结合NGVF信息和互信息的图像配准方法

对传统的互信患配准方法进行改进,提出一种新的结合灰度信息和空间信息的配准方法.该方法分为粗略和精确配准两个阶段,首先利用图像的空间信息进行粗配准,再利用ACMI进行精确配准.文中利用的空间信息不同于一般的图像梯度信息,它是归一化的GVF矢量场信息(NGVFI),把它用于粗配准,抗噪性能良好.试验证明该方法鲁棒性很好,特别是对大噪声图像,配准效果突出,精度能达到亚像素级.

An Image Registration Method Based on the Combination NGVF Information and Mutual Information

Based on the conventional MI approach,we propose a new registration method with the combination of intensity and spatial information.It includes two stages,one is coarse registration with spatial information of image,and the other is fine registration with ACMI.This spatial information is normalized gradient vector information (NGVFI) instead of ordinary gradient information,and it can enhance the noise resistibility greatly during coarse registration.The experimental results show that our method is very effective and robust,and the registration accuracy reaches sub-pixel successfully.