Abstract: Registration is widely used in medical imaging applications. It is time-consuming due to its huge computation and a new rapid registration framework was presented. The inputting data to the framework include two images: fixed image and moving image. The outputting data were the result image representing the differences between the fixed image and the moving image after registration. Besides the inputting and outputting data, the framework can be divided into four parts: interpolator, measurer, optimizer and transformer. Interpolator was used for evaluating moving image intensities at non-grid positions. Measurer evaluated how well the fixed image was matched by the transformed moving image. Optimizer can optimize the measure criterion and transformer exerted some transformations on the objective image. These four parts acted as different roles in medical images registration and constructed a simple, rapid and stable medical images registration framework. Compared with other registration frameworks, the proposed framework was quite simpler in structure but much quicker in image processing and application development. Good results have been obtained in practical applications.

Key words: framework, image registration, mutual information, Genetic Algorithm (GA)

摘要： 图像配准是医学影像处理中的一项常见技术，由于配准操作计算量大，非常耗时，因而设计了一种快速配准框架。输入框架的数据包括一幅固定图像和一幅待配准图像（移动图像），输出数据是包含两张图像差异结果的图像。除了输入和输出数据，整个框架包括四个组成部分：插值器、度量器、优化器和变换器。插值器用于测定移动图像映射后像素点的灰度值，度量器用于评价变换之后的移动图像和固定图像之间的匹配度，优化器用于优化度量规则，变换器对目标图像实施各种几何变换处理。这四个组成部分在图像配准操作中分别担任不同的角色，从而构建出一个简单、快速、稳定的医学影像配准框架。和其他图像配准框架相比，该框架在结构上更简单，在配准处理和程序开发方面更快捷，在实际应用中取得了良好的效果。

关键词: 框架, 图像配准, 互信息, 遗传算法