改进HD距离和遗传算法的图像匹配研究  

Research on Image Matching of Improved HD Distance Genetic Algorithm

作　　者：刘晋胜[1] 彭志平[1] 周靖[1] 

机构地区：[1]广东石油化工学院计算机与电子信息学院,广东茂名525000

出　　处：《电视技术》2011年第15期46-49,共4页Video Engineering

基　　金：广东省自然科学基金项目(8152500002000003)

摘　　要：针对高分辨力图像匹配点难于确定且运算量大的问题,提出一种加权Hausdorf(fHD)距离和多策略并行遗传算法相结合的图像匹配算法。算法采用概率加权Hausdorff距离,减少噪声、出格点等对匹配的影响,并利用多策略并行遗传算法的并行处理能力对概率加权Hausdorff距离进行寻优,提高匹配速度。试验结果表明,该算法对噪声、角度旋转及遮挡具有较强的稳健性,算法的匹配精度和运算速度都有一定的提高。Aiming at the problems of that matching points for high-resolution image is difficult to determine and the computation is large, an algorithm combined with weighted Hausdorff distance and multi-strategy parallel genetic algorithm is proposed. The probability weighted Hausdoff distance is used to reduce noise, outliers impact for matching, and the proposed algorithm takes the ability of parallel processing of multi-strategy parallel genetic algorithm to optimize Hansdorff distance for improving the matching speed. The experimental results show that the proposed algorithm has robustness for noise, rotation angle and block, and improve the matching speed and the computing speed effectively.

关 键 词：图像匹配 HAUSDORFF距离 多策略并行遗传算法 

分 类 号：TN911.73[电子电信—通信与信息系统]