Video Watermarking Algorithm For Anti-Frame Operations And Subtitle Tampering

With the development of network technology,the video transmission,storage and modification have become more and more convenient,which makes illegal video tampering easier,so the copyright protection of digital video has become pressing.Therefore,robustness to attacks of frame operations and subtitle tampering is becoming more and more important.The proposed watermarking algorithms in this dissertation are used to protect the oral history video,in which Chinese subtitles of the video are expressed by interviewer oral narrative.The main contributions are listed as following:(1)In the video digital watermarking algorithms,the frame operations include the frame insertion,recoding,frame rate changing and so on.To resist these attacks,the frame location algorithm based on Hamming code correction is proposed.This approach guarantees that the watermark information can be extracted by identifying watermarked frame.First,we use the simple linear iterative clustering(SLIC)method for video frames to extract its natural characteristics X(X is a 7-bit sequence with 0 and 1).Second,this approach takes the advantage of(7,4)Hamming code correction to modify one bit and form three bit targeting code S.Third,the copyright watermark is embedded into the location frames using discrete cosine transform-singular value decomposition(DCT-SVD)method.Experimental results show that this approach is effective to resist attacks of frame dropping,frames insertion and recoding.(2)The robust video double watermark algorithm based on speeded-up robust features(SURF)features is proposed,which can resist screen recording attacks and subtitle tampering attacks.The first watermark embedding process has three steps:1)extracting the SURF feature points from the non-subtitle region;2)SVD decomposition is executed in the luminance channel of video frame,and quantify the singular value after decomposition;3)embedding watermark is completed by modifying the singular values according to the Sudoku reference table.The first embedding strategy can not only resist the screen recording attack but also verify the watermark information to judge whether the video is attacked or not.The second watermark embedding process has three steps:1)extracting texture feature of subtitle region with the classical structural tensor;2)removing scatter points by morphological "opening and closing" operation;3)the words are obtained by an improved OTSU method and embedded in non-subtitle region using DCT-SVD decomposition method.The experimental results show that the second watermark strategy for video subtitles is robust to detect subtitle tampering.