关键词: 多目标跟踪, 离散小波变换, 背景差分法, Camshift算法

Abstract: In order to improve object recognition accuracy of video surveillance in the coal mine and to track moving targets effectively, wavelet transform is combined with background subtraction to improve Camshift algorithm, which applies to video surveillance in the coal mine. Firstly, three decomposition principles of wavelet of wavelet transform are adopted for image de-noising to get low-frequency images. Then, background subtraction is operated to detect moving targets. Finally, the improved Camshift algorithm is applied to track moving object in the video. Simulation results show that the improved Camshift algorithm reduces the randomness when initializing the candidate targets in original Camshift and improves the accuracy of object recognition and tracking, which ensures safe production of coal mine.

Key words: multi-target tracking, discrete wavelet transform, background subtraction, Camshift algorithm