Freeway Passive Ranging Algorithm

The ranging system of highway is to avoid the traffic accident by detecting and getting the right position of the obstacles. Visual ranging based on computer vision is very promising in this field.In this paper we chose the passive ranging which is based on bio-stereo vision from many methods of visual ranging,according to the feature of the highway. The main obstacle on highway - vehicle and its position can be detected by image segmentation pattern recognition and some pre-experience. The most difficulty of this paper is image segmentation which is also our main work.There are many methods of image segmentation,such as the segmentation based on prey image,the segmentation based on color image,the segmentation based on information theory,etc. But none of them can produce the desired result,in the complicated background of highway. After analyzing a lot of images of highway,a new algorithm which is a color image segmentation based on wiping off the area out of the road has been presented. Firstly,the road area can be detected in homogeneity image and the area out of the road will be wiped off. Then,the color image segmentation will be processing in the road area. After dealing with the segmentation result by some pre-experience,the object can be picked up completely. The relativity of the objects in the two road images can be got through image matching. Lastly,we can use the relativity to complete the ranging. In order to reducing the difficulty of processing,image pre-processing is necessary.The method in this paper will meet the requirement of ranging on highway well,which can be concluded from the result.