基于改进的SURF特征点的双目测距  

Binocular distance measurement based on improved SURF feature points

作　　者：朱照飞 刘伟[1] Zhu Zhaofei;Liu Wei(School of Materials and Energy,Guangdong University of Technology,Guangzhou 510006,China)

机构地区：[1]广东工业大学材料与能源学院

出　　处：《电子测量技术》2018年第12期133-138,共6页Electronic Measurement Technology

摘　　要：针对传统双目系统在复杂条件下的图像特征匹配算法精度低、计算量大、实时性差的问题,提出了一种基于改进的加速鲁棒特征（SURF）的特征匹配算法,并应用于双目视觉测量系统。首先对采集的左右视图进行极线约束,其次从左视图选择目标区域（ROI）,进行SURF特征向量检测,然后采用FLANN算法加速匹配,基于最大距离的阈值T初步筛选匹配点对,而后采用RANSAC算法去除误匹配。最后利用三角测量法进行距离计算,验证算法的可行性。实验表明改进的SURF算法获取的距离达到较高精度,实时性也较强。Aiming at the problem of low accuracy,large amount of calculation and poor real-time performance of traditional binocular system under complex conditions,in this paper a feature matching algorithm based on improved SURF（speed up robust feather）is proposed and applied to binocular visual measurement system.First of all,the left and right views of the acquisition are subjected to the polar constraint.Secondly,the target area（ROI）is selected from the left view to detect the SURF feature vector.Then the FLANN algorithm is applied to speed up the matching and the matching point pairs are initially screened based on threshold T of the maximum distance,then mismatches is removed by RANSAC algorithm.Finally,the triangulation method is used to calculate the distance to verify the feasibility of the algorithm.Experiments show that the improved SURF algorithm achieves higher accuracy and more powerful real-time performance.

关 键 词：加速鲁棒特征 双目测量 极线约束 图像匹配 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]