A Moving Object Extraction Algorithm Based on Hybrid Background Subtraction and Pixel Mean Technique

摘要

In this paper, a moving object extraction algorithm based on hybrid background subtraction and pixel mean is proposed to solve the problem that the existing object extraction algorithm can not adapt to the moving environment. The algorithm operation process includes target location identification, domain pixel mean value processing, image difference operation, background noise processing. Firstly, the processing frame is intercepted from the motion video and the thermal technology is used to recognize position of the moving target and get the preprocessing region. Then morphological recognition technique is used to recognize the target and establish the orthogonal rectangle enclosing the moving target. By means of domain pixel averaging, the background of the area in the orthogonal rectangle is modeled to obtain the background frame. Finally, the differential operation of processing frame and background frame is carried out, and then the target extraction results are obtained after the gray processing and binary processing. The experimental results show that this algorithm can effectively extract moving targets when it works in the motion environment, successfully merge the background subtraction method and the pixel mean value technique, and update the performance of the background subtraction method in the motion environment.