Abstract: Abstract: An improved Scale Invariable Feature Transformation (SIFT) matching algorithm based on second moment matrix is presented to solve the problems that SIFT result in low matching ratio when view point of image is changed. Feature points are detected in scale space, using affine second moment matrix point neighborhood is estimated, then feature vectors are computed by dominant orientation assignment to each feature point based on elliptical neighboring region, finally the feature vectors are matched by using Euclidean distance. The experimental results show that this algorithm is as robust as SIFT, but also acquires good performance on affine invariance of view point change, improves matching results greatly.

Key words: scale space, Second moment matrix, Affine invariance, Scale Invariable Feature Transformation (SIFT) algorithm

摘要： 摘要：为了解决了图像视角变化时造成的匹配率低的问题，作者提出了基于二阶矩的SIFT特征匹配算法。算法在尺度空间检测出特征点，用仿射的二阶矩来估计特征点的椭圆邻域，把椭圆邻域梯度的主方向作为该特征点的方向，生成特征向量，最后采用欧氏距离作为度量函数进行特征向量的匹配。实验表明，改进后的算法继承了SIFT算法对图像缩放、旋转等不变性，而且增强了图像对视角的仿射不变性，极大地改善了匹配效果。

关键词: 尺度空间, 二阶矩, 仿射不变, SIFT算法