摘要： 针对仿射扭曲图像对之间的匹配问题，传统算法对仿射变换的6个参数都进行归一化，在仿射畸变严重时匹配效果不佳，不具备完全的仿射不变性。为此，提出一种基于参数仿真的图像匹配算法。根据经纬度对摄像机坐标的角度参数进行仿真，利用Harris-Laplace算法提取特征点，使用尺度不变特征变换对其进行描述，由此仿真尺度参数，并归一化平移和旋转参数。实验结果表明，与MSER、Harris-Affine、Hessian-Affine算法相比，该算法的鲁棒性较强，能使所有参数都具有不变性，具有完全的仿射不变性。

关键词: 图像匹配, 仿射不变性, 特征点, Harris-Laplace算法, 特征提取

Abstract: Most traditional algorithms normalize the six parameters of an affine transformation, no fully affine invariant approach exists for the time being. Consequently, an affine invariant algorithm for image matching is proposed in this paper. This method simulates all image views obtained by varying the two camera axis orientation parameters. Then it covers the other four parameters by using Harris-Laplace detector and Scale Invariant Feature Transform(SIFT) descriptor. The new method allows to reliably identify features that have undergone significant affine transformations including large viewpoint and scale changes. This fact is substantiated by many experiments which present a comparative evaluation of different methods and results show that the new algorithm outperforms significantly the state-of-the-art methods SIFT, MSER, Harris-Affine, and Hessian-Affine algorithms.

Key words: image matching, affine invariance, feature point, Harris-Laplace algorithm, feature extraction

中图分类号: 

• TP391.41