摘要
针对传统SIFT算法很难兼顾匹配实时性与匹配准确率这一矛盾,提出一种融合彩色信息表征与光照补偿的Harris-SIFT算法。该算法在图像灰度化之前计算彩色信息补偿,并通过高斯权重函数得出光照补偿量,增强图像对比度,进而提高系统的匹配准确率。通过Harris角点检测算法与SIFT算法融合提高系统的快速性并剔除表征不好的特征点,改善系统的实时性。实验结果表明匹配算法在匹配准确率提高的同时,还提高了算法的实时性。
This study proposes a Harris-SIFT algorithm that combines the color information representation and illumination compensation to resolve the contradiction between matching real time and matching accuracy of the traditional SIFT algorithm.The algorithm calculates the color information compensation before the image is grayed.By using the Gaussian weight function,the illumination compensation amount is obtained,the image contrast is enhanced,and the matching accuracy is improved.The Harris corner detection and SIFT algorithms are combined to improve the system rapidity and eliminate badly characterized feature points,improving the real-time system performance.Further,the experimental results denote that the proposed matching algorithm improves the real-time performance of the algorithm while improving the matching accuracy.
作者
张庆鹏
曹宇
Zhang Qingpeng;Cao Yu(School of Automation,Harbin University of Science and Technology,Harbin,Heilongjiang 150080,China)
出处
《激光与光电子学进展》
CSCD
北大核心
2019年第19期112-118,共7页
Laser & Optoelectronics Progress
基金
黑龙江省普通本科高校青年创新人才培养计划项目(UNPYSCT-2015045)
