Abstract: This paper presents an image registration method based on edge features.A registration parameter is obtained by extracting the edge of an image via wavelet transformation and substituting the artificially selected edge points in the affine transformation model（selecting a different set of edge points would result in a different registration parameter）.Then，the interactive variance of two images is calculated under the condition of different registration parameters.The registration parameter corresponding to the minimum interactive variance is taken as the final registration parameter.Finally，the registered images are obtained by translating，rotating，and/or reducing/magnifying the images to be registered through using the affine transformation model again.

摘要： 采用一种基于边缘特征的图像配准方法，首先通过小波变换来提取图像的边缘，然后将人工选择的边缘点代入仿射变换模型，得到配准参数（每选择一组不同的边缘点，就会得到不同的配准参数）。在不同的配准参数条件下，计算两幅图像的交互方差。取交互方差最小时所对应的配准参数为最终的配准参数。最后再利用仿射变换模型对待配准图像进行平移、旋转、缩放得到最终的配准图像。