Background extraction model and shadow detection algorithm in traffic flow video detection
Zhi-hui Li,Chang-hai Zhang,Zhao-wei Qu,Dian-hai Wang
DOI: https://doi.org/10.3969/j.issn.1671-5497.2006.06.033
2006-01-01
Abstract:To alleviate the difficulties in the detection and recognition of the moving objects, even the possibility of the object misclassification, due to the effect of the variation of the moving object shadow and the background factors, a mixed Gaussian background update model based on the object level and a moving object shadow detection algorithm based on the RGB color variation degree were proposed. It performs the mixed Gaussian background update according to the object high-level semantic expressions, such as movement segmentation, object recognition, Kalman movement tracking, etc., in the light of spatio temporal features of the pixels, eliminates the effect of the prolonged traffic standstill due to the traffic control signs and the rushtime traffic congestion on the background extraction in the mixed Gaussian model based on the pixel level, avoids the object misclassification due to effect of the moving shadows. The results of experiment on the video pictures of different traffic conditions showed the proposed technique is robust and self-adaptive.