摘要
为了克服传统的局部特征匹配算法对图像的尺寸和旋转比较敏感的缺点,文章提出了一种基于Harris和SIFT的特征匹配算法。该算法首先进行Harris角点提取,利用圆的旋转不变性和多维向量来构造SIFT特征描述子,然后利用极线约束降低对匹配点的搜索空间,最后度量特征描述子之间的欧氏距离,运用双向匹配技术得到最终的匹配结果。实验结果表明,该算法能提高匹配精度,减少匹配时间。
In order to restrain the high sensitivity to image size and rotation as for conventional local matching algorithms,in this algorithm,Harris corner detector is used to extract feature point firstly,make use of the characteristic that circle is rotation-invariant and multi-dimensional vector to constructs the SIFT feature descriptor.Then,polar constraint is used to reduce the search space for match point.Matching results are obtained by measure the Euclidean distance between the feature descriptors and apply double-sided matching technology.Experiments demonstrated the algorithms can enhance matching accuracy and shorten the matching time.
出处
《价值工程》
2012年第6期2-3,共2页
Value Engineering
基金
国家自然科学基金重大专项项目(90820302)
国家博士点基金资助项目(200805330005)
