基于ASIFT算法的图像特征匹配

Image Feature Matching Based on ASIFT Algorithm

ASIFT算法原理是模拟仿射矩阵的两个参数,即相机与图像法线的倾斜角θ和旋转角?,从而精确的模拟出来所有由相机光轴方向改变所造成形变的图像,然后利用SIFT算法对这些模拟图像进行特征提取与匹配。ASIFT算法是一种完全仿射不变图像特征匹配算法,阐述ASIFT算法步骤,经度角和维度的采样过程,ASIFT采样范围和采样步骤。实验结果表明ASIFT算法可以对图像匹配实现有效准确的匹配。

The principle of the ASIFT algorithm is to simulate the two parameters of the affine matrix, that is, the inclination angle 0 and the rotation angle φ of the camera and the normal of the image, so as to accurately simulate all the images caused by the change of the optical axis of the camera, and then use the SIFT algorithm Feature extraction and matching of these simulated images. The ASIFT algorithm is a complete affine invariant image feature matching algorithm, which describes the ASIFT algorithm step, the longitude and dimension sampling process, the ASIFT sampling range and the sampling step. The experimental results show that the ASIFT algorithm can achieve accurate and accurate matching of image matching.