Improving Vehicle Detection and Analyzing Motion Features through Advanced Techniques

There is an increase in traffic flow due to development of vehicle technologies and transport systems. Detecting the movement of the vehicle in video sequence is difficult because of illumination condition, background images, and occlusion, unexpected object motion, change in form of the object pattern, non-rigid structures of object. This research work focuses on applying image processing concepts for detecting and tracking vehicles. This system can detect, recognize, track multiple vehicles and provide the information on the object vehicles. Video database was created by shooting video in hand DIGICAM. The videos were taken in the day time, cloudy time, late evening and at night. All the information were processed using MATLAB R2011a. Image processing technology is employed in the object detection. The potential objects can be detected, and at the same time, related information such as predicted direction, change in size, change in velocity, ellipse related parameters (major axis, minor axis and angle), static to static, static to moving, moving to moving, moving to static, sudden stop, sudden speed, occlusion with two or more separate objects, split of one object to more than one objects are recognized. Once the potential objects are detected, related information can be calculated and used to judge if the object vehicle is hazardous to the host vehicle and it is useful in parking lots and in the traffic signal.