一种多重影像线特征匹配算法的设计与仿真  

Design and Simulation of a Multi Image Line Feature Matching Algorithm

作　　者：崔戈辉 房金谱[2] CUI Ge-hui;FANG Jin-pu(Xi'an Technological University,Computer Science and Engineering College,Xi'an Shaanxi 710021,China;Xi'an Technological University,Xi'an Shaanxi 710021,China)

机构地区：[1]西安工业大学计算机科学与工程学院,陕西西安710021 [2]西安工业大学,陕西西安710021

出　　处：《计算机仿真》2022年第7期287-291,共5页Computer Simulation

摘　　要：由于当前已有算法未能对多重影像进行边缘检测,导致漏匹配率增加,匹配效率下降,匹配结果准确性较低。提出一种基于仿射不变性的多重影像线特征匹配算法,去除多重影像中的孤立像素,然后进行闭运算操作。根据得到的影像结合Sobel算子直接对原始影像进行边缘检测。将改进梯度相位计算方法和最小二乘方法进行拟合,并结合仿射不变特征提取多重影像线特征。采用SIFT算子对特征区域进行描述,实现多重影像线特征匹配。经实验测试证明,所提方法具有较高的匹配效率和正确匹配率,同时还能够有效减少漏匹配率,能够获取满意的匹配结果。At present,some traditional algorithms fail to perform edge detection on multiple images,leading to an increase in the missing matching rate and a decrease in the matching efficiency.Therefore,a line feature matching algorithm for multiple images based on affine invariability was put forward.Firstly,some isolated pixels were removed from multiple images,and then a closed operation was performed.After that,these images were combined with Sobel operators to directly detect the edge of the original image.Moreover,the improved gradient phase algorithm was integrated with the least square method.In the meanwhile,the affine invariant feature was introduced to extract the line features from multiple images.Finally,SIFT operator was used to describe the feature area.Thus,the line feature matching for multiple images was achieved.Experimental tests prove that the proposed method has higher matching efficiency and a correct matching rate.Meanwhile,this method can effectively reduce the leaky matching rate and get satisfactory matching results.

关 键 词：仿射不变性 多重影像线特征 匹配 边缘检测 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]