一种新的基于图论的图像匹配算法  

A NOVEL IMAGE MATCHING ALGORITHM BASED ON GRAPH THEORY

作　　者：伏雪[1] 马燕[1] 林涛[1] 

机构地区：[1]上海师范大学计算机科学与工程系,上海200234

出　　处：《计算机应用与软件》2016年第12期156-159,共4页Computer Applications and Software

基　　金：国家自然科学基金项目(61373004)

摘　　要：SIFT算法在图像匹配领域中占有重要地位,但是,利用SIFT算法提取的图像特征点,是分布在整幅图像中的,这就造成提取的特征点不集中。结合图论的方法,对SIFT算法提取的特征点进行处理,去除部分不集中的点,从而达到提高匹配效率的目的,将该方法命名为G-SIFT算法。G-SIFT算法提取的每一个特征点视为图的顶点,将这些顶点的一元关系视为图的边,并利用这些边的大小特点对特征点进行处理,使得处理后的特征点主要集中在物体上。实验证明,利用该方法对图像进行处理后,特征点匹配结果更加集中,匹配率最高提高了1.4%,匹配点集中在物体上的正确率最高提高了9.1%。SIFT algorithm plays an important role in the field of image matching. However, the feature points extracted by SIFT algorithm are distributedin the whole image, which leads to the problem that the feature points are not concentrated. Thus, a new G-SIFT algorithm based on the SIFT algorithm is proposed. The G-SIb^F algorithm combines the graph theory with the SIFT algorithm, removing the SIFT feature points, which are not concentrated. In the graph theory, every feature point is treated as a vertex andthe unary terms of these vertices are taken as edges of graph. Then the feature points are processed according to those edges. The experiments show that the feature points matching results are more concentrated, not only the matching rate is improved by 1.4% but also the concentration rate is improved by 9.1%.

关 键 词：SIFT G-SIFT 图论方法 特征点匹配 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]