Font Size: a A A

Research Of Compression Domain Video Keyframe Extraction Technology Based On Dual Judgement Rule

Posted on:2014-07-27Degree:MasterType:Thesis
Country:ChinaCandidate:X L ZhangFull Text:PDF
GTID:2268330392464293Subject:Circuits and Systems
Abstract/Summary:PDF Full Text Request
In the21st century, With the rapid development of technology, compressed video isbecoming more and more. For manage and applicate video information effectively, weneed establish a video retrieval and management system. Key frame extraction technologyis the most important technology of video retrieval technology. The paper research thetechnology of extract key frames in the video compression domain.Firstly, The paper state the research background and significance of the task, thenintroduced the current development status of video retrieval technology at home andabroad. The thesis analysis MPEG video structure; research the algorithm of key frameextraction technology and the current question of key frame extraction technology bycombining algorithm.Secondly, based on read a large number of articles about keyframe extractiontechnology and study video compression theory and standards in-depth, the thesis researchtwo key frame extraction algorithm. The first algorithm utilize DC coefficient of I frameand the motion vector of P-frame as the feature vectors of the image group, to determinethe similarity of the image group and the clustering, set threshold value, using a clusteringalgorithm to extract key frames. The second algorithm partially decode theMPEG-compressed video format, utilize the DC coefficients of I frames to determinewhich one of between the two I-frames is key frame area. Subsequently using P-framemacro-block type and the motion vector of the P-frame and B frames to determine the keyframe area whether have the gradual change, mutations and big motion. Big movement isfast moving objects, may lead to a miscarriage of judgement. The algorithm can judgelarge motion and gradual change frames which has big brightness change.Thirdly, The papers research the MPEG decoding, then structures key frameextraction experimental system on VC++6.0platform based on MPEG compressiondomain, write the pragrames of the two algorithms to extract the key frame and display theresult of key frame extraction. The system can read and decode MPEG video, display thebasic information and the composition of video, display the information of a frame, extract key frame and show the results of the key frame extraction.Finally, The paper select different types of video clips to test the experiment. Theexperiment show that the experimental results in accordance with expectations.
Keywords/Search Tags:MPEG compressed domain, DC coefficient, macro-block type, motion vector, key frames
PDF Full Text Request
Related items