Surf结合区域生长的特征点提取改进算法

Improved algorithm for feature point extraction based on combination of Surf and Regional Growth

为解决Surf算法提取边缘光滑图像特征点能力差的问题,提出了一种基于Surf与区域生长算法相结合的特征点提取改进算法Surf-RGA。通过局部平均值和Harr小波改进Surf算法,选取自动种子点,利用Matlab软件读取具有边缘光滑特点的篮球样本图像,分别运用Surf和Surf-RGA算法提取其特征点,验证算法的可行性。结果表明,Surf-RGA算法提取的光滑图像边缘特征点能力远好于原Surf算法,特征点提取数量增加率平均值约为420.69%,有效提升了Surf算法对图像特征点的提取能力。

This paper aims to address the poor ability of Surf algorithm to extract feature points from smooth edge images and proposes an improved algorithm Surf-RGA for feature point extraction based on the combination of Surf and Regional Growth algorithm.The study involves improving Surf algorithm by local average and harr wavelet and thereby selecting automatic seed points;reading basketball sample images with smooth edges using Matlab software;and extracting their feature points using Surf and Surf-RGA algorithms to verify the feasibility of the algorithm.The results show that Surf-RGA algorithm boasts a greater ability to extract smooth image edge feature points than the original Surf algorithm,with the average increase rate of feature points extraction of about 420.69%,providing the effectively improved ability of Surf algorithm to extract image feature points.