摘要
对于图像配准中剔除误匹配的问题,在研究随机抽样一致性算法(RANSAC)的基础上,针对RANSAC算法中具有3个不确定参数,较难设置这些参数以同时得到较高的匹配精度和较短的运行时间,并且匹配精度与运行时间具有一定随机性等问题,提出了采用相似距离比剔除误匹配的算法。实验结果表明:在对有旋转、平移、缩放的图像进行配准时,该算法能够充分利用正确匹配点,缩短了运行时间,具有良好的配准精度和鲁棒性。
To solve the problem of removing mismatched points in image registration, the Random Sample Consensus （RANSAC） algorithm was studied. Since the RANSAC algorithm has three uncertain parameters, and it is difficult to set the parameters appropriately to obtain higher registration precision and shorter operating time simultaneously, an algorithm using analogical distance ratio to remove the mismatched keypoints was proposed. Experimental results showed that： for registration of the images with scaling, rotating and translation, the proposed algorithm can make full use of the correct matching points and reduce the running time with a high registration precision and good robustness.
出处
《电光与控制》
北大核心
2012年第4期35-38,共4页
Electronics Optics & Control
