Comparative Research of Dynamic Target Detection Algorithms Based on Static Background

Dynamic target detection is a process of capturing moving objects in videos. Detection results are affected by various factors that limit the algorithms' further development. For example, mutual occlusion between objects can disrupt the overall structure of the target, which can easily lead to missed and false detections. In addition, when the light intensity increases or decreases, the pixel values of the input image also change, causing the results inaccurate. Considering the above factors, we need to understand how external conditions affect the accuracy and real-time performance of the algorithms to help us select the appropriate method in different scenes. In this paper, we propose a method for selecting suitable algorithms for different static scenes. It evaluates the number of dynamic targets and the light variation in the detected static background, which can help us select a suitable detection algorithm. First, we use the inter-frame difference, background difference, and the optical flow method to detect a single human target with a simple motion state in a closing scene. Then we apply these three algorithms to detect multiple targets in an open scene. In this situation, the algorithms need to deal with more complex information, bring a significant challenge for real-time and accuracy. During the process, the running frame rate is adjusted by modifying parameters in the experiments, improving the algorithms' real-time performance and accuracy. Finally, we compare the performance data of these algorithms in different scenarios and give an evaluation and analysis for each algorithm. In brief, we provide a basis and reference for selecting detection algorithms for different static scenes, which is of great significance.