结合拓扑纹理图像局部不变特征的医学影像光流场配准  

Medical image registration based on local feature of topological texture image and optical flow model

作　　者：于文勇[1] 康晓东[1] 葛文杰[2] 郭宏[1,3] 王昊[1,4] 耿佳佳[1] 

机构地区：[1]天津医科大学医学影像学院,天津300070 [2]河北工业大学信息工程学院,天津300400 [3]天津医科大学总医院,天津300070 [4]河北大学附属医院,河北保定071000

出　　处：《计算机应用》2014年第A01期206-210,共5页journal of Computer Applications

摘　　要：医学影像图像是一类拓扑纹理图像。在临床诊断和治疗中,常需要将不同模态的医学影像图像配准。为此提出一种结合拓扑纹理图像不变特征的医学影像光流场配准的方法。首先,以尺度不变特征变换(SIFT)描述子对待配准图像做局部特征描述,同时,以一致性提纯法对图像局部特征匹配对提纯;其次,以图像局部不变特征为参考,对图像进行基于光流场模型的弹性图像配准,同时,对配准模型予以单元分解,以避免图像局部拓扑纹理特征因被光滑而模糊配准的结果。仿真实验结果表明,新方法具有较好的医学影像图像配准效果。Medical image is a kind of topological texture image. In clinical diagnosis and treatment, it is often needed to match medical images of different models. This paper put forward a medical imaging optical flow field registration method in conjunction with invariant features of topology texture image. Firstly, Scale Invariant Feature Transform （SIFT） descriptor was used to describe local characteristics of image registration and image registration on local imaging characteristics is refined by the eonsistency purification method. Secondly, the image was matched on the basis of elastic image registration of optical flow model by reference to the image local invariant features, and the registration model was decomposed by units to avoid blurred registered result and loss of local topological image texture features because of being smoothed. The result of simulation experiment indicates that the new method has better effect on medical image registration.

关 键 词：图像配准 局部特征 尺度不变特征转换 特征提取 光流场模型 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]