Measurement Of Safe Driving Distance Based On Stereo Vision

With the development of economic in our country and the improvement of people’s daily life in the twenty-first century, The average family car has become an essential means of transport, when we are stepping into an "auto age," incidence of traffic accidents are rising, caused huge losses. In order to improve driver safety on the road, and help drivers judging and analyzing exactly in driving, stereo vision safe driving distance measurement method in this paper is based on theme of automotive anti-collision safety system.This is parallel to the two cameras placed on the vehicle, the cameras left and right to can take photos of the common field, this two camera used to detect the distance from obstacles in front. First established the camera imaging model according to the principle of stereo vision; then create the connection between spatial location and image through camera calibration; Second, feature extraction and stereo matching are deal between image captured from the two cameras, this step is very important because it directly affects the accuracy of distance; Finally, according to principles of the measured distance from the target about the vertical distance between the camera.In camera calibration, the projection matrix are deduced by the inner and the external parameters of camera, Through the projection camera calibration matrix, in the analysis comparing the various pros and cons of camera calibration technique, selected camera calibration target method based on the2D plane as this calibration method, to be compared with experiments of calibration kit, the calibration method of his paper is the basement and premise of binocular ranging system, and it also provides a degree of accuracy and quickness.The main purpose of camera calibration is the front of the camera target to target and shoot camera to establish the spatial relationship, the relationship between spatial location and then mapped to the cameras image, the actual spatial relationship with the video camera to image to establish the corresponding contact. Imaging camera to determine the geometric model of the nonlinear distortion parameters obtained, based on the minimum projection error criterion to optimize the parameters of internal and external, making the camera showing undistorted image.Then the cameras on the left and right image are tackled with feature extraction and stereo matching, the paper commonly use stereo matching algorithm analysis, the first candidate based on the average gray-scale window function obtained matching method, matching narrow the other unnecessary computation time, locked areas of interest match pairs. Then use Twice matching computation to two sides of window function image which we are interested in. Improve the match effectiveness without adding match runtime. because the target image based on different site, we can work out the optical parallax between two images when get Matching feature points between two sides images. Finally, work out the distance between obstale to the left and right camera using triangulation theory.In order to get the matched feature points in the two sides of image, because they are located around in different coordinates of images, the difference are calculated between two points, that is parallax. Finally, we use the triangulation principle to get the camera’s distance to the obstacle. The experiments results demonstrated the proposed stereo vision-based safe driving distance measurement method is feasible in theory and can support backup safety functions in transportation system accessibility.When driving on the highway, it will be susceptible to weather changes, which will affect the Image acquisition, resulting in increased driving distance measurement errors. So image pre-processing is needed at first, it used for capturing a clearer image and improving the picture on the computer in order to get more accuracy distance between the locations of obstacles in front. Through many times of experiments, this pre-processing can get a higher precision measurement results, also can improve the accuracy of ranging and reduce the range of error.