基于对齐度和模糊梯度相似性的图像配准

Image Registration Based on Alignment Metric and Fuzzy Gradient Similarity

提出一种基于对齐度和模糊梯度相似性的图像配准方法.该方法利用小波多尺度积提取边缘图像和特征点,然后利用角度直方图估计旋转角度,再使用提出的基于对齐度和模糊梯度相似性的配准准则来确定匹配点对.实验结果表明,该方法精确性较高,并具有一定的抗噪性.

A new approach for image registration is proposed based on alignment metric and fuzzy gradient similarity. The proposed approach applies wavelet mult-scale product to extract edge image and feature point. Then angle histogram is calculated to estimate the rotation angle, and then proposed registration rule based on alignment metric and fuzzy gradient similarity is applied to determine matching pairs. Experiments demonstrates the accuracy、efficiency and robust of the approach.