摘要
针对在图像匹配要求实时性较高的情况下匹配率低的问题,提出了一种基于Harris自相关矩阵的迹和改进灰度值特征的高速匹配算法.该算法首先根据像素点位于角点时Harris自相关矩阵的迹最大的特点,用Harris自相关矩阵的迹检测待匹配图像的子区域角点,避免计算Harris响应函数,提高了图像匹配的实时性;然后在子区域的角点基础上利用灰度值特征检测出最高阶特征点,在参考图像中检测出灰度值与子区域最高阶特征点灰度值相同的像素点,并计算这个像素点和最高阶特征点的灰度值与其邻域灰度值之和的比值,当比值相同时则这个像素点被定义为参考图像中的特征点;最后以子区域特征点为匹配依据将参考图像和待匹配图像进行匹配.该算法与SIFT算法、SURF算法的仿真结果表明,在实时性要求较高的情况下该算法可得到高匹配率.
Aimed at the problem that the match rate is low when high real-time performance is required for image matching,a high-speed image matching algorithm is proposed on the basis of the trace of Harris autocorrelation matrix and improved gray scale value feature.In this algorithm,according to the feature that the trace of the Harris autocorrelation matrix is the largest trait when the pixel spot is located at the corner spot,the subdomain corner spots of the image to be matched are detected first by using the trace of the Harris autocorrelation matrix without the computation of Harris response function,so that the real-time performance of image matching is improved.Then,based on the corner spots,the highest-order feature spots are detected by using the gray scale feature,the pixel spots with identical gray scale value to that of the highest-order feature spot in the template image are detected,and the ratios of gray scale values of these pixel spots and the highest order feature spot to the sum of neighborhood gray scale value are calculated.When the ratio is the same,that pixel spot is defined as the feature spot in the template image.Finally,the subdomain feature spot is taken as matching basis to match the reference image with the image to be matched.It is shown by the simulation with the algorithm proposed in this article,SIFT algorithm,and SURF algorithm that a high matching rate can be obtained with the algorithm proposed when the real-time requirement is higher.
作者
赵小强
张源峰
ZHAO Xiao-qiang;ZHANG Yuan-feng(College of Electrical and Information Engineering,Lanzhou Univ.of Tech.,Lanzhou 730050,China;Key Laboratory of Gansu Advanced Control for Industrial Processes,Lanzhou Univ.of Tech.,Lanzhou 730050,China;National Experimental Teaching Center of Electrical and Control Engineering,Lanzhou Univ.of Tech.,Lanzhou 730050,China)
出处
《兰州理工大学学报》
CAS
北大核心
2018年第5期108-113,共6页
Journal of Lanzhou University of Technology
基金
国家自然科学基金(61763029
61873116)
甘肃省基础研究创新群体基金(1506RJIA031)
