Recognition of Navigation Path of Exhibition Robot Based On Visual Navigation

摘要：

This paper proposed a method of robot path recognition based on visual navigation. Beforehand, we lay the navigation path which robot walks along on the floor, and use a camera which on the robot to collect navigation path image in real-time. First of all, preprocessing the collected image and using OSTU method to calculate the best threshold value for image binarization, and then de-nosing with median filter and mathematical morphology. Second, we use the Sobel operator for image segmentation to segment the edge of the navigation path. At last, because the edge of the navigation path is not a pixel wide and a pixel is consists of three bytes, so the midpoint of each line in the edge image we find out may be more than one. We separate the midpoints into two groups, and then calculate the center of gravity of each group. The connection of the two points is the navigation line. The results of experiments show that the method in this paper has good feasibility to recognize the navigation path.

关键词： visual navigation mathematical morphology image segmentation path recognition

作者: Huiming Lv Jiwen Dong Xiuli Li Lei Zhao

作者单位: School of Information Science and Technology University of Jinan Jinan, Shandong Province, China

会议类型: 国际会议

会议名称: 2010 International Conference on Circuit and Signal Processing(2010年电路与信号处理国际会议 ICCSP 2010)

会议地点: 上海

会议语种:英文

页码: 182-185

在线出版日期: 2010-12-25（万方平台首次上网日期，不代表论文的发表时间）