摘要
研究了基于互信息量的图像拼合方法,对现有的概率分布估计方法及搜索策略进行了比较,并提出用梯度二值化图像互相关配准做预处理方法,并就融合后图像的显示问题提出一种基于边界法向插值的新方法.应用这些方法拼合实际XA医学图像,实现序列下肢步进图像的准确拼合.
This paper looks further at the method of medical image patching based on mutual information. We have compared the methods of density estimation with the searching strategy of image registration. At the same time, a new way of image preprocessing based on grads information has been used as well as a novel display method, which introduced directions of the edge curve into image interpolation. A good patching result has been made in our scheme.
出处
《北方交通大学学报》
CSCD
北大核心
2004年第2期38-40,45,共4页
Journal of Northern Jiaotong University
基金
北京交通大学"十五"重点基金项目资助(2002Z004)
关键词
图像拼合
图像配准
互信息量
图像熵
image patching
image registration
mutual information
image entropy