Using Three-Frame Difference Algorithm to Detect Moving Objects

Abstract

Detection technology for moving targets is an important direction in computer vision research. This paper proposes a method to improve moving target detection, which is a three-frame difference algorithm combining edge information. This difference algorithm can detect moving objects with high accuracy. For the real-time image captured by the camera, the algorithm not only can reduce the processing time, but also judge the moving object with the fastest speed and the highest accuracy. In the process of the algorithm, Expansion and corrosion in mathematical morphology are used to remove noise from images. Due to the detection of moving targets, there will inevitably be some interference factors, and these factors will have a more serious impact on the detection sample. In order to eliminate the influence of noise on the processing results, the algorithm first divides the image into binary image, expansion and corrosion treatment of the image, and then according to the different pixel points between the three frame images, the shape of the object can be determined by the information of space and time, so as to identify the moving object, give an alarm and save the image to the local area.