基于角点检测图像配准的一种新算法

提出了一种新的基于角点检测的图像配准方法，其核心思想是采用一种快速的基于图像灰度的角点检测新算法，通过沿圆弧曲线扫描获取角点信息，然后根据这些角点信息建立图像间角点的对应关系，并由此得到初配准参数，最后通过迭代过程以提高配准的精度。理论分析和实验结果表明，该算法对图像间的旋转角度没有限制，配准精度高而且计算量较小。

A New Method of Image Registration Based On Corner Detection

A new method of image registration is presented based on a fast corner detection algorithm on grey level image.Corner information can be achieved by scanning along the circle. After the corners of different frames of the same scene are detected, the matching corner pairs between these frames can be found. Based on these pairs, initial registration parameters can be estimated. And then, an iterative scheme is performed to refine these registration results. The experiments show that this method has no restriction to the rotation angle between the frames and is computationally efficient and that the registration results are of high precision.