CN1317889C - A digital video watermark detection method based on independent component analysis - Google Patents
A digital video watermark detection method based on independent component analysis Download PDFInfo
- Publication number
- CN1317889C CN1317889C CNB200410036047XA CN200410036047A CN1317889C CN 1317889 C CN1317889 C CN 1317889C CN B200410036047X A CNB200410036047X A CN B200410036047XA CN 200410036047 A CN200410036047 A CN 200410036047A CN 1317889 C CN1317889 C CN 1317889C
- Authority
- CN
- China
- Prior art keywords
- video
- watermark
- frame
- component analysis
- main
- 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
Landscapes
- Image Processing (AREA)
- Editing Of Facsimile Originals (AREA)
Abstract
The present invention belongs to the technical field of video and multimedia signal processing. The present invention has the technical scheme that an original uncompressed video to be embedded with a watermark is divided by shots to obtain different segments when the watermark is embedded; the components of frames in each segment are separately decomposed; main separate component components are selected from signals obtained by decomposition and are embedded into the watermark; the separate components are analyzed and inversely transformed to obtain a video embedded with the watermark. When the watermark is detected, a video to be detected just like embedding the watermark is divided by shots, the separate components are analyzed to obtain the main separate components, and then, the watermark is extracted. The present invention overcomes the defects of complex step, poor robustness, etc. in the prior art and has the advantages of high efficiency, good robustness, etc.
Description
(1) technical field
The present invention relates to a kind of video digital watermark method, belong to vision signal processing, multimedia signal processing technique field based on independent component analysis.
(2) technical background
Along with multimedia technology and rapid development of network technology and their popularizing in real society, the multimedia digital product is produced more and more easily, obtains, is duplicated and propagates, when having made things convenient for validated user to use, also caused serious piracy, impinge issue, this just makes the copyright problem of multimedia digital product become urgent problem.Digital watermark technology puts forward at this problem, is the main mode that solves copyright problem now.Digital watermarking is exactly under situation about not discovered by the people, in media product, hide or embed the information that can represent copyright, then by to the extraction of copyright information, confirm to reach authentication and protection for the digital product copyright.The most basic, the topmost requirement of digital watermarking is exactly a robustness, i.e. watermark still can correctly extract, detect through after the common signal processing; Even under the situation of malice or deliberately attack, can both correctly extract and detect.
Digital video watermarking has become the emphasis of Digital Watermarking Theory research and practical application now, video stationary problem in time be in the video watermark inevitably and will at first need the problem considered, whether it directly influences watermark can correctly detect, and most watermarking projects require that all there is good robustness in the asynchronism(-nization) step.Therefore, how to improve technological difficulties and the emphasis that the robustness in asynchronism(-nization) step is also just become present video watermark research.Existing video watermark process way to solve the problem hereto generally has two kinds: when (1) embeds in watermark, set up a time template according to the characteristic of video, also be embedded into this template in the video; When detecting, reach the synchronous again of video by the template that detects.(2) when watermark embeds,, in each frame, periodically embed the watermark of particular state according to the sequential of frame of video; When detecting, by to the determining of watermark state, come the positioning video frame, reach in time synchronous again of video.All there is same shortcoming in these two kinds of methods, when promptly carrying out watermark detection, need position complex steps to frame of video; Though and if frame of video change, the situation that the content of whole video remains unchanged (changing as frame per second), these two kinds of methods just all can not correctly be carried out synchronous on the video time, also just can not correctly extract watermark.
(3) summary of the invention
In order to overcome the shortcoming that prior art exists, the present invention proposes a kind of video watermark process based on independent component analysis, utilize independent component analysis to extract the content characteristic of video, watermarked on these features, improved water mark method greatly for the asynchronism(-nization) robustness in step, and simplified the embedding and the testing process of watermark, improved efficient.
The present invention mainly realizes that by watermarked and two processes of extraction watermark concrete steps are as follows:
(1) watermarked
In watermarked, earlier adopt the method for overall histogram comparison to carry out camera lens the video of unpressed any frame per second, arbitrary frame size and cut apart; Poor according to the frame that obtains, determine segmentation threshold, cut apart and obtain video-frequency band.The formula that calculates frame difference institute foundation is:
Wherein (i, i+1) frame between expression i frame and the i+1 frame is poor, H for D
1(j), j=1 ..., n represents the number of pixels of j gray scale of i frame, and n is total number of greyscale levels, and P is that the total pixel in the frame is counted; For color framing, can adopt the Y component in the yuv format to carry out histogrammic comparison.Frame in each video-frequency band is changed into vectorial form, these vectors are carried out prewhitening handle; Adopt quick independent component analysis rule to decompose for the vector that obtains after the albefaction, obtain isolated component, and therefrom select main isolated component; These principal components are regarded as image, on these components, carry out the embedding of watermark, thereby obtain watermarked principal component.These principal components that embedded watermark are changed back vector form again, replace original main isolated component, carry out quick independent component analysis inverse transformation then and just obtained watermarked video.
(2) extract watermark
When extracting watermark, earlier with unpressed video to be detected, carry out camera lens and cut apart, video is divided into video-frequency band.Utilize quick independent component analysis rule, these sections are carried out main independent component analysis, obtain main isolated component.From these components, carry out the extraction of watermark, obtain detected watermark.
Adopt the present invention, can choose corresponding camera lens dividing method at the characteristics of video scene and obtain segmentation effect preferably; When carrying out the independent component analysis feature extraction, also can choose the feature that suitable independent component analysis rule obtains more embodying video at the characteristics of frame of video and the requirement of watermarking project; When carrying out watermark and embedding, can obtain being suitable for the video watermark process of concrete condition requirement according to the embedding that different modes is carried out in the application scenario of video watermark, the requirement that will reach etc.The present invention has robustness preferably to most of common watermark attacks and Video processing, and when the video time of occurrence is nonsynchronous, does not need to carry out the location of single frames, and efficient is higher, robustness is better.
(4) description of drawings
Fig. 1 is that watermark embeds block diagram.
Fig. 2 is the watermark extracting block diagram.
Fig. 3 is an original watermark.
Fig. 4 is that the frame of every adjacent two frames in the test video is poor.
Fig. 5 is the frame in each section in the original video.
Fig. 6 is the frame of each section in the watermarked video.
Fig. 7 is the watermark that extracts in each section of video under the various attack.
(5) embodiment
Embodiment
Select for use one section advertisement that three camera lenses are arranged as the test video, frame per second is 25fps, and frame sign is 272 * 352, and totalframes is 73 frames.Watermark is the bianry image that Shandong University's printed words are arranged, and size is 32 * 32, as accompanying drawing 1.
Earlier adopting the method for overall histogram comparison to carry out camera lens this section video cuts apart.According to frame difference formula, it is poor to calculate frame, as accompanying drawing 2, determines that segmentation threshold is 0.2, cuts apart to obtain three video-frequency bands.Owing to be color framing, select for use the Y component in the yuv format to carry out histogrammic comparison here.
Frame in each video-frequency band is changed into vectorial form, and vector length is 272 * 352=95744, adopts principal component analysis to carry out prewhitening these vectors and handles.The vector that albefaction obtains adopts quick independent component analysis rule to decompose, and selects main isolated component from the isolated component that obtains, and converts the form of principal component frame to, and the ranks number is respectively 272 and 352.
On these principal component frames, adopt the SWE rule to carry out the embedding of watermark, thereby obtain watermarked principal component.Replace original main isolated component with these principal components, carry out quick independent component analysis inverse transformation then and just obtained watermarked video.Original frame of video and watermarked frame of video are respectively as accompanying drawing 3, accompanying drawing 4.
Extract watermark under the various situations below, and all after having passed through the MPEG2 encoding and decoding, video carries out: (a) 3 * 3 medium filterings, (b) frame size dwindles 2 times, (c) frame is rotated counterclockwise 0.5 degree, and (d) density is 0.02 salt-pepper noise, and (e) average is 0, variance is 0.001 Gaussian noise, (f) collusion attack of each video-frequency band inside (g) is passed through MPEG4 encoding and decoding LOF again, (h) in the never watermarked original video; The attack in asynchronism(-nization) step: (i) frame of each video-frequency band inside exchange, optional 2 consecutive frames exchange in per 5 frames, (j) lose a frame arbitrarily in per 6 frames, (k) extract frame per second from 25fps to 30fps, duration is constant, and frame number increases to 85 from 73, and (l) frame per second is from 25fps to 24fps, duration is constant, and frame number is kept to 69 from 73.
The watermark that extracts is listed in accompanying drawing 5.The watermark that extracts under respectively corresponding (a)～(l) situation of accompanying drawing 5 (a)～(l).(a)～(l) three watermarks in each subgraph, the watermark that from first, second and third camera lens, extracts after corresponding respectively from left to right every kind of attack.
In the top example, in the video of watermark was arranged, testing result can both reproduce original watermark well.This has just illustrated the detection that watermark can both be correct under the situation that various common processing, attack and asynchronism(-nization)s go on foot, and robustness is fine; Do not having under the situation of watermark, the situation that flase drop is surveyed can not occur.And, so just fundamentally prevented the generation of the copy-attack in the video because method of the present invention when watermark embeds, has been carried out the feature extraction based on video content.
Claims (1)
1, a kind of digital video watermarking method based on independent component analysis is characterized in that concrete steps are as follows:
(1) watermarked
In watermarked, with the video of unpressed any frame per second, arbitrary frame size, carry out camera lens and cut apart, video is divided into different sections; Frame in each section is carried out isolated component decompose, from the isolated component that obtains, choose main isolated component composition, and convert the form of principal component frame to, watermarked on these principal component frames; Become to assign to replace original corresponding main isolated component composition with these main isolated components that embedded watermark, carry out the independent component analysis inverse transformation again, obtained embedding the video of watermark;
(2) extract watermark
When extracting watermark, earlier uncompressed video to be detected is carried out camera lens and cut apart, obtain video-frequency band, each video-frequency band is carried out main independent component analysis, obtain main isolated component; From the main isolated component that obtains, extract watermark.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB200410036047XA CN1317889C (en) | 2004-11-02 | 2004-11-02 | A digital video watermark detection method based on independent component analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB200410036047XA CN1317889C (en) | 2004-11-02 | 2004-11-02 | A digital video watermark detection method based on independent component analysis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1604635A CN1604635A (en) | 2005-04-06 |
| CN1317889C true CN1317889C (en) | 2007-05-23 |
Family
ID=34664385
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB200410036047XA Expired - Fee Related CN1317889C (en) | 2004-11-02 | 2004-11-02 | A digital video watermark detection method based on independent component analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1317889C (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4398943B2 (en) * | 2006-01-20 | 2010-01-13 | 株式会社東芝 | Digital watermark detection apparatus, digital watermark detection method, and digital watermark detection program |
| US10169838B2 (en) | 2016-08-01 | 2019-01-01 | International Business Machines Corporation | Multiple source watermarking for surveillance |
| CN112634116A (en) * | 2020-12-02 | 2021-04-09 | 四川大学 | Semi-fragile digital watermark embedding and extracting method based on FastICA |
-
2004
- 2004-11-02 CN CNB200410036047XA patent/CN1317889C/en not_active Expired - Fee Related
Non-Patent Citations (5)
| Title |
|---|
| 一种新颖的基于非压缩数字视频的水印盲检测算法 愈能海，赵卓，曹楠楠，刘政凯，电路与系统学报，第8卷第3期 2003 * |
| 基于独立分量分析德盲视频水印方案 孙建德，刘琚，电子学报，2004年第9期，第32卷第9期 2004 * |
| 基于独立分量分析的数字图像水印研究 林月娟，黄静霞，许慰玲，微机发展，第14卷第8期 2004 * |
| 独立元分析及其在视频分割中的应用 詹?赵杰煜，汪增福，计算机仿真，第21卷第6期 2004 * |
| 独立元分析及其在视频分割中的应用 詹?赵杰煜，汪增福，计算机仿真，第21卷第6期 2004;基于独立分量分析德盲视频水印方案 孙建德，刘琚，电子学报，2004年第9期，第32卷第9期 2004;基于独立分量分析的数字图像水印研究 林月娟，黄静霞，许慰玲，微机发展，第14卷第8期 2004;一种新颖的基于非压缩数字视频的水印盲检测算法 愈能海，赵卓，曹楠楠，刘政凯，电路与系统学报，第8卷第3期 2003 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1604635A (en) | 2005-04-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Meghanathan et al. | Steganalysis algorithms for detecting the hidden information in image, audio and video cover media | |
| Sherly et al. | A compressed video steganography using TPVD | |
| Tabassum et al. | A digital video watermarking technique based on identical frame extraction in 3-Level DWT | |
| CN103152578B (en) | H.264 video watermark based on mixed encoding and decoding embeds and extracting method | |
| CN108682425B (en) | Robust digital audio watermark embedding system based on constant watermark | |
| CN101950405B (en) | Video content-based watermarks adding method | |
| KR20020071927A (en) | Method and arrangement for embedding a watermark in an information signal | |
| CN105657431A (en) | Watermarking algorithm based on DCT domain of video frame | |
| CN1241389C (en) | Object-oriented self-synchronizing water mark detecting method for multimedia product | |
| CN1317889C (en) | A digital video watermark detection method based on independent component analysis | |
| CN102413381B (en) | Video watermark based on optical flow method and digital holographic | |
| CN1655616A (en) | Audio-embedded video frequency in audio-video mixed signal synchronous compression and method of extraction | |
| CN114363719B (en) | Content self-adaptive robustness video watermarking method based on Bessel-Fourier moment | |
| CN101685496B (en) | Video segment duplication detecting method | |
| Shukla et al. | Overview of Scene Change Detection-Application to Watermarking | |
| CN114168908A (en) | Copyright protection technology based on audio and video analysis | |
| CN102065287A (en) | T-shaped video watermark embedding and extracting method | |
| Agung et al. | Video scene characteristic detection to improve digital watermarking transparency | |
| Wu et al. | Steganalysis of msu stego video based on discontinuous coefficient | |
| CN110827188A (en) | Color image rapid blind digital watermarking method and extraction method | |
| Zhang et al. | A new robust watermarking algorithm based on intra-frame difference | |
| Amor et al. | A block artifact distortion measure for no reference video quality evaluation | |
| Deshpande | Robust DCT based video watermarking algorithms for assorted watermarks | |
| Liu et al. | An improved spatial spread-spectrum video watermarking | |
| KR101025311B1 (en) | Robust high definition video watermarking system based on self synchronizing siginal against composite distortions |
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 | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20070523 Termination date: 20091202 |