KAZE算法在巡线机器人障碍物检测中的应用  

Inspection Robot Obstacle Detection Based on KAZE Algorithm

作　　者：张蒙 ZHANG Meng(School of Electrical Information Engineering, HoHai University WenTian College, Maanshan, Anhui 243000, Chin)

机构地区：[1]河海大学文天学院电气信息工程系,安徽马鞍山243000

出　　处：《湖南城市学院学报（自然科学版）》2017年第3期61-64,共4页Journal of Hunan City University:Natural Science

摘　　要：巡线机器人障碍物检测技术是计算机视觉与自主巡检系统的研究热点之一,对保证输电系统安全可靠运行具有重要意义.传统的检测算法鲁棒性低,不能适应复杂环境下障碍物的检测要求.针对以上问题,提出一种基于KAZE算法的巡线机器人障碍物检测方法,该算法在非线性尺度空间中进行特征点提取,利用最近邻匹配准则和RANSC算法对由M-SURF算法生成的特征向量进行检测.实验结果表明,基于KAZE算法的巡线机器人障碍物检测效果较好,具有良好的的鲁棒性.Inspection robot obstacle detection technology is one of the hot spots of computer vision and autonomous inspection system. It is important for ensuring the safe and reliable operation of transmission system. Low robustness of traditional detection algorithm can not meet the requirements on obstacle detection when dealing with complex environment. To solve the above problem, an inspection robot obstacle detection method based on KAZE algorithm is given. The feature extraction is done in the nonlinear scale space, and feature vectors are formed with M-SURF algorithm. Feature vectors detecting is worked through the nearest neighbor algorithm as matching criteria and RANSC generated algorithm. Experimental results show that the effect of inspection robot obstacle detection algorithm based on KAZE can make a good performance with better robustness.

关 键 词：KAZE算法 巡线机器人 障碍物检测 鲁棒性 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]