Application Study Of Digital Image Processing In ITS

At present, intelligent transportation system(ITS) is a foreland research task which applied the electronics to the traffic field in the world. Image processing plays an important role in ITS. Applying image processing techniques to ITS is a challenging field, which has great theoretical significance and practical value. Image processing technology in the ITS utilizes method of image processing together with computer vision, obtains useful information which ITS needs. Due to running the red light, road accidents occur easily at road junctions. In order ot overcome the disadvantage of ILD(inductive Loop Detectot),the paper surveil the vehicles through tracking ,and recognize the license plate when the vehicle run the red light.The algothirom is also applied to detect other peccancy :yellow-line covering,converse running,over-speed running.Traffic surveillance must be working in difference weather, light, and condition. In day time,we detect the vehicle through three-frames difference,and track the vehicle through the Kalman filter.At last the paper analyse the vehicle occlusion .In night time,the paper give two solutions.In the first solution,the paper extracts the headlights.And we can track the vehicle through tracking the headlights. In second solution,we first give a improved Retinex algorithm,which not only can eliminate the influence of headlights and their beams, but also can enhance the image. After the preproceesing, we can detect the vehicle by two-frame difference,and track the vehicle. As shown in experiments, the two solutions can exactly track moving vehicles at night.We also research the question of license plate location. A improved direction region distance measure based approach to color edge detection is proposed.And suppressing the non-plate region by edge-color pair. At last, localizing the license plate by structural and texture features. As shown in experiments,the method can locate the plate under the difference background.Next the paper apply a matching function to segment the license plate character which can improve the character segmentation reliability and real time.Befor segmentation, enhanceing the image by improved Retinex,and correcting the tilting plate.