Detection Algorithm of Vehicles Based on Image Statistical Characteristics

Aiming at the complexity of image recognition and the fluidity of the image itself, a new method to detect and track the target vehicles with the statistical features of the image gray was proposed. The one-to-one correspondence relationship between the target vehicle and the actual vehicle was built by the lane line symmetry and car installation angle. It will take the full image average grey value, target domain area, the perimeter between target domain and background domain border as the statistical characteristics. In addition, several order center distance of the target field distribution function was taken as one of statistical characteristics. The similarity between the extracted image characteristic parameter and the image characteristic parameter from the database at different distance is taken to identify the target vehicles accurately. The results shows that the algorithm cans effectively detecting and tracking the target vehicle in front of the host vehicle 90m apart.