基于阈值和B样条插值的MR图像增强算法

提出了利用双门限分割灰度级后再进行三次B样条插值非线性变换的MR图像增强算法,将图像灰度级按两个灰度阈值分割为目标区、过渡区、背景区,对这三个不同的区域采用不同的灰度变换方法。在徐军等提出的对比度和目标细节评价标准基础上,提出了一个新的图像质量客观评价标准来评价图像质量。该客观标准可以动态调节对比度和细节的权重参数!,具有交互性。通过寻求该标准最优时的三次B样条插值非线性变换来增强MR图像。实验表明,和目前主要的灰度图像增强算法相比,用该算法增强后的图像不仅提高了图像对比度,也加强了目标的细节,而且具有交互性,特别适合于MR图像处理。

MR Image Enhancement Algorithm Based on Threshold and B-Spline Interpolation

A new algorithm for MR image enhancement is proposed.First the bi-threshold segmentation is implemented,in which the gray-level of a MR image will be segmented into three sections:object section,transition section and background section.Second the cubic B-Spline interpolation transform is carried out on the three sections.For the best visual effect,different transforms will be taken for different sections.Then,an assessment criterion based on luminance contrast and detail,which can adjust the weight parameter dynamicly,is proposed for evaluating the image enhancement effect.Finally,the best cubic B-Spline interpolation transformation is searched for ideal enhancement effect.Experimental results show that the algorithm outperforms some other enhancement methods from the view point of the luminance contrast as well as the detail assessment parameter.In addition,the algorithm is interactive,especially suitable for MR image processing.