Motion Detection Of Sports Vehicles Based On Frame Difference Method And AdaBoost

With the development of human society, the need of sevices and management from intelligent transportation system have been increasing day by day, so the research on this subject has been attached much more importance to both domestic and international. The vehicle detection technology is the foundation of intelligent transportation systems. Based on analysis and summary of the traditonal target detection technology and system, a novel method which combines frame difference and AdaBoost method is present in this paper.The main work of it shows as follows:1.Different from the conventional frame difference method, the main purpose of this frame difference method is to obtain the information of gray level characteristic information form original frame image and show it simply, so the binarization operation is not taken, the frame difference images with rich gray level charactercistic information are put into the following processing steps.2.Remove the dotted noise from frame difference image with image filter, so the information of motion is protruded.3.Extract vehicle samples and non-vehicle samples from frame difference images manualy, then train weak classifiers of AdaBoost method with them, a final strong classifier will come after some iteration steps. Thanks to the proceding steps, the moving target has been highlighted, so the learning steps to recognize target are greatly simplified.After the experiment, this identification system for road vehicles achieved good recognition rate and acceptable recognition speed.