100 likes | 340 Views
Shot boundary detection based on frame histograms analysis. Tomsk Polytechnic University. Vakulenko M.D. 1 , Kovalenko D.A . 2 , Tolkunov S.V. 2 , Master Students Gr. 1 -8BM10, 2 -8VM13. Scientific advisor: Dr. Sergey V. Axyonov. Motivation
E N D
Shot boundary detection based on frame histograms analysis Tomsk Polytechnic University VakulenkoM.D.1, Kovalenko D.A. 2, TolkunovS.V.2, Master Students Gr.1 -8BM10, 2 -8VM13 Scientific advisor: Dr. Sergey V. Axyonov Shot boundary detection based on frame histograms analysis
Motivation • Presence of huge unstructured and unindexed video archives in the world • Video scenes search problem • Need for automated video content analysis • Target • Automated scenes splitting by the shot boundary detection based on histograms difference analysis • Tasks • Create and implement an algorithm that is capable of detection base cuts in the movies based on hard threshold. • Create and implement an algorithm that is capable of detection base cuts in the movies based on adaptive threshold. • Estimate the effectiveness of the best developed algorithm for some different movies (quality, sizes). Motivation, Targets and Tasks Shot boundary detection based on frame histograms analysis
Application Shot boundary detection based on frame histograms analysis
Different Cuts Shot boundary detection based on frame histograms analysis
Overview of the Model Shot boundary detection based on frame histograms analysis
Elements of analyzed vector: Mathematical Model According to obtained vector C adaptive threshold can be calculated in following form: - correlation function - vectors obtained from histograms of red, green and blue component of the i-th frame , , - weighting factor - threshold value n – number of neighbor of analyzed element - mathematicalexpectation - variance Video segmentation
cvCreateFileCapture • Get frames with some step from video file • Calculate histogram difference between neighboring frames • Store it in the array • Calculate adaptive threshold • Peaks which we received are places of cuts While(!EOF) { Algorithm cvQueryFrame cvCalcHist cvCompareHist } cvAdaptiveThreshold Shot boundary detection based on frame histograms analysis
Examples Shot boundary detection based on frame histograms analysis
Developed needed mathematical model for movie shot boundary detection. Created a C++ program that implements the model based on the OpenCV library. The software was tested to detect shot boundary for the different media types (different movie quality, different cut types). The system can detect about 90% described cuts in the movies. Results Shot boundary detection based on frame histograms analysis