Study On Warning Of Safe Distance At The Traffic Lights Intersection Based On Binocular Vision

With the rapid development of economy, automobile becomes the main traffic tool now, which brings convenience and comfortable to people, at the same time, it also has a great threat of people’s lives and property safety, especially at the traffic lights intersection. In order to improve driving safety of vehicle and reduce the number of traffic accidents at the traffic lights intersection and assist the driver to make accurate judgements, this paper designs and implements the warning study of safe distance at the traffic lights intersection based on binocular vision.This paper adopts the technologies that have binocular stereo vision system, image processing and fusion principle of BP neural network. Firstly, two cameras are used to capture the traffic lights’information that are two pieces of binocular images at the left and right from different perspectives, which are processed. We gets gray images with the algorithm of gray processing, and then two pieces of binocular images are processed through Smooth, corrosion and open and close operation. We position the interested area, and get the color of the traffic light, countdown seconds of green light and two matching points of the corresponding features by using the method of stereo matching. Following this, we obtain the internal and external parameters of cameras by camera calibration technique. Then, according to principle of triangulation and parallax theory, the distance from the camera to preceding vehicle, the distance from the camera to traffic lights are obtained and Using imaging differences of the same target in projection of different images, the three-dimensional information of measured target and depth information are obtained. It is that the distance from the camera to preceding vehicle and the distance from the camera to traffic lights are obtained. Then, the BP neural network technology is applied to warning of safe distance at the traffic lights intersection. By Training for network structure, we can establish the mapping relationship between the characteristics of road and the actual road type. The speed module acquires vehicle’s speed again. Through the effective fusion of the above information available, it has enough safe distance that the vehicle passes the he traffic lights intersection. Finally, the buzzer is used warning of risk based on the hardware platform of ARM chips.After a great deal of experiments, the experimental result shows that This paper’s design of traffic light intersection safety distance warning based on binocular vision system to research the car traffic light intersection safety distance warning function, simple and effective, and the characteristics of strong robustness, good real-time. It has the high precision of measurement and laid a foundation for the further research of automobile safety system.