A Research On All-weather Vehicle Detection Method Based On Video

The rapid development of China's economy and the increasing number of vehicles bring a series of serious traffic problems.Intelligent transportation system is an effective way to improve traffic problems.Vehicle detection is an important research topic in intelligent transportation systems.However,vehicle detection is affected by light,weather,and various complex environmental conditions,there are many difficulties,so the study of vehicle detection is of great significance.Aiming at the problem of poor daytime and nighttime vehicle detection under severe lighting conditions,an all-weather on-road vehicle detection method is proposed.The detection of all-weather vehicles mainly includes: daytime detection of vehicles,night detection of vehicles,and adaptive switching of day and night detection modes.The main research contents and results are summarized below:(1)For vehicle detection at daytime,training different layers of classifiers based on the same samples by adaboost algorithm,choosing optimal classifier by contrast.Processing detection area and making irregular detection area to rectangles.Then detecting vehicle by multi-scale detection in detection area.At last,recovering the results of detection to the original video.This method is effective to avoid the missing detection caused by irregular detection area and increase the detection rate.(2)For vehicle detection at nighttime,because the environment of road at nighttime is more complex,considering of driving side by side,stopping side by side and changing lanes,the method combine the image processing and virtual loop technology is proposed.Detecting vehicle by detecting vehicle headlights,and this method is useful to avoid the false matching between different vehicles.(3)Illumination intensity will be effected by the weather changes and alternation of day and night,for switching of vehicle detection at daytime and nighttime,an adaptive switching method is proposed,the detection mode will be able to switch adaptive under different illumination intensity.This method increased the flexibility of vehicle detection and increased the detection rate during the illumination intensity change.The all-weather vehicle detection method can automatically switch different detection modes according to different detection environments to carry out vehicle detection and achieve good detection results.