基于相位相关和Harris纹理分类的SIFT图像拼接算法  

SIFT Image Stitching Algorithm Based on Phase Correlation and Harris Texture Classification

作　　者：陶磊 郝晓丽 王松涛 张磊 唐泽恬 TAO Lei;HAO Xiaoli;WANG Songtao;ZHANG Lei;TANG Zetian(School of Physics and Electrical Engineering,Liupanshui Normal University,Liupanshui,Guizhou 553004,China)

机构地区：[1]六盘水师范学院物理与电气工程学院,贵州六盘水553004

出　　处：《计算技术与自动化》2025年第1期12-18,共7页Computing Technology and Automation

基　　金：贵州省教育厅自然科学研究项目(黔教合KY字〔2022〕046号);贵州省高等学校教学内容和课程体系改革项目(2023302);六盘水师范学院一流本科专业建设点项目(LPSSYylzy2202);六盘水师范学院大学生科研训练计划项目(2022DK204)。

摘　　要：针对传统尺度不变特征变换(SIFT)算法计算量大的问题,提出了一种基于相位相关和Harris纹理分类的SIFT图像拼接算法。首先使用相位相关法获取待拼接图像的重叠区域,其次使用Harris角点响应函数计算待拼接图像的纹理复杂度,并且通过设定阈值将图像分为强纹理区和弱纹理区,最后在强纹理区域使用SIFT算法进行特征点的检测、描述子的生成以及特征点的匹配,以实现图像拼接。实验结果表明,与传统的SIFT算法相比,提出的算法平均拼接速度提升了85%;与现有的SIFT算法相比,提出的算法平均拼接速度提升了20%,因此该算法在对拼接效率有较高要求的领域具有潜在的应用价值。Aiming at the problem of large computation of traditional scale invariant feature transform(SIFT)algorithm,this paper proposed a SIFT image stitching algorithm based on phase correlation and Harris texture classification.Firstly,the phase correlation method is used to obtain the overlapping region of the image to be stitched.Secondly,the Harris corner point response function is used to calculate the texture complexity of the image to be stitched,and the image is divided into strong texture region and weak texture region by setting the threshold.Finally,the SIFT algorithm is used in strong texture areas for feature point detection,descriptor generation,and feature point matching to achieve image stitching.Experimental results show that compared with the traditional SIFT algorithm,the average stitching speed of the proposed algorithm is increased by 85%;Compared with the existing SIFT algorithm,the average splicing speed of the proposed algorithm is increased by 20%.Therefore,the proposed algorithm has potential application value in the field with high requirements for splicing efficiency.

关 键 词：相位相关 Harris纹理分类 SIFT算法 图像拼接 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]