摘要
文中提出了一种新的图像配准方法,可以很好解决图像配准中的平移和旋转问题(刚性变换问题)。算法的实现是首先得到待配准两幅图像的轮廓特征点(本文使用SUSAN边缘检测算法);再剔除虚假的特征点(噪声点)后,根据像素间的连通性判别准则,得到图像的主轮廓;最后以两幅图像的主轮廓信息,配准图像。实验证明,算法实现速度快,精度较高。
A new algorithm about image regisu：ation is presented in this paper, which has solved translations and rotations transformations （so-called rigid transformations） problems in image registration process well, This approach is as followed： first, we use SUSAN edge-detected method to obtain the feature points （interesting points） of the two images, and from which, we eliminate the bad-feature points （the noise points）; second, by using connectedness-judged criteria, acquire the main outlines of the two images; third and last, we use the information extracted from the main outlines to register the two images. The experiments have test the effect of the approach.
出处
《微计算机信息》
北大核心
2007年第02X期267-268,240,共3页
Control & Automation
