A novel method for real-time object detection and multiple persons tracking
Real-time robust algorithms of object detection and multiple people tracking for outdoor visual surveillance scenes are researched. Firstly the background is constructed and updated based on a new adaptive Gaussion model considering pixels chroma components and luminance component in HSV colorspace. Then foreground pixels are detected based on the background model. After moving objects are extracted, the Camshift algorithm-called Auto-Camshift which is based on H-S 2D histogram is extended to track multiple pedestrians. Occlusion during tracking is handled by the location estimation method according to pedestrians moving velocity. Finally experiment results for outdoor surveillance scenes show that we are able to reliably detect and track multiple people in non-crowded surveillance scenes with real-time speed.
Background reconstruction Object extraction Noise Removal Auto-Camshift algorithm
Guishan Xiang Xuanyin Wang
School of Information and Electronic Engineering, Zhejiang University of Science and Technology, Han State Key Laboratory of Fluid Power Transmission and Control, Zhejiang University, Hangzhou, China
国际会议
杭州
英文
354-358
2009-04-08(万方平台首次上网日期,不代表论文的发表时间)