Vehicle And Forward Distance Detection Based On Computer Vision

With the development of the global economy and the surge in car ownership, traffic accidents rate is rising rapidly. Vehicle safety, as an important research topic of public concern, attracts the universal attention of the society and the government. The development of intelligent transportation system and intelligent vehicle is an effective measure to solve the trafic safety problem of our country. Vehicle distance measurement technique is the research fronts and hotspots of intelligent transportation system and intelligent vehicle system and it is also one of the effective techniques to solve the frequent traffic accidents.This paper puts forward a method of vehicle and forward vehicle distance detection based on computer vision. It uses a singer CCD camera as the input device, processes the input images with computer, detects the shaded area of the image as the assumed vehicle region, then train a vehicle classifier based on Adaboost algorithm to determine and verify the assumed vehicle region, through that we can detect the location of vehicles in the image. Meanwhile, this paper adopts a vehicle distance model based on the geometrical reasoning method, combining with the camera calibration, we can establish the corresponding relationship between the image coordinates and the actual road surface coordinates, then we can calculate the actual distance between the two vehicles.In this paper, we developed a video application program of vechicle and vechicle distance detection based on computer vision, using the open source computer vision library (OpenCV) under VC++6.0. We use OpenCV as the the base library to develop the video application program, rewriting or calling the function of the library, we can use C or C++ language to develop our own video applications, as a result, solve the long cycle and low efficiency defects of the development of video applications.In order to test the system, we capture the videos of real road and process the video with the system; meanwhile, we make real-time vehicle and vehicle distance detection. The experimental results show that the method of vehicle and vehicle distance detection is feasible and effective under a certain extent.