Suche Bilder Maps Play YouTube News Gmail Drive Mehr »
Anmelden
Nutzer von Screenreadern: Klicke auf diesen Link, um die Bedienungshilfen zu aktivieren. Dieser Modus bietet die gleichen Grundfunktionen, funktioniert aber besser mit deinem Reader.

Patentsuche

  1. Erweiterte Patentsuche
VeröffentlichungsnummerCN103226571 A
PublikationstypAnmeldung
AnmeldenummerCN 201310100320
Veröffentlichungsdatum31. Juli 2013
Eingetragen26. März 2013
Prioritätsdatum26. März 2013
Veröffentlichungsnummer201310100320.X, CN 103226571 A, CN 103226571A, CN 201310100320, CN-A-103226571, CN103226571 A, CN103226571A, CN201310100320, CN201310100320.X
Erfinder张玉双
Antragsteller天脉聚源(北京)传媒科技有限公司
Zitat exportierenBiBTeX, EndNote, RefMan
Externe Links:  SIPO, Espacenet
Method and device for detecting repeatability of advertisement library
CN 103226571 A
Zusammenfassung
The invention discloses a method and a device for detecting repeatability of an advertisement library. The method comprises the steps of conducting shot breakdown treatment on an advertisement to be detected, extracting an image characteristic of a key frame of each shot after the shot breakdown treatment, and judging whether the advertisement library stores the advertisement to be detected according to the extracted image characteristic and a preset characteristic library. According to the method and the device, the image characteristic of an advertisement video is extracted by taking the shots as units, and motion characteristics and integral color distribution characteristics of the shots are kept, so that the advertisements with the same shot and shot sequence can be detected, and particularly the advertisements with the same shot and different shot sequences can be prevented from being judged to be the same.
Ansprüche(10)  übersetzt aus folgender Sprache: Chinesisch
1.一种广告库重复性检测方法,其特征在于,包括以下步骤: 对待检测广告进行分镜头处理; 分别对分镜头处理后的每个镜头的关键帧提取图像特征; 根据提取的图像特征以及预设的特征库,判断广告库中是否已存储有待检测广告;其中,特征库中存储有广告库中每个广告的每个镜头的关键帧的图像特征。 An ad libs reproducible detection method comprising the steps of: detecting the ad storyboard treatment process; were key points for each shot after shot processing frame image feature extraction; the image feature extraction and pre- of signatures to determine whether the ad library to be detected stored advertisement; where signatures are stored in the image feature keyframe ad library each shot each ad.
2.根据权利要求1所述的方法,其特征在于,根据提取的图像特征以及预设的特征库,判断广告库中是否已存储有待检测广告的步骤包括: 在预设的特征库中查找与提取的图像特征匹配的图像特征; 判断对于待检测视频的每个图像特征是否均查找到匹配的图像特征; 当对于待检测视频的每个图像特征均查找到对应的匹配的图像特征时,判断匹配的图像特征是否属于同一广告,以及匹配的图像特征对应的镜头的顺序与待检测广告的镜头的顺序是否一致; 当匹配的图像特征属于同一广告并且匹配的图像特征对应的镜头的顺序与待检测广告的镜头的顺序一致时,确认广告库中已存储有待检测广告。 2. The method according to claim wherein the image feature extraction and pre signatures to determine whether the ads have been stored in the library to be detected advertisement comprises: Find feature with preset library image feature extraction of image feature matching; determining characteristics for each image to be detected whether all find the video image feature matching; when each image feature to be detected for both video find corresponding image feature matching, the judge image feature matching belong to the same advertising, and image feature matching sequence corresponding to the lens to be detected is consistent with the sequence of shots of advertising; when the image feature matching feature images belong to the same advertising and matching with the corresponding sequence of shots to be When consensus sequence detection advertising lens, confirm Ad Library has been stored to be detected ad.
3.根据权利要求2所述的方法,其特征在于,在预设的特征库中查找与提取的图像特征匹配的图像特征的步骤包括: 将提取的图像特征编码生成索引值; 在特征库中查找具有所述索引值的图像特征; 计算查找到的图像特征与提取的图像特征之间的相似度; 将查找到的与提取的图像特征之间的相似度大于预设阈值的图像特征,作为与提取的图像特征匹配的图像特征。 3. The method according to claim 2, wherein the step of searching the image feature extraction and matching of image features include preset features a library: the extracted image feature coding to generate an index value; in the feature library Find the image feature having the index value; calculating the similarity find the image feature extraction and image feature between; to find the similarities with the image feature extraction is greater than a predetermined threshold of image feature, as image feature extraction and image feature match.
4.根据权利要求1所述的方法,其特征在于,所述方法还包括: 当广告库中未存储有待检测广告时,将待检测广告存储至广告库。 4. The method according to claim 1, wherein the method further comprises: when the advertising to be detected is not stored in the library advertising, store advertising will be tested to ad libs.
5.根据权利要求4所述的方法,其特征在于,所述方法还包括: 当广告库中未存储有待检测广告,而待检测广告中包含广告库中的广告时,将广告库中包含于待检测广告的广告删除。 5. The method according to claim 4, wherein the method further comprises: when the advertising to be detected is not stored in the library advertising, and advertising to be detected ad library contains advertising, the advertising contained in the library Delete ad to be detected.
6.一种广告库重复性检测装置,其特征在于,包括: 分镜头处理模块,用于对待检测广告进行分镜头处理; 提取图像特征模块,用于分别对分镜头处理后的每个镜头的关键帧提取图像特征; 判断模块,用于根据提取的图像特征以及预设的特征库,判断广告库中是否已存储有待检测广告;其中,特征库中存储有广告库中每个广告的每个镜头的关键帧的图像特征。 An ad libs reproducible detection apparatus comprising: a shooting processing module for advertising to be detected storyboard process; image feature extraction module for key storyboard each shot separately processed frame extraction key where signatures are stored in the library each shot ads for each ad; the image feature; determining module for image feature extraction and pre signatures to determine whether the ads have been stored in the library to be detected Advertising Characteristics of image frames.
7.根据权利要求6所述的装置,其特征在于,所述判断模块包括: 查找单元,用于在预设的特征库中查找与提取的图像特征匹配的图像特征; 第一判断单元,用于判断对于待检测视频的每个图像特征是否均查找到匹配的图像特征; 第二判断单元,用于当对于待检测视频的每个图像特征均查找到对应的匹配的图像特征时,判断匹配的图像特征是否属于同一广告以及匹配的图像特征对应的镜头的顺序与待检测广告的镜头的顺序是否一致; 确认单元,用于当匹配的图像特征属于同一广告并且匹配的图像特征对应的镜头的顺序与待检测广告的镜头的顺序一致时,确认广告库中已存储有待检测广告。 7. The device according to claim 6, wherein the determination module comprises: a search unit, is used to find the image feature extraction and image feature matching preset features a library; a first judging unit, with is determined for each image feature to be detected whether the video image feature matching to find the average; a second judgment unit for, when the image feature to be detected for each of the video images are to find the corresponding feature matching, the matching is determined belong to the same image feature if advertising and image feature matching sequence corresponding to the order of the lens and the lens to be detected are the same advertisements; confirmation unit for image feature when matching ads and belonging to the same image feature matching corresponding lens order sequence to be detected when the lens is consistent advertising, advertising library has been confirmed to be detected stored advertising.
8.根据权利要求7所述的装置,其特征在于,所述查找单元用于: 将提取的图像特征编码生成索引值; 在特征库中查找具有所述索引值的图像特征; 计算查找到的图像特征与提取的图像特征之间的相似度; 将查找到的与提取的图像特征之间的相似度大于预设阈值的图像特征,作为与提取的图像特征匹配的图像特征。 8. The apparatus according to claim 7, wherein said search unit is used: the extracted image feature code generation index value; look image feature having the index value in the feature database; calculation to find similarity image feature extraction and image feature between; to find the similarities with the image feature extraction is greater than a predetermined threshold of image features, as the image feature extraction and image feature match.
9.根据权利要求6所述的装置,其特征在于,所述装置还包括: 存储模块,用于当广告库中未存储有待检测广告时,将待检测广告存储至广告库。 9. The apparatus according to claim 6, characterized in that said apparatus further comprises: a storage module for, when the advertisement is not stored in the library to be detected advertisement, the advertisement to be detected to store advertisement library.
10.根据权利要求9所述的装置,其特征在于,所述装置还包括: 删除模块,用于当广告库中未存储有待检测广告,而待检测广告中包含广告库中的广告时,将广告库中包含于待检测广告的广告删除。 10. The apparatus according to claim 9, wherein the apparatus further comprises: When you delete a module for the library when the advertising to be detected is not stored advertising, and advertising to be detected ad library contains advertising, will Advertising library contains to be detected ad removed.
Beschreibung  übersetzt aus folgender Sprache: Chinesisch

一种广告库重复性检测方法和装置 An advertisement reproducible detection method and device library

技术领域 Technical Field

[0001] 本发明涉及视频处理领域,更具体地,涉及一种广告库重复性检测方法和装置。 [0001] The present invention relates to the field of video processing, and more particularly, to an ad reproducible detection method and device library.

背景技术 Background

[0002] 随着视频播放媒体的不断发展,视频广告也越来越丰富,电视台、网站、应用中都会播出视频广告。 [0002] With the continuous development of video playback media, video advertising has become increasingly diverse, television stations, Web sites, applications, video ads will be aired.

[0003] 为了对视频广告进行编辑和管理,通常会建立广告库。 [0003] In order to edit and manage video ads, typically set up ad libs. 建立广告库时,通过人工将广告存储至广告库中。 The establishment of ad libs, manually store the ad to the ad library.

[0004] 但是,由于广告库中的广告数据量巨大,人工编辑广告库时,经常出现相同广告重复入库的问题。 [0004] However, because of the huge amount of data advertisement ad libraries, manual editing ad libraries, often repeating the same ad storage problems. 因此,亟需一种能够判断广告库中是否已存储有相同广告的方法。 Therefore, we need a way to determine whether the ad libraries already stored with the same advertising methods.

发明内容 DISCLOSURE

[0005] 有鉴于此,本发明实施例的目的是提出一种广告库重复性检测方法及装置,能够判断广告库中是否已存储有与待检测广告相同的广告。 [0005] In view of this, the purpose of an embodiment of the present invention is to provide a reproducible ad libs detecting method and device, can determine whether advertising is already stored in the library with the same ad to be detected.

[0006] 为了达到上述目的,本发明实施例提出一种广告库重复性检测方法,包括以下步骤: [0006] In order to achieve the above object, an embodiment of the present invention proposes a method for detecting the repeatability advertising library, comprising the steps of:

[0007] 对待检测广告进行分镜头处理; [0007] to be detected ad storyboard process;

[0008] 分别对分镜头处理后的每个镜头的关键帧提取图像特征; [0008] The key for each shot separately storyboard processed frame image feature extraction;

[0009] 根据提取的图像特征以及预设的特征库,判断广告库中是否已存储有待检测广告;其中,特征库中存储有广告库中每个广告的每个镜头的关键帧的图像特征。 [0009] The image feature extraction and pre signatures to determine whether the ad library to be detected stored advertisement; where signatures are stored in the image feature keyframe ad library each shot each ad.

[0010] 本发明实施例中以镜头为单元提取广告视频的图像特征,以镜头为一个单元,可以保留镜头的运动特性和整体颜色分布特征,而一般的关键帧只有图像的空间特征,因此能够检测到镜头及镜头顺序都相同的广告,尤其可以避免将镜头相同但镜头顺序不同的广告判断为相同的广告。 [0010] Generally keyframes only the spatial characteristics of the image of the motion characteristics and overall color distribution in the lens of the image feature extraction unit commercial video to the camera as a unit, you can keep the camera embodiment of the present invention, but, it is possible to detected on the lens and lens are the same order of advertising, especially to avoid the same lens but with different lenses order advertising judged to be the same ad.

[0011] 作为上述技术方案的优选,根据提取的图像特征以及预设的特征库,判断广告库中是否已存储有待检测广告的步骤包括: [0011] As a preferred aspect of the above, according to the image feature extraction and pre signatures, determining whether or ad libs have been stored to be detected advertising include:

[0012] 在预设的特征库中查找与提取的图像特征匹配的图像特征; [0012] Finding image feature extraction and image feature matching feature in the default database;

[0013] 判断对于待检测视频的每个图像特征是否均查找到匹配的图像特征; [0013] It is determined for each image feature to be detected whether the video was to find the matching image features;

[0014] 当对于待检测视频的每个图像特征均查找到对应的匹配的图像特征时,判断匹配的图像特征是否属于同一广告以及匹配的图像特征对应的镜头的顺序与待检测广告的镜头的顺序是否一致; [0014] When the characteristics for each image are to be detected video feature matching to find the corresponding image, it is determined whether the image feature matching and image ads feature the same order as the corresponding matched lens and the lens to be detected ad order are the same;

[0015] 当匹配的图像特征属于同一广告并且匹配的图像特征对应的镜头的顺序与待检测广告的镜头的顺序一致时,确认广告库中已存储有待检测广告。 [0015] When the image feature matching feature images belong to the same advertising and matching the corresponding sequence shot sequence to be detected is consistent with the advertising of the lens, confirm Ad Library has been stored to be detected ad.

[0016] 作为上述技术方案的优选,在预设的特征库中查找与提取的图像特征匹配的图像特征包括:将提取的图像特征编码生成索引值;在特征库中查找具有所述索引值的图像特征;计算查找到的图像特征与提取的图像特征之间的相似度;将查找到的与提取的图像特征之间的相似度大于预设阈值的图像特征,作为与提取的图像特征匹配的图像特征。 [0016] As a preferred aspect of the above, look for image feature extraction and image feature is included in the pre-match features a library: the extracted image feature coding to generate an index value; Finds the index value in the feature library similarity calculation find the image feature extraction and image feature between;; image feature will find similarity with the image feature extraction is greater than a predetermined threshold of image features as feature extraction and image matching image features. 本方案提高了查找效率。 The program improves the search efficiency.

[0017] 作为上述技术方案的优选,所述方法还包括:当广告库中未存储有待检测广告时,将待检测广告存储至广告库。 [0017] As a preferred aspect of the above, the method further comprises: when the advertising to be detected is not stored in the library advertising, store advertising will be tested to ad libs. 本方案根据判断结果自动进行存储操作。 The program automatically store operation based on the result.

[0018] 作为上述技术方案的优选,所述方法还包括:当广告库中未存储有待检测广告,而待检测广告中包含广告库中的广告时,将广告库中包含于待检测广告的广告删除。 [0018] As a preferred aspect of the above, the method further comprises: when the advertising to be detected is not stored in the library advertising, and advertising to be detected ad library contains advertising, the advertising contained in the library to be detected ad Delete. 本方案避免了将包含广告库中的广告的待检测广告存储至广告库后出现重复广告的情况。 The program avoids the situation after the library will contain advertising advertising advertising storage to be detected repeated ad ad library.

[0019] 本发明实施例还提出一种广告库重复性检测装置,包括: [0019] The present invention also provides a library repetitive advertising detecting device, comprising:

[0020] 分镜头处理模块,用于对待检测广告进行分镜头处理; [0020] The sub-lens processing module, used to treat detect ad storyboard process;

[0021] 提取图像特征模块,用于分别对分镜头处理后的每个镜头的关键帧提取图像特征; [0021] The image feature extraction module for key storyboard each shot separately processed frame image feature extraction;

[0022] 判断模块,用于根据提取的图像特征以及预设的特征库,判断广告库中是否已存储有待检测广告;其中,特征库中存储有广告库中每个广告的每个镜头的关键帧的图像特征。 [0022] The judgment module for image feature extraction and pre signatures to determine whether the ad library to be detected stored advertisement; where signatures are stored in the library each shot ads for each ad key Characteristics of image frames.

[0023] 作为上述技术方案的优选,所述判断模块包括: [0024] 查找单元,用于在预设的特征库中查找与提取的图像特征匹配的图像特征; [0023] As a preferred aspect of the above, the determination module comprises: [0024] The search unit, is used to find the image feature extraction and image feature matching feature in the default database;

[0025] 第一判断单元,用于判断对于待检测视频的每个图像特征是否均查找到匹配的图像特征; [0025] The first determination unit for determining image characteristics for each image feature to be detected whether the video to find all matching;

[0026] 第二判断单元,用于当对于待检测视频的每个图像特征均查找到对应的匹配的图像特征时,判断匹配的图像特征是否属于同一广告以及匹配的图像特征对应的镜头的顺序与待检测广告的镜头的顺序是否一致; Sequence [0026] second judgment unit for, when the image feature to be detected for each video image features are to find the corresponding matching, the image feature matching to judge whether belonging to the same advertising and image feature matching corresponding lens It is consistent with the sequence to be detected is advertising the lens;

[0027] 确认单元,用于当匹配的图像特征属于同一广告并且匹配的图像特征对应的镜头的顺序与待检测广告的镜头的顺序一致时,确认广告库中已存储有待检测广告。 [0027] The confirmation unit for image feature when matching belong to the same advertising and the image feature matching corresponding sequence shot sequence to be detected is consistent with the advertising of the lens, confirm Ad Library has been stored to be detected ad.

[0028] 作为上述技术方案的优选,所述查找单元用于: [0028] As the above-described preferred aspect, the search unit for:

[0029] 将提取的图像特征编码生成索引值; [0029] The extracted image features code generation index value;

[0030] 在特征库中查找具有所述索引值的图像特征; [0030] to find the image feature having the index value in the feature library;

[0031] 计算查找到的图像特征与提取的图像特征之间的相似度; [0031] calculating the similarity find the image feature extraction and image feature between;

[0032] 将查找到的与提取的图像特征之间的相似度大于预设阈值的图像特征,作为与提取的图像特征匹配的图像特征。 [0032] will find similarity with the image feature extraction is greater than a predetermined threshold of image features, as the image feature extraction and image feature match.

[0033] 作为上述技术方案的优选,所述装置还包括: [0033] As the above-described preferred aspect, the apparatus further comprising:

[0034] 存储模块,用于当广告库中未存储有待检测广告时,将待检测广告存储至广告库。 [0034] The storage module for the library when the advertising to be detected is not stored advertising, store advertising will be tested to ad libs.

[0035] 作为上述技术方案的优选,所述装置还包括: [0035] As the above-described preferred aspect, the apparatus further comprising:

[0036] 删除模块,用于当广告库中未存储有待检测广告,而待检测广告中包含广告库中的广告时,将广告库中包含于待检测广告的广告删除。 [0036] Delete module for the library when the advertising to be detected is not stored advertising, and advertising to be detected ad library contains advertising, the advertising contained in the library to be detected ad removed.

[0037] 本发明实施例的其它特征和优点将在随后的说明书中阐述,并且,部分地从说明书中变得显而易见,或者通过实施本发明而了解。 [0037] Other features and advantages of embodiments of the present invention will be set forth in the description which follows, and in part will be obvious from the description, or be learned by practice of the invention. 本发明的目的和其他优点可通过在所写的说明书、权利要求书、以及附图中所特别指出的结构来实现和获得。 The objectives and other advantages of the present invention may be realized and obtained by means of the written description, claims and drawings in the structure particularly pointed out.

[0038] 下面通过附图和实施例,对本发明实施例的技术方案做进一步的详细描述。 [0038] By the following figures and examples, technical solutions of the embodiments of the invention are described in further detail. 附图说明 Brief Description

[0039] 附图用来提供对本发明实施例的进一步理解,并且构成说明书的一部分,与本发明的实施例一起用于解释本发明,并不构成对本发明实施例的限制。 [0039] FIG to provide further understanding of the embodiments of the invention, and constitute a part of this specification, the embodiments of the invention, serve to explain the present invention is not limited to the embodiments constitute the present invention. 在附图中: In the drawings:

[0040] 图1是本发明实施例中的广告库重复性检测的主要方法流程图; [0040] FIG. 1 is a primary method described in the ad library repetitive testing flow chart of the present invention;

[0041] 图2是本发明实施例中的广告库重复性检测的具体方法流程图; [0041] FIG. 2 is a specific method described in the ad library repetitive testing flow chart of the present invention;

[0042] 图3是本发明实施例中的另一种广告库重复性检测的具体方法流程图; [0042] FIG. 3 is a specific method described in another ad libs reproducible detection flowchart present invention;

[0043] 图4是本发明实施例中的广告库重复性检测装置的主要结构示意图 [0043] FIG. 4 is a schematic diagram of the main structure of the ad libs repetitive detection apparatus embodiment of the present invention

[0044] 图5是本发明实施例中的广告库重复性检测装置的具体结构示意图。 [0044] FIG. 5 is a detailed schematic diagram of the ad libs reproducible detection apparatus embodiment of the present invention.

具体实施方式 DETAILED DESCRIPTION

[0045] 以下结合附图对本发明的优选实施例进行说明,应当理解,此处所描述的优选实施例仅用于说明和解释本发明实施例,并不用于限定本发明实施例。 [0045] The following accompanying drawings of the preferred embodiment of the present invention will be described, it should be understood that the preferred embodiments described herein are only intended to illustrate and explain the embodiments of the present invention and are not intended to limit the embodiment of the invention.

[0046] 如图1所示为本发明优选实施例中的广告库重复性检测的主要方法流程,包括: [0046] FIG. 1 shows a preferred embodiment of the invention the main process flow example ad libraries repetitive testing, comprising:

[0047] 步骤SlOl:对待检测广告进行分镜头处理。 [0047] Step SlOl: treat detect ad storyboard process.

[0048] 步骤S102:分别对分镜头处理后的每个镜头的关键帧提取图像特征。 [0048] Step S102: The key of each shot separately treated storyboard frame image feature extraction.

[0049] 步骤S103:根据提取的图像特征以及预设的特征库,判断广告库中是否已存储有待检测广告;其中,特征库中存储有广告库中每个广告的每个镜头的关键帧的图像特征。 [0049] Step S103: The image feature extraction and pre signatures to determine whether the ad library to be detected stored advertisement; where signatures are stored in the library keyframe ads each shot of each ad image features.

[0050] 其中,广告库中存储有待检测广告是指广告库中存在这样的广告:与待检测广告完全相同,或者包含与待检测广告完全相同的片段。 [0050] where advertising needs to be stored in the library means to detect the existence of such advertising Advertising Advertising library: the same ad to be detected, or a fragment of the same ad to be detected.

[0051] 本发明实施例中以镜头为单元提取广告视频的图像特征,以镜头为一个单元,可以保留镜头的运动特性和整体颜色分布特征,而一般的关键帧只有图像的空间特征,因此能够检测到镜头及镜头顺序都相同的广告,尤其可以避免将镜头相同但镜头顺序不同的广告判断为相同的广告。 [0051] Generally keyframes only the spatial characteristics of the image of the motion characteristics and overall color distribution in the lens of the image feature extraction unit commercial video to the camera as a unit, you can keep the camera embodiment of the present invention, but, it is possible to detected on the lens and lens are the same order of advertising, especially to avoid the same lens but with different lenses order advertising judged to be the same ad.

[0052] 下面通过两个典型实施例对本发明实施例中的广告库重复性检测方法进行详细说明。 [0052] The following examples by two typical examples of ad libs reproducible detection method of the present invention will be described in detail.

[0053] 如图2所示为本发明实施例中广告库重复性检测方法的详细实施例,具体地,包括以下步骤: [0053] As shown in Figure 2 is a detailed embodiment examples Ad Library reproducible detection method embodiment of the invention, in particular, comprising the steps of:

[0054] 步骤S201:预先建立特征库。 [0054] Step S201: features pre-built library.

[0055] 特征库中存储有广告库中每个广告的每个镜头的关键帧的图像特征。 [0055] The feature library stores the image feature keyframe ad library each shot each ad.

[0056] 首先将广告库中的广告进行分镜头处理,然后提取每个镜头的关键帧,对关键帧提取图像特征作为该镜头的图像特征。 [0056] First, the ad libraries ads storyboard process, then extracted key frame each shot, image feature extraction for key frame as the image feature of the lens. 常用的关键帧提取方法有对视频隔帧采样、基于帧间差提取关键帧。 Commonly used methods for extracting key frame every other frame of video sampling, extracting key frames based on frame difference. 提取的特征可以是图像的颜色、边缘、纹理、特征点、指纹等。 Extraction can be characterized by the color of the image, edge, texture, feature points, such as fingerprints.

[0057] 在本实施例中,优选的,以镜头单元的短时信息表达图像(TemporallyInformative Representative Image, TIRI)作为每个镜头的关键巾贞,并提取关键巾贞的指纹,将此指纹作为关键帧的图像特征,以此来建立特征库。 [0057] In the present embodiment, it is preferable to short-term information express image lens unit (TemporallyInformative Representative Image, TIRI) as each shot is crucial towel Ching, Ching towel and extract key fingerprint, this fingerprint as a key image feature frame, in order to establish the feature library.

[0058] 优选的,由于广告数据较大,因此还可以对特征库建立索引表。 [0058] Preferably, since the advertising data is large, it is also possible indexing table signatures. 索引是指能够从整体内容中直接查找到某个部分的目录。 Index refers to the ability to find the content directly from the whole to a part of the catalog. 在建立索引时,考虑到图像特征多为高维向量,首先对图像特征向量进行量化,产生低维的数据(通常为一个整数),量化的方法有哈希算法以及数字指纹的方法等;量化后的数值即为索引值。 When creating an index, taking into account the image feature mostly high-dimensional vector, the first image feature vector quantization, resulting in low-dimensional data (usually an integer), quantitative methods are hashing algorithm and digital fingerprint method; quantization After the value is the index value. 将图像特征按照索引值排列(即具有相同索引值的图像特征排为一行或一列),这样就形成了索引表。 The image features are arranged according to the index value (ie image features rows with the same index value for the row or column), thus forming the index table. 此外,存储图像特征时将其与对应的镜头ID以及广告ID关联保存,因此,可以从特征库中获知:存储的图像特征是哪个广告的第几个镜头的图像特征。 In addition, when storing image feature corresponding lens ID and advertisement ID associated with preservation, so you can learn from the feature library: the image is stored in the image feature characteristic which ad the first few lenses. 例如,存储图像特征时采用以下结构: For example, when the stored image features the following structure:

[0059] [0059]

Figure CN103226571AD00071

[0060] 其中video_id为广告编号,shot_id为镜头编号(镜头在广告中的顺序编号),fingerprint为图像特征。 [0060] where video_id ad number, shot_id number for the lens (the lens sequence number in the ad), fingerprint image feature.

[0061] 步骤S202:对待检测广告进行分镜头处理。 [0061] Step S202: to be detected ad storyboard process.

[0062] 对待检测广告进行镜头检测,作分镜头处理。 [0062] shot detection ads to be detected, as the storyboard process. 镜头是一组帧序列,同一个镜头中的图像帧比较接近,特征基本不变。 Is a group of lens frame sequence, the same image frame shot closer, characterized in essentially the same. 若相邻帧之间发生较剧烈的特征变化,就认为发生了镜头切换。 If occur between adjacent frames feature more intense change, it considers that a cutscene. 目前镜头检测方法主要有基于直方图、模板匹配、基于统计的方法等。 Currently shot detection methods are mainly based on the histogram matching, templates, methods based on statistics. 待检测广告由一个或多个镜头组成,将待检测广告分为以镜头为单元的视频片段,并且根据它们在待检测广告中的顺序为它们加上标号,例如依次为l,2,3...n。 Ad to be detected by a lens consisting of one or more, will be detected as advertisements into the lens unit to a video clip, and according to their advertisements to be detected in order of their reference numerals plus, for example, followed by l, 2,3. ..n. 因此,待检测广告用镜头序列[S1,S2,...Sn]来表示。 Therefore, advertising with a camera sequence to be detected [S1, S2, ... Sn] to represent. 以镜头为单元,可以保留镜头的运动特性和整体颜色分布特征,具有空间特征和时间特征,而一般的关键帧只有图像的空间特征。 With the lens as a unit, you can keep the camera motion characteristics and overall color distribution, spatial and temporal characteristics, and general keyframes only spatial characteristics of the image.

[0063] 步骤S203:确定每个镜头的关键帧。 [0063] Step S203: Identify key frames of each shot.

[0064] 在本实施例中,分别将每个镜头的TIRI作为每个镜头的关键帧。 [0064] In the present embodiment, respectively, each shot TIRI as keyframes for each shot. TIRI的优点是包含了视频的时间和空间特点,通过计算镜头内的图像的加权平均值,得到一幅融合的图像,它不仅能反映了镜头的颜色分布信息,还反映了镜头的运动特性。 TIRI advantage is the inclusion of time and space characteristics of the video, by calculating the weighted average of the image within the lens, to obtain a fusion image, it can not only reflect the color distribution information of the lens, also reflects the motion characteristics of the lens.

[0065] 令Inumk为第k帧图像在(m,η)位置的像素灰度值,TIRI图像的像素值可以由以下公式表达: [0065] for the k-th order Inumk frame image pixel gray value (m, η) position, the pixel values of the image can be expressed TIRI by the following equation:

[0066] [0066]

Figure CN103226571AD00072

[0067] 其中,N为一个镜头包括的图像帧数,ω,为加权函数。 [0067] wherein, N is a number of image frames including the lens, ω, is a weighting function. 可以采用不同的加权方法来获得TIRI图像,如常数加权、线性加权和指数加权等。 You can use different methods to obtain TIRI weighted image, such as a constant weight, linear weighted and exponentially weighted like. 在本实施例中选择指数加权函数Oj5 — Y , Y — 1.1。 Select the index weighting function Oj5 in this embodiment - Y, Y - 1.1.

[0068] 步骤S204:分别对每个镜头的关键帧提取图像特征。 [0068] Step S204: The key for each frame of each shot image feature extraction.

[0069] 在本实施例中,对镜头的TIRI图像提取图像指纹作为镜头的图像特征。 [0069] In this embodiment, the lens of TIRI image as the image feature extraction Fingerprint lens.

[0070] 首先,将TIRI图像分成Μ*Ν块,每一块的大小为2w*2w(w为设置的窗口大小),计算每一块的水平方向和垂直方向的DCT系数。 [0070] First, the TIRI Μ * Ν image into blocks, each block of size 2w * 2w (w is set window size), calculated for each block in the horizontal direction and the vertical direction of the DCT coefficients.

[0071] 水平方向DCT系数为:β u = I1BijjV ; [0071] horizontal DCT coefficients: β u = I1BijjV;

[0072] 垂直方向DCT系数为叫,」=V1BijjI ; [0072] vertical DCT coefficient is called, "= V1BijjI;

[0073] 其中I为单元向量,Bijj为图像块,V为余弦向量: [0073] where I is the unit vector, Bijj image block, V cosine vector:

[0074] V= [cos (0.5 η /2w), cos (1.5 π /2w),...cos (J1-0.5 π /2w) ]τ [0074] V = [cos (0.5 η / 2w), cos (1.5 π /2w),...cos (J1-0.5 π / 2w)] τ

[0075] 其次,将以上得到的Μ*Ν*2个系数按顺序组成一个向量f,对f进行量化。 [0075] Next, the above obtained Μ * Ν * 2 coefficients in order to form a vector f, of f quantify. 量化方法有很多种,例如可以是取向量的平均值或中值。 There are many methods to quantify, for example, may be oriented average or median amount. 本实施例中取向量的平均值a,将f的每一维与a比较,得到一个由0,I组成的二值向量f',即为图像的指纹特征: Examples orientation of the average amount of the present embodiment a, the f each dimension compared with a, get a made 0, I composed a binary vector f ', is the image of the fingerprint characteristics:

Figure CN103226571AD00081

[0077] 步骤S205:在预设的特征库中分别查找提取的每个图像特征所匹配的图像特征。 [0077] Step S205: Find image features are extracted from each image feature matched preset feature library.

[0078] 可以通过计算相似度的方法在特征库中查找匹配的图像特征。 [0078] You can find the matching image features in the feature library by the method of computing the similarity.

[0079] 如步骤201中所述,优选的,特征库中的图像特征以索引表的形式存储。 [0079] As described in step 201, it is preferable, features a library of image features in the form of index table storage. 因此,相应的,在本步骤中,首先将待检测广告的图像特征编码为一个整数作为索引值(应保证与特征库的索引表采用同样的编码方法),然后在索引表中查找该索引值对应的图像特征,与每个查找到的具有相同索引值的图像特征进行相似计算,并将最相似的索引项作为匹配结果O Therefore, the corresponding, in this step, first, the picture will be detected ad features encoded as an integer value as an index (the index table should ensure signatures using the same encoding method), and then look for the index values in the index table corresponding image features, and each look to image features with the same index values were similar calculations and index entries as the most similar match results O

[0080] 由于指纹特征是由一系列O和I组成,因此可采用汉明距离来计算相似度。 [0080] Since the fingerprint is characterized by a series consisting of O and I, and therefore can be used to calculate the similarity Hamming distance. 假设指纹特征的向量的长度为N,两个向量相同位置上的值相同的个数M,则两个向量的相似度 Suppose fingerprint feature vector of length N, the number of the same value M at the same position on the two vectors, the similarity of the two vectors

比值 Ratio

Figure CN103226571AD00082

当s大于预设阈值T时,认为两个指纹特征匹配,本实施例中T取0.9。 When s is greater than a predetermined threshold T, characterized in that the two fingerprints match, this embodiment T 0.9.

[0081] 步骤S206:判断对于待检测视频的每个图像特征是否均查找到匹配的图像特征,若是,执行步骤S207,若否,执行步骤S210。 [0081] Step S206: judging for each image feature to be detected whether the video image features are to find matches, if so, step S207, and if not, to step S210.

[0082] 步骤S207:判断匹配的图像特征是否属于同一广告,若是,执行步骤S208,若否,执行步骤S210。 [0082] Step S207: Analyzing image feature matching belong to the same ad, and if so, to step S208, and if not, to step S210.

[0083] 从特征库中查找到的图像特征为: [0083] Find feature library to the image features are:

Figure CN103226571AD00083

[0085] 通过读取其中的videojd获取该图像特征所属的广告编号,同一广告的图像特征具有相同的广告编号。 [0085] videojd get by reading one of the characteristics of the image ad number belongs to, the image feature the same ad with the same number of ads.

[0086] 步骤S208:判断匹配的图像特征对应的镜头的顺序是否与待检测广告的镜头顺序一致,若是,执行步骤S209,若否,执行步骤S210。 [0086] Step S208: Analyzing image feature matching sequence corresponding lens coincides with the lens to be detected order advertising, and if so, to step S209, and if not, to step S210.

[0087] 根据图像特征 [0087] The image features

Figure CN103226571AD00084

中的shot_id可以确定每个图像特征所对应的 The shot_id may be determined for each image feature corresponding

镜头编号。 Lens number. 可以将匹配的图像特征按照其匹配的待检测广告的图像特征的顺序排列,若排列后的图像特征的镜头编号为连续递增时,则表示连续匹配的图像特征所对应的镜头顺序与待检测广告的镜头顺序一致。 Images can be in the order of image feature matching features its matching advertising arrangement to be detected, if the arrangement of the camera after the image feature numbered continuously increasing, it represents a continuous matching of image characteristics corresponding to the sequence to be detected lens advertisement consistent shot sequence.

[0088] 例如,当待检测广告分为3个镜头时,若排列后的图像特征对应的镜头编号为shot_3、shot_4、shot_5,则与待检测广告的镜头顺序一致,即待检测广告与该广告中的shot_3、shot_4、shot_5组成的片段完全一致;若排列后的图像特征对应的镜头编号为sh0t_3、sh0t_4、sh0t_6,则与待检测广告的镜头顺序不一致(比待检测广告的后两个镜头之间多出一个镜头);若排列后的图像特征对应的镜头编号为shot_3、shot_5、shot_4,贝U与待检测广告的镜头顺序不一致(后两个镜头的先后顺序不一样)。 [0088] For example, when advertising to be detected is divided into three lens, the image feature if arranged after the corresponding lens number is shot_3, shot_4, shot_5, it is consistent with the sequence to be detected shot advertising that advertising be detected with this ad The shot_3, shot_4, exactly shot_5 fragment thereof; if arranged after the image feature corresponding lens number is sh0t_3, sh0t_4, sh0t_6, then be detected inconsistencies order advertising shots (to be detected than after two shots of advertising among more than a shot); if arranged after image characteristics corresponding lens number is shot_3, shot_5, shot_4, Tony U be detected inconsistencies order advertising shots (two shots after the order is not the same).

[0089] 步骤S209:确认广告库中已存储有待检测广告。 [0089] Step S209: Confirm Ad Library has been stored to be detected ad. [0090] 步骤S210:确认广告库中未存储有待检测广告。 [0090] Step S210: Confirm advertising is not stored in the library to be detected ad.

[0091] 通过本发明实施例,可以检测出广告库中是否存在与待检测广告完全相同的广告,或者是否存在包含与待检测广告完全相同的片段的广告。 [0091] The embodiments of the present invention can detect the existence of ad library to be detected with identical ads advertising, or whether there is to be detected fragment contains the same advertising ads.

[0092] 优选的,还可以根据上述确认结果执行以下步骤: [0092] Preferably, you can perform the following steps to confirm the above results:

[0093] 当确认广告库中未存储有待检测广告时,将待检测广告存储至广告库中。 [0093] When it is confirmed ad library to be detected is not stored advertising, store advertising will be tested to the ad library.

[0094] 在本发明的另一个实施例中,还考虑到以下情况:当广告库中未存储有待检测广告,但待检测广告中包含广告库中的广告。 [0094] In another embodiment of the present invention, it is also taken into account the following circumstances: When an ad is not stored in the library to be detected ad, but to be detected advertisement ad library contains advertising. 如图3所示,该实施例包括以下步骤: As shown in Figure 3, this embodiment comprises the steps of:

[0095] 步骤S301:预先建立特征库。 [0095] Step S301: features pre-built library.

[0096] 步骤S302:对待检测广告进行分镜头处理。 [0096] Step S302: to be detected ad storyboard process.

[0097] 步骤S303:确定每个镜头的关键帧。 [0097] Step S303: Identify key frames of each shot.

[0098] 步骤S304:分别对每个镜头的关键帧提取图像特征。 [0098] Step S304: The key for each frame of each shot image feature extraction.

[0099] 步骤S305:在预设的特征库中分别查找提取的每个图像特征所匹配的图像特征。 [0099] Step S305: Find image features are extracted from each image feature matched preset feature library. [0100] 步骤S306:判断对于待检测视频的每个图像特征是否均查找到匹配的图像特征,若是,执行步骤S307,若否,执行步骤S309。 [0100] Step S306: judging for each image feature to be detected whether the video image features are to find matches, if so, step S307, and if not, to step S309.

[0101] 步骤S307:判断匹配的图像特征是否属于同一广告,若是,执行步骤S308,若否,执行步骤S309。 [0101] Step S307: Analyzing image feature matching belong to the same ad, and if so, to step S308, and if not, to step S309.

[0102] 步骤S308:判断匹配的图像特征对应的镜头的顺序是否与待检测广告的镜头顺序一致,若是,执行步骤S313,若否,执行步骤S314。 [0102] Step S308: Analyzing image feature matching lens corresponding sequence to be detected is consistent with the advertising sequence shot, and if so, to step S313, and if not, to step S314.

[0103] 匹配的图像特征的排列顺序是与待检测广告的镜头顺序一致的(按照匹配关系对应)。 Order of image features [0103] matching is consistent with the sequence to be detected ad lenses (in matching relation counterpart). 获取匹配的图像特征对应的镜头的顺序,判断匹配的图像特征对应的镜头的顺序是否与待检测广告的镜头顺序一致。 Get matching image feature sequence corresponding lens to determine the matching image features corresponding to the order of the lens to be detected is consistent advertising shot sequence.

[0104] 步骤S309:将匹配的图像特征按所属广告划分。 [0104] Step S309: the image feature will match the press's ad division.

[0105] 步骤S310:判断划分后的图像特征是否是其所属广告的全部图像特征,若是,执行步骤S311,若否,执行步骤S314。 [0105] Step S310: the image feature determines whether the divided feature is their respective full image ads, and if so, step S311, and if not, to step S314.

[0106] 步骤S311:判断划分后的图像特征所对应的镜头的顺序是否连续递增,若是,执行步骤S312,若否,执行步骤S314。 [0106] Step S311: Analyzing image features corresponding to the divided lens sequence is continuously increasing, and if so, to step S312, and if not, to step S314.

[0107] 步骤S312:将待检测广告存储至广告库,并将广告库中的广告A删除。 [0107] Step S312: to be tested to ad libs store advertisements and advertising ad library A erased.

[0108] 步骤S313:不将待检测广告存储至广告库。 [0108] Step S313: not to be tested ad store to ad libs.

[0109] 步骤S314:将待检测广告存储至广告库。 [0109] Step S314: store advertising to be tested to ad libs.

[0110] 本实施例能够将待检测广告存储至广告库的同时,将广告库中包含于待检测广告的广告删除,避免广告库中存在重复的广告。 At the same time [0110] This embodiment can detect ad will be stored to the ad library library contains the ad to be detected ad delete, avoid ad library duplicate ads.

[0111] 如图4示为本发明实施例中的广告库重复性检测装置,包括: [0111] FIG. 4 shows the present case, the ad libs reproducible detection device invention, comprising:

[0112] 分镜头处理模块401,用于对待检测广告进行分镜头处理; [0112] storyboard processing module 401, used to treat detect ad storyboard process;

[0113] 提取图像特征模块402,用于分别对分镜头处理后的每个镜头的关键帧提取图像特征; [0113] image feature extraction module 402, for each key for each shot storyboard processed frame image feature extraction;

[0114] 判断模块403,用于根据提取的图像特征以及预设的特征库,判断广告库中是否已存储有待检测广告;其中,特征库中存储有广告库中每个广告的每个镜头的关键帧的图像特征。 [0114] judging module 403 for the image feature extraction and pre signatures to determine whether the ad library to be detected stored advertisement; where signatures are stored Ad Library each shot of each ad image feature keyframes.

[0115] 优选的,判断模块403包括:[0116] 查找单元,用于在预设的特征库中查找与提取的图像特征匹配的图像特征; [0115] Preferably, the judging module 403 includes: [0116] search unit, is used to find the image feature extraction and image feature matching preset features a library;

[0117] 第一判断单元,用于判断对于待检测视频的每个图像特征是否均查找到匹配的图像特征; [0117] The first determination unit for determining image characteristics for each image feature to be detected whether the video to find all matching;

[0118] 第二判断单元,用于当对于待检测视频的每个图像特征均查找到对应的匹配的图像特征时,判断匹配的图像特征是否属于同一广告以及匹配的图像特征对应的镜头的顺序与待检测广告的镜头的顺序是否一致; Sequence [0118] second judgment unit for, when the image feature to be detected for each video image features are to find the corresponding matching, the image feature matching to judge whether belonging to the same advertising and image feature matching corresponding lens It is consistent with the sequence to be detected is advertising the lens;

[0119] 确认单元,用于当匹配的图像特征属于同一广告并且匹配的图像特征对应的镜头的顺序与待检测广告的镜头的顺序一致时,确认广告库中已存储有待检测广告。 [0119] confirmation unit for image feature when matching belong to the same advertising and the image feature matching corresponding sequence shot sequence to be detected is consistent with the advertising of the lens, confirm Ad Library has been stored to be detected ad.

[0120] 优选的,查找单元用于: [0120] Preferably, the search unit for:

[0121] 将提取的图像特征编码生成索引值; [0121] The extracted image features code generation index value;

[0122] 在特征库中查找具有所述索引值的图像特征; [0122] to find the image feature having the index value in the feature library;

[0123] 计算查找到的图像特征与提取的图像特征之间的相似度; [0123] calculating the similarity find the image feature extraction and image feature between;

[0124] 将查找到的与提取的图像特征之间的相似度大于预设阈值的图像特征,作为与提取的图像特征匹配的图像特征。 [0124] will find similarity with the image feature extraction is greater than a predetermined threshold of image features, as the image feature extraction and image feature match. [0125] 优选的,如图5所示,所述装置还包括: [0125] Preferably, as shown in FIG. 5, the apparatus further comprising:

[0126] 存储模块404,用于当广告库中未存储有待检测广告时,将待检测广告存储至广告库。 [0126] storage module 404 for when the ad is not stored in the library to be detected advertising, store advertising will be tested to ad libs.

[0127] 优选的,如图5所示,所述装置还包括: [0127] Preferably, as shown in FIG. 5, the apparatus further comprising:

[0128] 删除模块405,用于当广告库中未存储有待检测广告,而待检测广告中包含广告库中的广告时,将广告库中包含于待检测广告的广告删除。 When the [0128] deleting module 405 for advertising to be detected is not stored in the library when advertising, and advertising to be detected ad library contains advertising, the advertising contained in the library to be detected ad removed.

[0129] 本领域内的技术人员应明白,本发明的实施例可提供为方法、系统、或计算机程序产品。 [0129] Those skilled in the art should understand, embodiments of the present invention may provide a method, system, or computer program product. 因此,本发明可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。 Accordingly, the present invention may be entirely hardware embodiment, an entirely software embodiment, or a combination of forms of embodiment of software and hardware aspects. 而且,本发明可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器和光学存储器等)上实施的计算机程序产品的形式。 Furthermore, the present invention can be used in one or more of which contains a computer usable program code, computer usable storage media (including, but not limited to, disk storage and optical storage, etc.), implemented on a computer program product.

[0130] 本发明是参照根据本发明实施例的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。 [0130] The present invention has been described in accordance with a method embodiment of the present invention, apparatus (systems), and flow computer program products and / or block diagram to describe. 应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。 It should be understood by the computer program instructions, and a combination of the flowchart and / or block diagram each of the processes and / or block flow and / or block diagram of the process and / or box. 可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器,使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。 These computer program instructions can be provided to a general purpose computer, special purpose computer, embedded processor or other programmable data processing apparatus to produce a machine, such that the instructions executed by a computer or other programmable data processing apparatus generating In the apparatus for implementing a process flow diagram or more processes and / or block diagram block or blocks a specified function.

[0131] 这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的计算机可读存储器中,使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品,该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。 [0131] These computer program instructions may also be stored in a computer can boot the computer or other programmable data processing apparatus to function in a particular manner readable memory so that stored in the computer readable instructions in the memory to produce articles of manufacture including instruction means The instruction means implemented in a process flow diagram or more processes and / or block diagram block or blocks a specified function.

[0132] 这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上,使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理,从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的步骤。 [0132] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus, so that the implementation of a series of steps on the computer or other programmable apparatus to produce a computer implemented, resulting in a computer or other programmable apparatus Instruction is provided on the implementation of a process for implementing the flowchart or more processes and / or block diagram of a block or blocks functions specified steps.

[0133] 显然,本领域的技术人员可以对本发明进行各种改动和变型而不脱离本发明的精神和范围。 [0133] Obviously, those skilled in the art of the present invention may be various changes and modifications without departing from the spirit and scope of the invention. 这样,倘若本发明的这些修改和变型属于本发明权利要求及其等同技术的范围之内,则本发明也意图包含这些改动和变型在内。 Thus, if such modifications and variations of the present invention belongs to the claims of the invention and their equivalents technology range, the present invention is also intended to encompass such changes and modifications included.

Patentzitate
Zitiertes PatentEingetragen Veröffentlichungsdatum Antragsteller Titel
CN101162470A *16. Nov. 200716. Apr. 2008北京交通大学Video frequency advertisement recognition method based on layered matching
CN102890778A *21. Juli 201123. Jan. 2013北京新岸线网络技术有限公司Content-based video detection method and device
US8079054 *14. Apr. 200813. Dez. 2011Adobe Systems IncorporatedLocation for secondary content based on data differential
US20080126191 *8. Nov. 200629. Mai 2008Richard SchiaviSystem and method for tagging, searching for, and presenting items contained within video media assets
Referenziert von
Zitiert von PatentEingetragen Veröffentlichungsdatum Antragsteller Titel
CN104185060A *26. Febr. 20143. Dez. 2014无锡天脉聚源传媒科技有限公司Video de-duplication method and device
CN104185060B *26. Febr. 20147. Juli 2017无锡天脉聚源传媒科技有限公司一种视频排重的方法及装置
Klassifizierungen
Internationale KlassifikationG06F17/30
Juristische Ereignisse
DatumCodeEreignisBeschreibung
31. Juli 2013C06Publication
28. Aug. 2013C10Entry into substantive examination
18. Jan. 2017C12Rejection of a patent application after its publication