Video Jitter Detection Algorithm Based on Principal Component Analysis
Songhua Xie, Hui Nie · 2015
A simple and efficient detection algorithm is proposed for the problem of image jitter detection in video quality diagnosis system. In the algorithm the gray projection feature is selected to calculate the correlation of motion vector, which effectively solve the threshold selection problem. PCA is used to calculate the main direction of displacement vector. And the projection relationship of motion vectors in the main direction of displacement vector is used to judge the jitter. This way effectively utilizes the statistical characteristics, enhances the ability of anti-noise and reduces the rate of false positives. Experimental results show that the proposed algorithm can effectively detect the jitter of video images, and can meet the needs of the actual work.