一种自适应阈值分块BRISK的图像配准方法

Method of image registration based on adaptive threshold and block BRISK algorithm

提出一种基于自适应阈值BRISK的图像配准方法。首先,根据图像复杂度设置全局初始阈值,并检测图像关键点;然后将图像划分为若干均匀的子图像块,设置子图像块关键点数量范围,根据每个子图像块的关键点数量增加或删除关键点,并对关键点间距离进行约束,使关键点均匀化;最后匹配关键点,使用RANSAC算法剔除误匹配并估算变换模型参数。实验结果表明,本文提出的方法能够自适应地设定阈值,并获得分布均匀的关键点,从而提高图像配准的自动化程度。

An image registration method based on the adaptive threshold BRISK is proposed. Firstly, global initial threshold is set according to image complexity and image keypoints are detected. Then, image is divided into a number of uniform sub-image blocks, the ranges of keypoints of sub-image blocks are set, keypoints are added or deleted based on each sub-image block＇s keypoints number, and the distance between keypoints is limited. Finally, we match the extracted keypoints and use RANSAC to remove the mismatches, and calculate the transformation matrix. Experimental results show that the proposed method can adaptively set thresholds and get keypoints with even distribution, and increase the automation degree of image registration.