LSD井下视频图像线特征匹配算法改进  

Improvement of line features matching algorithm for LSD downhole video image

作　　者：毛昕蓉[1] 杨兴林 张小红[1] 韩晓冰[1] MAO Xinrong;YANG Xinglin;ZHANG Xiaohong;HAN Xiaobing(College of Communication and Information Engineering,Xi’an University of Science and Technology,Xi’an 710054,China)

机构地区：[1]西安科技大学通信与信息工程学院,陕西西安710054

出　　处：《西安科技大学学报》2022年第6期1224-1231,共8页Journal of Xi’an University of Science and Technology

基　　金：国家自然科学基金项目(61901358);陕西省计划项目(2019SF-260)。

摘　　要：图像特征提取匹配做为视觉SLAM(Simultaneous Localization and Mapping)的重要组成部分,在井下无人巡检机器人上应用广泛。针对井下环境复杂,光照不足,现有特征提取匹配算法存在匹配率低,进而导致视觉SLAM定位精度低的问题。通过对现有LSD(Line Segment Detector)线特征匹配算法进行改进,采用对比度亮度和对数变换算法对采集的视频图像帧进行图像增强,利用Canny边缘提取算法对增强后的视频图像帧进行图像边缘信息提取后进行LSD线特征提取匹配,与原始算法进行平均匹配率对比分析。结果表明:在连续300帧井下视频图像匹配过程中,改进算法的平均匹配率为99.88%,原始算法的平均匹配率为88.42%,其平均匹配率提升11.46%。说明改进的LSD井下视频图像线特征提取匹配算法具有更高的匹配精度且更适用与井下无人巡检机器人进行无人巡检工作。As an important part of visual SLAM(Simultaneous Localization and Mapping),image feature extraction and matching was widely used in downhole unmanned inspection robots.In view of the complex downhole environment and insufficient illumination,the existing feature extraction and matching algorithm is low in matching rate,resulting in a low visual SLAM positioning accuracy.The existing LSD(Line Segment Detector)line feature matching algorithm was improved.And the contrast brightness and log transformation algorithm was used to enhance the collected video image frames,and the canny edge extraction algorithm was used to extract the enhanced video image frame edges after the LSD line feature extraction and matching,with the average matching rate by the original algorithm compared.The results show that the average matching rate of the improved algorithm was 99.88%,and the original algorithm was 88.42%,which was increased by 11.46%,which suggests that the improved LSD downhole video image line feature extraction and matching algorithm is higher in matching accuracy and is more suitable for unmanned inspection with the downhole unmanned inspection robot.

关 键 词：LSD算法 Canny边缘提取 线特征匹配 井下视频图像 匹配率 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]