面向移动设备的图像匹配算法适用性研究  

Study on the Applicability of Image Matching Algorithm for Mobile Equipment

作　　者：黄碧辉 吴勇[1,2] 郑森源 

机构地区：[1]福建师范大学地理科学学院,福建福州350007 [2]福建省陆地灾害监测评估工程技术研究中心,福建福州350007

出　　处：《海南师范大学学报（自然科学版）》2017年第4期407-413,共7页Journal of Hainan Normal University（Natural Science）

基　　金：国家自然科学基金青年项目(41201417);福建省自然科学基金项目(2016J01186);福建省省属公益类科研院所专项(2017R1034-4);福建省测绘地理信息局局校合作科技项目(2016JX02);福建师范大学地理科学学院研究生科研创新基金资助

摘　　要：图像匹配技术作为增强现实的核心问题受到广泛关注.现有的研究主要集中在PC端图像匹配算法的鲁棒性和具体实现方面,而对计算能力较弱、易受环境因素影响的移动终端上实现图像匹配的鲁棒性、适用性等方面探讨较少.针对该问题,选取常用的ORB、BRISK、SURF、FAST_FREAK、FAST_SURF以及SURF_FREAK图像匹配算法,以移动终端为实验平台,从匹配效率与鲁棒性两个方面对算法进行比较分析,并采用正确匹配点对和匹配分数两个指标对算法进行综合性能评价.实验结果表明:ORB算法具有较高的运行效率,同时该算法对旋转、光照、尺度、视角等变换具有较好的鲁棒性,适合于移动增强现实的实时匹配应用.Image matching technology, as the core issue of augmented reality, has received widespread attention. Now most research is mainly about how to implement algorithm on the computer and robustness of the algorithm, while the perform-ance ,robustness and applicability of the image matching algorithm implemented on mobile devices that has weaker perform-ance and easily affected by outdoor environmental factors are few discussed. To solve this problem, we select the commonly used ORB, BRISK, SURF, FAST_FREAK, FAST_SURF and SURF_FREAK image matching algorithm which uses mobile terminal as the experimental platform to compare and analyze the algorithm from efficiency and robustness, and the numberof correct matches and matching score of the two indices are used to evaluate the performance of the algorithm when matc-hing image is transformed. The experimental results show that ORB can meet the requirement of real - time performance.At the same time, the ORB is robust to rotation, illumination, scale and angle of view, and it can better meet the practical requirements of mobile augmented reality.

关 键 词：图像匹配 特征点检测 移动终端 增强现实 鲁棒性 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]