SURF算法在局域盲人图像导航中的应用

局域盲人图像导航即在建立基准图像信息特征库的基础上，对实时拍摄图像提取特征并与基准图像特征库进行匹配，得到相应地理信息并语音输出给盲人，从而实现导航的一种方法。本文基于SURF算法对图像特征进行匹配，首先提取SURF特征点，然后采用Hessian矩阵迹快速索引匹配以及匹配点距离差平方和的相似性度量方法进行匹配。实验表明SURF匹配算法优于SIFT匹配算法，并可实现快速、鲁棒、准确匹配，为实现盲人局域图像导航奠定理论基础。

Application of SURF Algorithm in Local Image Navigation for the Blind

Local image navigation for the blind, a method of navigation which is based on the establishment of database of reference image information characteristics, captures the featuros of real-time image and matches to reference signature images to get the correspond- ing geographic information and output to the blind. Based on SURF algorithm, this article matches image features, it extracts the feature point of SURF, lastly indexes to match by Hessian matrix trace, and then matches by similarity measure with the sum of square of points distance difference. The results show that SURF matching algorithm is better than SIFT matching algorithm, it is fast, robust and accurate, which lays a theoretical basis for achieving local image navigation for the blind.