Compression and Security of Surveillance Videos Exercise 6 – Shot Change Detection M 陳威佑
Background My method Experiment results Conclusion Outline
Background Detecting shot change by my own method
Calculate the average histogram difference between current frame and previous frame, and give a threshold to determine if the current is a shot (threshold=200) Calculate the PSNR between current frame and previous frame, and give a threshold to determine if the current is a shot (threshold=17 dB) If satisfy these two condition, mark the current frame as a shot My Method
Experiment Results Detected shot number: 40, 67,83,129, 169, 177, 195, 336, 362, 395, 435, 767, 1130, 1137, 1145, 1179, 1225, 1268, 1313 Correct: 14 Missed: 11 False: 5 Recall rate: 56% Precision rate: 73.68%
Use histogram and PSNR to detect shot change is very sensitive to suddenly luminance change Conclusion Frame 66 Frame 67 Frame 1136 Frame 1137
The End