摘要
基于Harris角点的视网膜图像配准中,Harris角点检测算法无法针对不同的图像设定通用的阈值,检测到的角点也可能存在过多过密的问题。针对这些问题,提出一种改进的Harris角点检测算法,利用角点响应函数的均值来得到自适应阈值,并通过自动控制高斯模糊窗口大小,使角点数量在合理的范围。该算法采用Matlab语言来编程实现。实验结果表明,该方法能较好地配准各种视网膜图像,配准速度约为通用双引导就近点搜索算法的51%。
In the process of retinal image registration based on Harris corner detection, the Harris comer detec- tion algorithm fails to set a universal threshold for each image, and it is a problem that corner detected may be too many. To solve these problems, an improved Harris comer detection algorithm is mean of comer response function to get an auto - adaptive threshold and make the reasonable range by automatically adjusting the size of the Gaussian blur window. Matlab language. Experimental results show that this method can register a variety rate of about 51% of that by the dual bootstrap- ICP algorithm. proposed. This algorithm uses the number of comer points within a The algorithm is implemented in of retinal images at a registration
出处
《电子科技》
2017年第2期119-122,共4页
Electronic Science and Technology
