基于SURB结合随机抽样一致算法在鞋面匹配中的应用

Application of SURB Combined with Random Sample Consensus Algorithm in Shoe Uppers Matching

针对鞋面匹配中存在的尺度变化、光照变化以及噪声干扰等问题,提出基于加速稳健特征和对象请求代理(SURF-ORB)算法结合随机抽样一致(RANSAC)算法的鞋面匹配检测算法。采用SURF算法提取鞋面图像特征点;通过ORB算法对提取到的特征点进行描述,得到描述子;采用汉明距离完成初匹配,再结合RANSAC算法对由噪声干扰和光照变化而产生的误匹配点进行剔除,获得较为精准的匹配点对。结果表明:当鞋面图像中存在尺度变化、光照变化和噪声干扰等影响时,该算法能够准确匹配,具有较强的稳健性。

Aiming at the problems of scale change, illumination change and noise interference in the uppers matching, a shoe upper matching detection method based on the speeded-up robust features-object request broker （SURF-ORB） algorithm combined with random sample consensus （RANSAC） algorithm is presented. The feature points of the uppers image are extracted by SURF. The descriptors are obtained and the feature points are described by the ORB algorithm. In order to obtain more accurate matching points, the initial matching is completed by using the Hamming distance, and then by combining the RANSAC algorithm, the mismatching points generated by noise interference and illumination changes are eliminated. The experimental results show that the algorithm can effectively match and has strong robustness when there are scale change, illumination change and noise interference in the shoe uppers image.