Research And Implementation Of Vehicle Side Collision Warninging

With the development of economy,people become more and more wealth and keep pursuing high standard lifestyle.More and more people and family buy their own cars as a kind of transportation.Therefore,the quantity of cars increasing dramatically.However,this tendency also leads to bad traffic jam.Moreover,with the development of technology,various kinds of systems and technologies have been invented to improve vehicle safety.During the driving process,more than 80%of the environmental information is catched by driver' s eyes.Comparing with radar,ultrasonic,infrared and other detectors,the visual sensor is more similar with the driver's eyes.The concept of vision-based intelligent navigation system has been proposed in the 1960s.But due to the limitations of technological development,backward computer hardware and limited computing power,complex operations could not be performed and computer vision-related application is limited at that time.Entering the 21st century,science and technology are developing faster and faster,and the computational complexity of computers is greatly improved.Under this background,the image processing related algorithms have been widely studied and proposed.Meanwhile,the detection of moving objects and moving objects applications,such as tracking have also been developed,which provide a theoretical basis.Therefore,the assisted driving system based on computer vision and image processing has very meaningful and bright development prospects.The goal of this paper is to expand the lateral view of the vehicle through video stitching by combining multiple image processing techniques,while detecting and tracking the laterally traveling vehicles,and predicting whether the lateral vehicles will collided with the car by obtaining their distance and speed direction.In this paper,the SURF algorithm is modified to improve the efficiency of the algorithm.Meanwhile,combining with the actual situation of the experiment,the calculation area is reduced,the calculation amount of the system is reduced.Finally,the image projection process is simplified by the table look-up method,in order to achieve the real-time requirement.At the same time,this paper also proposes a vehicle tracking algorithm by combining Camshift algorithm and Kalman filtering,which improves the success rate of vehicle tracking and provides a theoretical basis for vehicle lateral collision avoidance warning.Finally,the above process is combined,and the knowledge of certain distance measurement and speed measurement is used to realize the design and implementation of the early warning prototype system of vehicle lateral collision prevention.