A traffic parameter detection based on computer vision

Aiming at the characteristics of the road video traffic flow, mainly study the light change on the impact of the moving vehicle recognition system, and analysis the changes of the spatial information, which raised a moving object extraction algorithm based on background adaptive updating. The algorithm model fully considered the graded influence which produced by light field to video image sequences, realized self-adapted extraction to background, and overcomed the problem which was difficult to extract vehicle target by using fixed background. With the algorithm of projection principle and two-dimensional combined statistics, the problem of traffic parameter extraction in complex background was solved. Experimental results show that the algorithm is effective for traffic parameters extraction, and have very strong robustness in the conditions of the big space, complex background and different illumination. These algorithms can truely reflect the prevailing traffic condition, which also provide a new detection method for intelligent transportation systems.