摘要
提出了一种仿射医学图像融合方法。以医学图像的象素灰度信息为融合依据,首先根据参考图像的象素值对参考图像进行划分;然后根据该划分对浮动图像进行划分,并假设被划分在同一组中的象素有相似的灰度值;最后迭代地对浮动图像进行仿射变换使得浮动图像中的被划分在同一组中的各个象素间的标准方差最小,实现融合图像象素之间的匹配。由于算法充分利用图像的全部象素的灰度信息,不需对图象进行滤噪等预处理,从而提高算法的精度;该方法采用仿射变换,因而不仅可纠正刚体形变而且可纠正放缩形变,适用于单模及模板融合。
An affine alignment algorithm is proposed for the processing of medical images. The reference image is segmented into groups according to the gray scale of pixels of the medical image. Supposedly the pixels with the same gray scale are segmented into the same group, the study image is segmented in the same way. The study image is processed iteratively by the affine transformation in order to minimize the standard variance of pixels in the same group. Finally, the pixels of the original images can be matched properly. Because the total intensity information of each pixel is used in the algorithm, it is not necessary to perform other image processings, such as de -noising. So the original information of the image could be kept as much as possible. The affine transformation algorithm can be used to correct the rigid and non-rigid displacement. The algorithm is applicable to the mono-and multi - modality image fusion.
出处
《黑龙江大学自然科学学报》
CAS
2004年第1期75-78,共4页
Journal of Natural Science of Heilongjiang University
基金
黑龙江大学杰出青年科学基金资助项目(2002)