A Moving Object Detection Algorithm of Jitter Video

Abstract

In view of the problem that camera shake cannot accurately detect moving target in natural environment, this paper proposes a moving target detection algorithm combining grey projection, background difference and continuous frame difference. The algorithm calculates the gray projection after removing the peripheral edge pixels from the image frame, and calculates the interrelation between the images according to the obtained gray projection curve, and completes the jitter sequence correction. The fusion strategy of background difference and continuous three frame difference method is proposed to enhance the motion target area. The fusion differential image was segmented using Ostu method to detect the foreground moving target. After the public dither video sequence experiment and comparison and verification with different algorithms, it can be concluded that this algorithm can accurately detect the moving target in camera dither scene, and ensure a good detection effect while the detection speed is fast and the calculation amount is small.