摘要
图像拼接技术是近年来医学图像处理的研究热点之一,在眼科方面具有广泛的研究价值。笔者根据眼底图像的特点,提出了一种基于模板匹配法的眼底图像拼接方法。首先在两幅图像的重叠区域中分别抓取一对初始匹配点,分别以抓取点为中心取不同大小的模板区域和搜索区域,然后根据模板匹配法的基本原理,确定最佳匹配点,通过平移和旋转变换完成图像之间的配准,最后采用重叠区域线性过渡的图像融合方法消除拼缝,实现图像间的平滑过渡。经实验验证,该方法实现了不同视角眼底图像之间的拼接,且计算速度快,精度高。
Image mosaicing is a hot spot of medical image processing technology in recent years, and has broad value of the research in ophthalmology. This paper proposed a fundus image mosaic method based on template matching according to the characteristics of fundus images. First, grab a pair of initial matching points in the overlapping area of two images, use these points as the center to get different size of template region and search region respectively, then fix the optimum matching points according to the template matching method. And then adjust the images through translation and rotation. At last eliminate edge joint by linear transition of overlapping regional to achieve a smooth transition between images. After experimental verification, fundus images of different perspectives were matched by this method which also has quick calculation speed and high accuracy.
出处
《计算机技术与发展》
2010年第4期148-150,154,共4页
Computer Technology and Development
基金
国家高技术研究发展计划(863计划)(2006AA02080477)
关键词
眼底图像
图像配准
图像融合
模板匹配法
fundus image
image registration
image fusion
template matching
