be divided into pixel, and pixel is the minimum unit of image.Each picture frame comprises an integer sheet, and each image sheet represents the image information of one group of n continuous sweep row, for example 16 continuous sweep row.In this case, each data block is corresponding to the 16x16 block of pixels of the image of crossing over described frame.And a frame can be split into even number and odd number sheet, thereby forms even number half-frames and odd number field.In addition, image pixel uses the color component system of red, green, blue (RGB) to represent usually.Yet, because that human eye changes brightness is more responsive, and colourity is changed not too responsive, the YCbCr color space is used for the presentation video pixel usually in video compress.Described YCbCr color space is the linear transformation of RGB component, and wherein Y is a chromatic component, and Cb and Cr are color component.If frame is split into the even/odd frame, corresponding to Y, Cb and Cr component, picture frame can be made up of three even number half-frames and three odd number fields.
In above description, sheet can also be represented one group of continuous sweep row outside 16 continuous sweep row.And if block can be divided again, and described data block can be n x m piece, and wherein n is not equal to m.In addition, the different colours space with identical or different number color component can be used to the presentation video pixel.Yet, the purpose that the block size of 16 x, 16 pixels and YCbCr color space will be used to explain below.
Refer again to Fig. 1, VBSDCT module 110 is transformed into frequency domain with described digital image information from the spatial domain, and generates the DCT coefficient with corresponding block size assignment (BSA) information.VBSDCT module 110 divide as required and processing block and sub-piece in digital image information.Fig. 3 shows process 300, to determine whether that piece is divided again.In process 300, obtain the variance yields (310) of DCT piece.Then, described variance is compared (320) with the big or small set threshold value TBS that is described.Based on the comparison of described variance yields and described threshold value TBS, determine whether that described quilt divides again.That is, if described variance yields greater than TBS, a then described quilt divides (330 and 340) again.Otherwise described is not divided (350) again.At this, similar procedure can be used on the sub-piece, to determine whether that described sub-piece is divided again.In this case, the described variance and the big or small set threshold value of sub-piece are for this reason compared.
If block is divided, and how BSA information indicator dog is divided again.For example, described BSA information can indicate 16 of 16 x to be subdivided into 8 of four 8 x, and shown in Fig. 4 A, perhaps, 8 of 8 x also may be further divided into 4 of four 4 x, shown in Fig. 4 B.In other systems, BSA information can indicate 16 of 16 x to be subdivided into 8 of four 8 x, shown in Fig. 4 C.As shown in the figure, among Fig. 4 C among 8 of 8 x one is subdivided into 4 of four 4 x, and based on system configuration and/or demand, further is divided into 2 of four 2 x, shown in Fig. 4 C.Then, quantization modules 120 quantizes described DCT coefficient, and, the DCT coefficient that VLC 130 utilizes the variable-length encoding technique compresses to quantize.
In image decompressor 200, the image information that VLD module 210 decompresses and compressed, the image information of the described decompression of quantization modules 220 inverse quantizations, and, anti-VBSDCT module 230 is utilized block size assignment information, and described inverse quantization image information is converted to the spatial domain by frequency domain.Processor 250 determines whether that two pieces of described image are adjacent block, as shown in Figure 5.Then, processor 250 is based on block edge activity or busy degree, and determining whether need be to the processing of deblocking of described two adjacent blocks.The processing of deblocking if desired, one or more common edge pixels of described two adjacent blocks are by 240 filtering of de-blocking filter module.Then, the image information after the processing is exported to display and/or is stored to be used for demonstration.
Fig. 6 shows a process 600, and it is used for the image that the block-based compression of processing and utilizing is compressed.In process 600, determine whether that two adjacent blocks are all divided (610) again.At this, described BSA information can be used to determine whether that described two adjacent blocks are divided again.If described two adjacent blocks are all divided again, then on one or more edge pixels of described two adjacent blocks, use deblocking filter (620).
Fig. 7 shows another process 700, and it is used for the image that the block-based compression of processing and utilizing is compressed.In process 700, determine whether that two adjacent blocks are all divided (710) again.If determine that described two adjacent blocks are all divided again, then do not use deblocking filter.Yet if described two adjacent blocks are all divided again, that is, at least one in described two adjacent blocks do not divided again, determines whether that further a quilt in described two adjacent blocks divides (720) again.If a quilt in described two adjacent blocks divides again, then on one or more pixels of described two adjacent blocks, use first deblocking filter (730).At this, described first deblocking filter can be 2 average filters (two pointaveraging filter), and it is used on two edge pixels of described two adjacent blocks.If described two adjacent blocks are not all divided again, then on one or more pixels of described two adjacent blocks, use second deblocking filter.
More specifically, obtain difference (740) between the edge pixel of one or more correspondences between described two adjacent blocks.Difference represents to cross over the variance of block boundary, and can utilize various technology to be acquired and/or to obtain.Can obtain the simple first order difference between two respective edges pixels of two adjacent blocks.In other embodiments, can obtain and use second order difference.Described one or more difference is compared (750) with threshold value TD.Based on the comparison of described one or more differences and threshold value TD, select described second deblocking filter (760).
Described threshold value TD depends on brightness usually, and, for different systems and/or dissimilar images, can be provided with in advance.In one embodiment, the average of described two adjacent blocks on average can be used as described threshold value TD.Perhaps, described threshold value TD can be poor in the average of described two adjacent blocks.Described threshold value TD can also be optimised, changes with the brightness (intensity) that utilizes scale factor to handle in the image, and this α is proportional to as the contrast of giving a definition, wherein μ
cBe the average of current block, μ
nFor having comprised the average of obtaining the piece of employed edge pixel in the described difference.
α＝(|μ
c-μ
n|)/μ
n
The span of described α is 0 to 1.
In addition, in one embodiment, the difference between three edge pixels of described two adjacent blocks is acquired and is compared with threshold value TD.If at least two in the described difference greater than TD, select Gaussian filter.That is, if in described three differences three greater than TD, use 6 Gaussian filters on six edge pixels of described two adjacent blocks.If two in described three differences greater than TD, then on four edge pixels of described two adjacent blocks, use 4 Gaussian filters.If one in described three differences greater than TD, then on two edge pixels of described two adjacent blocks, use average filter.
Refer again to Fig. 2, thereby processor 250 can determine whether and need deblock.As above discuss, as long as system allows, processor 250 also can be selected different deblocking filters according to the feature of described adjacent block.Thereby de-blocking filter module 240 can comprise the wave filter of one or more types, for example, but is not limited to average filter and/or Gaussian filter.
In addition, as above discuss, available ABSDCT realizes VBSDCT module 110.Below will utilize the block size of 16 x, 16 pixels to describe the compress technique of having used ABSDCT.Usually, each brightness and chromatic component are sent to block interleaver (block interleaver) (not shown).In one embodiment, to shown in the 8D, the piece of 16 x 16 is submitted to described block interleaver as Fig. 8 A, and it is arranged in the image example in 16 of described 16 x, to generate piece and to form data sub-block, analyzes and be used for DCT.One 16 x 16DCT is applied to single order, and four 8 x 8DCT are applied to second order, and 16 4 x 4DCT are applied to three rank, and 64 2 x 2DCT are applied to quadravalence.Described DCT operation has reduced spatial redundancy intrinsic in the image source.After carrying out DCT, the major part of image signal energy is tended to concentrate in the DCT coefficient of minority.
For 16 of described 16 x and each sub-piece, coefficient behind the analytic transformation is to determine coding described or the needed bit number of sub-piece.Then, to need to select the piece that minimum number bits encodes or the combination of sub-piece, with the presentation video fragment.For example, can select two 8 x 8 sub-pieces, six 4 x, 4 sub-pieces, and eight 2 x 2 sub-pieces, to represent described image segments.Then, the combination of piece of being chosen or sub-piece is correctly arranged in regular turn.
Analyze the coefficient after the conversion, and the combination of selecting described or sub-piece is to represent described image segments.Like this, can produce the block size assignment information of the block size assignment in the expression n x n piece.For 16 x, 16 data blocks, the ABSDCT technology generates the data be called PQR information, and these data are illustrated in the block size assignment in 16 of 16 x.Described PQR information is the variable bit width data, and describes the degree that 16 quilts of 16 x divide again.The R bit of PQR field represents whether 16 of described 16 x are subdivided into 8 of four 8 x.Shown in Fig. 9 A, if described R bit is " 0 ", described is kept perfectly.In this case, do not need further PQR information, described PQR field only is 1 bit long.If described R bit is " 1 ", 16 of then described 16 x are subdivided into 8 of four 8 x, and shown in Fig. 9 B, and at least four extra bits will be present in the described PQR field.
Described four additional bit are called as " Q " information.The bit of each Q is represented 8 of 8 x are further divided into 4 of four 4 x.For the bit of each Q that is set up, also have the bit of other four " P ", it is used to indicate whether to have 4 of 4 x to be subdivided into 2 x 2.So the length of PQR data can be 1 to 21 bit, this depends on the block size assignment in 16 of described 16 x.If 8 of each 8 x are divided again, then described PQR information will be 21 bit long.Figure 10 A-D shows some examples of 16 of 16 x corresponding with the PQR data.
Therefore, each piece can be divided into the sub-piece that size is 8 x, 8,4 x 4 and/or 2 x 2, and this depends on allocation criteria.Described criterion of dividing n x n piece again is described variance, and is as follows:
If the piece variance of n x n piece exceeds certain threshold value, this n x n piece of pixel can be subdivided into 4 n/2 x n/2 pieces.At this, for 10 bit image, the value of described average is in (0,1023) scope.Like this, described image is divided into 12 bins (bin), and one group of threshold value is used to each bin of each color component.And, can determine described threshold value based on the statistics of collecting from the number of dissimilar picture frames.The typical threshold setting as shown in figure 12.
Figure 11 shows instantiation procedure 1100, and it is used to 16 of 16 x to generate PQR information.For each piece, obtain average and variance V16 (1110).Described variance V16 and the appropriate threshold value T16 that is used for corresponding average are compared (1115).If described variance V16 is not more than threshold value T16, the R value of PQR data is set to 0, and described process finishes (1120).Otherwise described R value is set to 1 (1125).Then, be in described four 8 x, the 8 sub-pieces 0 to 3 each, obtain variance V8 (i), { i=1 is to 4}, shown in Fig. 9 B, and each variance V8 (i) and suitable threshold T8 compared, and thinks that described PQR data determine Q value (1130 to 1140).If variance V8 (i) is not more than threshold value T8, corresponding Q (i) is set to 0 (1145).Otherwise the value of described Q (i) is set to 1 (1150).Be set in four 4x, the 4 sub-pieces of 8 of each 8 x of 1 each for Q (i) then, obtain variance V4 (j), { j=1 is to 4}, and each variance V4 (j) and suitable threshold T4 are compared, and thinks that described PQR data determine described P value (1155 to 1165).If variance V4 (j) is not more than threshold value T4, corresponding P (j) is set to 0 (1170).Otherwise described P (j) value is set to 1 (1175).
Like this, described PQR information can be generated, and is used to deblock in the image decompressor such as image decompressor 200.Described PQR information is used to the edge content in definite image.Described marginal information in the piece is big more, and described block size is more little, and described PQR sign indicating number is long more.Figure 14 shows instantiation procedure 1400, and it is used for the image of processing and utilizing ABSDCT compression, and Figure 13 shows the various definition of using in process 1400.
When determining whether,, obtain described PQR information (1410) for each piece to two adjacent blocks use deblocking filters.If each PQR bit is all greater than 5 bits (1415), this process finishes.Also promptly, each piece all is confirmed to be and is divided, and has been considered to comprise enough marginal informations.Otherwise, if in the described PQR bit one is greater than 5 bits, at { x1,2 average filters of the last use of y1} (1420 and 1425).If described PQR bit all is not more than 5 bits, then obtain difference d1, d2 and d3 (1430).If d1, d2 and d3 are greater than threshold value TD, then at { x1, x2, x3, y1, y2,6 Gaussian filters of the last use of y3} (1435 and 1440).If d1 and d2 are greater than threshold value TD, then at { x1, x2, y1,4 Gaussian filters of the last use of y2} (1445 and 1450).If d1 is greater than threshold value TD, then at { x1,2 average filters of the last use of y1} (1455 and 1460).
In process 1400, described embodiment is not limited to average filter and/or Gaussian filter.Also can use the various wave filters outside average filter and/or the Gaussian filter.
As shown, can easily in decompressor, realize the de-blocking filter module.So, can reduce pseudo-shadow significantly, and improve visual quality for images.It may be noted that when de-blocking filter module 240 as shown by from anti-VBSDCT module 230 with when processor 250 is realized discretely, in de-blocking filter module 240, anti-VBSDCT module 230 and the processor 250 one or its combination can realize together.
And illustrated embodiment can make up by hardware, software, firmware, middleware, microcode or its and realize.When realizing in software, firmware, middleware or microcode, the key element of described embodiment is program code or code snippet in order to carry out required task, and it can be stored in (not shown) in the machine readable media.Code snippet can be represented the combination in any of process, function, subroutine, program, routine, subroutine, module, software package, class or instruction, data structure or program statement.By transmitting and/or reception information, data, variable, parameter or memory contents, code snippet can combine with another code snippet or hardware circuit.Information, variable, parameter, data etc. can transmit, pass on or send by any suitable mode, and described mode comprises that storage is shared, message transmission, token transmission, Network Transmission etc.And described machine readable media can realize at the product that is used for computer system, and can be so that the machine readable code device is embodied as wherein.
In addition, although use variable block size DCT that described embodiment is described, the aforesaid technology of deblocking also can realize in having the DCT of fixed block size.In this case, described BSA information can be generated, but actual DCT can not be used for.But described BSA information will be used in image decompressor, to determine whether and need deblock for two adjacent blocks.
It may be noted that previous embodiment as just example, and can not be interpreted as it is limitation of the present invention.Description to described embodiment is intended to describe, and the scope of unrestricted claim.Equally, this programme can be applied to the device of other type easily, and many replacements, change and variation are conspicuous for those skilled in the art.
Claims (22)
1. be used for the method for the image that the block-based compression of processing and utilizing compresses, it comprises:
Determine whether that two pieces are adjacent block;
If determine that described two pieces are adjacent block, determine whether that described two adjacent blocks are all divided again;
If determine that described two adjacent blocks are all divided again, determine whether that a quilt in these two adjacent blocks divides again;
Divide again if determine a quilt in described two adjacent blocks, on one or more edge pixels of these two adjacent blocks, use first deblocking filter; And
If determine that these two adjacent blocks are not all divided again, on one or more edge pixels of these two adjacent blocks, use second deblocking filter.
2. according to the process of claim 1 wherein, determine whether that two adjacent blocks are all divided again to comprise:
Obtain in described two adjacent blocks the variance yields of each;
Described variance yields and first threshold are compared; And
Based on the described comparison of described variance yields and described first threshold, determine whether that described two adjacent blocks are all divided again.
3. according to the process of claim 1 wherein, determine whether that two adjacent blocks are all divided again to comprise:
Obtain block size assignment; And
Utilize described block size assignment to determine whether that described two adjacent blocks are divided again.
4. according to the process of claim 1 wherein, utilize described first deblocking filter to comprise:
On two edge pixels of described two adjacent blocks, use 2 average filters.
5. according to the method for claim 1, further comprise:
If described two adjacent blocks are not all divided again, obtain one or more differences of one or more edge pixels of described two adjacent blocks;
The described one or more differences and second threshold value are compared; And,
Based on the described comparison of described one or more differences and described second threshold value, select described second deblocking filter.
6. according to the method for claim 5, wherein, describedly obtain one or more differences and comprise:
Obtain the first order difference between the edge pixel of described two adjacent blocks.
7. according to the method for claim 5, wherein, describedly obtain one or more differences and comprise:
Obtain the second order difference between described two adjacent tile edges pixels.
8. according to the method for claim 5, wherein, describedly obtain one or more differences and comprise:
Obtain the difference between three edge pixels of described two adjacent blocks; And wherein, select described second deblocking filter to comprise:
If at least two in the described difference greater than described second threshold value, use Gaussian filter.
9. method according to Claim 8, wherein, use Gaussian filter to comprise:
If described difference greater than described second threshold value, is used 6 Gaussian filters on six edge pixels of described two adjacent blocks.
10. method according to Claim 8, wherein, use Gaussian filter to comprise:
If two in the described difference greater than described second threshold value, on four edge pixels of described two adjacent blocks, use 4 Gaussian filters.
11. method according to Claim 8 wherein, selects described second deblocking filter also to comprise:
If one in the described difference greater than described second threshold value, on two edge pixels of described two adjacent blocks, use 2 average filters.
12. be used for the device of the image that the block-based compression of processing and utilizing compresses, it comprises:
Be used to determine whether that two pieces are the device of adjacent block;
If determine that described two pieces are adjacent block, determine whether described two devices that adjacent block is all divided again;
If determine that described two adjacent blocks are all divided again, determine whether a device that quilt divides again in these two adjacent blocks;
Divide again if determine a quilt in described two adjacent blocks, on one or more edge pixels of these two adjacent blocks, use the device of first deblocking filter; And
If determine that these two adjacent blocks are not all divided again, on one or more edge pixels of these two adjacent blocks, use the device of second deblocking filter.
13., wherein, describedly be used to determine whether that the device that two adjacent blocks are all divided again comprises according to the device of claim 12:
Be used for obtaining each the device of variance yields of described two adjacent blocks;
Be used for device that described variance yields and first threshold are compared; And
Based on the described comparison of described variance yields and described first threshold, determine whether described two devices that adjacent block is all divided again.
14., wherein, describedly be used to determine whether that the device that two adjacent blocks are all divided again comprises according to the device of claim 12:
Be used to obtain the device of block size assignment; And
Be used to utilize described block size assignment to determine whether described two devices that adjacent block is divided again.
15. the device according to claim 12 further comprises:
If described two adjacent blocks are not all divided again, obtain the device of one or more differences of one or more edge pixels of described two adjacent blocks;
Be used for device that described one or more differences and second threshold value are compared; And,
Based on the described comparison of described one or more differences and described second threshold value, select the device of described second deblocking filter.
16. according to the device of claim 15, wherein, the described device that is used to obtain one or more differences comprises:
Be used to obtain the device of the difference between three edge pixels of described two adjacent blocks; And wherein, describedly be used to select the device of described second deblocking filter to comprise:
If at least two in the described difference greater than described second threshold value, use the device of Gaussian filter.
17. according to the device of claim 16, wherein, the device of described use Gaussian filter comprises:
If described difference greater than described second threshold value, is used the device of 6 Gaussian filters on six edge pixels of described two adjacent blocks.
18. according to the device of claim 16, wherein, the device of described use Gaussian filter comprises:
If two in the described difference greater than described second threshold value, on four edge pixels of described two adjacent blocks, use the device of 4 Gaussian filters.
19., wherein, describedly be used to select the device of described second deblocking filter further to comprise according to the device of claim 16:
If one in the described difference greater than described second threshold value, on two edge pixels of described two adjacent blocks, use the device of 2 average filters.
20. be used to handle the device that uses the image that block-based compression compresses, it comprises:
Processor, first deblocking filter and second deblocking filter, it is configured to:
Determine whether that two pieces are adjacent block;
If determine that described two pieces are adjacent block, determine whether that described two adjacent blocks are divided again;
If determine that described two adjacent blocks are all divided again, determine whether that a quilt in these two adjacent blocks divides again;
Divide again if determine a quilt in described two adjacent blocks, on one or more edge pixels of these two adjacent blocks, use first deblocking filter; And
If determine that these two adjacent blocks are not all divided again, on one or more edge pixels of these two adjacent blocks, use second deblocking filter.
21. according to the device of claim 20, wherein, described processor utilizes block size assignment information to determine whether that two adjacent blocks are divided.
22. according to the device of claim 20, wherein, described processor determines whether that based on the variance yields of each piece two adjacent blocks are divided.
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US45603003P | 2003-03-17 | 2003-03-17 | |
| US60/456,030 | 2003-03-17 | ||
| US10/802,285 | 2004-03-16 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1774722A CN1774722A (en) | 2006-05-17 |
| CN100424717C true CN100424717C (en) | 2008-10-08 |
Family
ID=36760957
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2004800102531A Expired - Fee Related CN100424717C (en) | 2003-03-17 | 2004-03-17 | Method and apparatus for improving video quality of low bit-rate video |
Country Status (5)
| Country | Link |
|---|---|
| CN (1) | CN100424717C (en) |
| AR (1) | AR043643A1 (en) |
| TW (1) | TWI373734B (en) |
| UA (1) | UA87660C2 (en) |
| ZA (1) | ZA200507507B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI444047B (en) * | 2006-06-16 | 2014-07-01 | Via Tech Inc | Deblockings filter for video decoding , video decoders and graphic processing units |
| WO2011126277A2 (en) | 2010-04-05 | 2011-10-13 | Samsung Electronics Co., Ltd. | Low complexity entropy-encoding/decoding method and apparatus |
| KR20110112168A (en) * | 2010-04-05 | 2011-10-12 | 삼성전자주식회사 | Method and apparatus for video encoding based on internal bitdepth increment, method and apparatus for video decoding based on internal bitdepth increment |
| EP3113107B1 (en) * | 2015-07-02 | 2018-12-05 | Continental Automotive GmbH | Static soiling detection and correction |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5107345A (en) * | 1990-02-27 | 1992-04-21 | Qualcomm Incorporated | Adaptive block size image compression method and system |
| CN1264482A (en) * | 1997-07-16 | 2000-08-23 | 三星电子株式会社 | Signal adaptive filtering method,signal adaptive filter and computer readable medium for storing program therefor |
| CN1281617A (en) * | 1997-10-25 | 2001-01-24 | 三星电子株式会社 | Image data post-processing method for reducing quantization effect, apparatus thereof |
| US6529634B1 (en) * | 1999-11-08 | 2003-03-04 | Qualcomm, Inc. | Contrast sensitive variance based adaptive block size DCT image compression |
| US20030044080A1 (en) * | 2001-09-05 | 2003-03-06 | Emblaze Systems Ltd | Method for reducing blocking artifacts |
-
2004
- 2004-03-17 CN CNB2004800102531A patent/CN100424717C/en not_active Expired - Fee Related
- 2004-03-17 AR ARP040100894 patent/AR043643A1/en active IP Right Grant
- 2004-03-17 TW TW93107090A patent/TWI373734B/en active
- 2004-03-17 UA UAA200509654A patent/UA87660C2/en unknown
-
2005
- 2005-09-16 ZA ZA200507507A patent/ZA200507507B/en unknown
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5107345A (en) * | 1990-02-27 | 1992-04-21 | Qualcomm Incorporated | Adaptive block size image compression method and system |
| CN1264482A (en) * | 1997-07-16 | 2000-08-23 | 三星电子株式会社 | Signal adaptive filtering method,signal adaptive filter and computer readable medium for storing program therefor |
| CN1281617A (en) * | 1997-10-25 | 2001-01-24 | 三星电子株式会社 | Image data post-processing method for reducing quantization effect, apparatus thereof |
| US6529634B1 (en) * | 1999-11-08 | 2003-03-04 | Qualcomm, Inc. | Contrast sensitive variance based adaptive block size DCT image compression |
| US20030044080A1 (en) * | 2001-09-05 | 2003-03-06 | Emblaze Systems Ltd | Method for reducing blocking artifacts |
Also Published As
| Publication number | Publication date |
|---|---|
| AR043643A1 (en) | 2005-08-03 |
| TWI373734B (en) | 2012-10-01 |
| CN1774722A (en) | 2006-05-17 |
| UA87660C2 (en) | 2009-08-10 |
| TW200508987A (en) | 2005-03-01 |
| ZA200507507B (en) | 2007-06-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11758149B2 (en) | In-loop filtering method and apparatus for same | |
| KR100983996B1 (en) | Method and apparatus for improving video quality of low bit-rate video | |
| EP2777255B1 (en) | Method and device for optimizing encoding/decoding of compensation offsets for a set of reconstructed samples of an image | |
| KR101807810B1 (en) | Image information encoding and decoding method | |
| JP5384696B2 (en) | Adaptive block size DCT image compression based on variance | |
| KR101979387B1 (en) | Method for encoding and decoding image information | |
| CN100424717C (en) | Method and apparatus for improving video quality of low bit-rate video |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: DE Ref document number: 1087227 Country of ref document: HK |
|
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: GR Ref document number: 1087227 Country of ref document: HK |
|
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20081008 |