Three-dimensional Reconstruction And Road Region Detection Method Based On Image Segmentation

Among the relevant researches on mobile robot, navigation technology is one of its core technologies. Detecting obstacles based on the surrounding information is an important element of navigation technology research on mobile robot. A large proportion of moving environments of mobile robot are complex environments, therefore, analyzing obstacles in complex environments has great significance.This paper studies accessible region detection based on visual mobile robot. One kind of obstacle detecting method has been tried, which combines color image segmentation based on fuzzy c-means clustering and stereo vision. The main researches include the following two parts: color image segmentation based on fuzzy c-means clustering and stereo vision.Polysegment is introduced to carry out precise estimation to the clustering number of image, which resolves the problem of parameter initialization of fuzzy c-means clustering. On this basis, through adopting histogram technique and anisotropic weight information, make full use of the color information of the image to improve traditional fuzzy c-means clustering, which surmounts the deficiency of sensitivity to noise.In the part of stereo vision obstacles detection, the vertical scan line in image segment region is used as the research object. Make three-dimensional reconstruction to space point to obtain the three-dimensional information of point in the image, and then determine whether it is obstacle region according its height value. Since the used stereo visual system is a parallel placed dual-camera system by controlling the rotation of camera and cloud platform, the pixel point matching speed of image pair is increased and the three-dimensional reconstruction method of space point is improved.At last, the result shows that this method can adapt to complex environment and reliably detect the accessible region along advancing direction of the mobile robot.