Based on the split the three-dimensional image reconstruction and road region detection method

Mobile robot technology research , navigation technology is one of its core technology . According to the information of the surrounding environment , the detected road region is an important content of mobile robot navigation technology research . Large proportion in the the mobile robot driving environment , the complex unknown environment , so the analysis has an important significance of the road in these environments . This thesis is based on the vision of the mobile robot road detection , try a combination of a color image segmentation based on fuzzy C-means clustering and three-dimensional visual road area detection method . The main research topics include: color image segmentation based on fuzzy C-means clustering , stereo vision obstacle detection of two parts . Introduction of polynomial image segmentation techniques for the accurate estimation of the number of clusters , and solve the problem of fuzzy C-means clustering parameter initialization . Weight information on this basis , by using anisotropic rights , and take full advantage of the color of the color image information to improve the traditional fuzzy C -means clustering to overcome the shortcomings of the algorithm for noise sensitive . Vertical scan lines of stereo vision obstacle detection part of the use of image segmentation area as a research object , through improved three-dimensional reconstruction method of space point to obtain three-dimensional information of the image in pixels , and then determine whether the obstacle region based on the height value . Stereo vision system is similar to the system of dual cameras placed in parallel , to improve the image on pixel matching speed has been improved , and three-dimensional reconstruction method of space point . Finally, experimental results show that the method can adapt to the different changes in the environment , and more reliable to detect the region of the road in front of the mobile robot travels .