Research Of Robot Path Planning In Foggy Weather Based On Binocular Vision

Movement of mobile robots in foggy weather mainly depends on information of the surrounding environment captured by cameras and how the path planned according to the captured information.Binocular vision system achieves its function with the help of binocular camera.It can get the depth information of surrounding environment by capturing left side and right side image,which provides information support for movement of mobile robot in foggy weather.Based on binocular vision,this paper focuses on the enhancement of foggy image and binocular stereo matching algorithm.The main work is as follows:(1)In view of the disadvantages that the traditional foggy image enhancement method used a filter with fixed scale parameters without taking the fog concentration into consideration,this paper proposed an improved Retinex algorithm based on the dark channel prior theory.Based on dark channel prior theory,this algorithm used gaussian filter of different scales to different fog concentration areas according to the relationship between filter scale parameter and the mean value of sub-block dark channel.Then this algorithm enhanced image with the help of Retinex theory,and finally further optimized information of details by restoring color information and enhancing image edges with improved fractional differentiation.Experimental results showed that this algorithm could improve sharpness and contrast of image,make image details richer and had stronger edge retention ability.What's more,the restored images had better layer effect and overall effect.(2)In view of the disadvantages that the traditional stereo matching algorithm based on Census transformation had in noise suppression and image details,this paper proposed an improved Census stereo matching algorithm based on window mutual correlation information.Firstly,this algorithm got the mean value of each pixel gray value in the window,made it as the reference value.At the meantime,added a noise threshold on the basis of reference value as another constraint to improve robustness of this algorithm.Then calculated weighted matching cost by means of SAD,which can reduce noise interference,enrich image details and reduce matching error rate.Experimental results showed that compared with other traditional algorithm,thisalgorithm had lower matching error rate,higher matching precision and obvious improvement effect in the area of depth-discontinuity.This paper designed a mobile robot with Arduino controller and binocular camera that can work in foggy weather.Based on the image enhancement and binocular stereo matching algorithm proposed in this paper,the task of binocular distance measurement and path planning and obstacle avoidance were accomplished.Experiment results proved that methods proposed in this paper could guarantee system stability and consistency.