基于点线特征的帧间匹配流视觉里程计研究  

Research on Inter-Frame Matching Flow Visual Odometer Based on Dotted Line Feature

作　　者：钱潘优 甘屹[1] QIAN Pan-you;GAN Yi(School of Mechanical Engineering,University of Shanghai for Science and Technology,Shanghai 200082,China)

机构地区：[1]上海理工大学机械工程学院,上海200082

出　　处：《软件导刊》2020年第5期1-8,共8页Software Guide

基　　金：国家自然科学基金项目(51375314)。

摘　　要：针对机器人快速运动时视觉里程计精度严重下降问题,提出基于点线特征的帧间匹配流视觉里程计(Point and line of frame math,PL-FM)算法,以提高机器人在快速运动情形下的定位精度。PL-FM算法通过对图像的预处理去噪,在特征点提取时引入灰度值权重,从而降低快速运动时光照的影响。将特征点匹配问题转化为向量计算,从而减少匹配时间,在帧间匹配流则采用衰减关键帧计算位姿,从而提高关键帧利用率。通过4组实验对比,证明PL-FM算法误差精度提高70%,时间效率提高75%,保证了移动机器人的定位实时性,实现了低误匹配率及较高的定位精度。In order to solve the problem that the accuracy of visual odometer is seriously degraded during rapid motion of the robot,a point and line of frame math(PL-FM)algorithm based on dotted line feature is proposed to improve the positioning accuracy of the ro⁃bot in fast motion.The PL-FM algorithm denoises the image by preprocessing,and introduces the gray value weight in the feature point extraction to reduce the illumination effect during fast motion.The feature point matching problem is transformed into vector calculation to reduce the matching time and match between frames.The stream uses the attenuation keyframe to calculate the pose to improve the utilization of keyframes.Through four sets of experimental comparisons,it is proved that the PL-FM algorithm improves the error accu⁃racy by 70% and the time efficiency by 75%,which can ensure the real-time positioning of the mobile robot and achieve low mismatch rate and high positioning accuracy.

关 键 词：预处理去噪 灰度值权重 帧间匹配流 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]