Research On Provable Data Possession Technology For Video File

With the popularity of smartphones and digital cameras,the popularity of short videos,and the spread of massive repetitive videos on the Internet,the number of digital media files such as pictures and videos that users need to store is developing rapidly.To reduce local storage costs,users usually choose to outsource local digital files to the cloud storage servers.Users need to verify whether the server saves its outsourced documents to ensure the integrity of outsourced documents without downloading the whole outsourced file.The massive duplicate videos in the cloud storage server promote the development of video duplication detection and deduplication technology,which can detect duplicate videos and keep only one of the multiple duplicate videos in the cloud storage server.With the popularization of video editing software,a large number of videos obtained by users from the network are released again after a series of simple secondary editing.Therefore,the scope of video de-duplication technology in recent years has expanded from the same videos to perceptually similar videos.Ordinary users allow the server to store only similar files to the original file.However,for other types of users,such as medical imaging,remote sensing,and other fields,their files usually are original,and cannot be tampered with.If the file is tampered with,this type of user hope to be able to locate the tampering.Therefore,there are the above two requirements for provable data possession protocol about video files,but there are few publicly reported research results related to video file provable data possession protocol.This thesis takes data possession proof about video files as the research objects,the provable data possession protocols are proposed for similar video scenes and locate tampering scenes,and explained both the two schemes in terms of security,storage overhead,time overhead,etc.The specific works are as follows:(1)Provable data possession protocol for video is proposed based on random LocationAware hash aiming at the problem that traditional data Possession Proof cannot perform on similar video verification.To solve this,video feature extraction and similarity discrimination technology are studied.This PDP protocol randomly selects some random blocks in video frames,discrete cosine transform is used on these random blocks to extract local key features of the video frame and cooperates with the global features created by the gray level co-occurrence matrix to create verification labels.During the challenge stage,the client uses the position of the random block as a challenge,the verification stage compares the correlation coefficient thresholds and judging similarity.A provable data possession proof protocol can be obtained,which achieves the purpose of extending the provable data possession protocol to the field of similar videos.After experimental analysis,the scheme is safe,can judge similar videos,and has good communication overhead,computational overhead,and storage overhead.(2)Provable data possession protocol for video is proposed based on fragile watermarking and digital signature technology aiming at meeting the demand for outsourced video in the fields of medical imaging and remote sensing.To solve this,chaos mapping,fragile digital watermarking algorithm,and digital signature technology are studied.The protocol uses the digital signature of the video frame as a provable data possession label and uses the video frame with watermark information to locate the tampering,with the help of the sensitivity of the watermark,to detect whether cloud files are strictly consistent with user source files.The experimental data shows that the total storage cost of the image and digital signature after the vulnerability watermark is embedded is much smaller than the original image.This protocol can accurately detect and locate tampering,with good accuracy and storage cost.