基于改进遗传算法的医学图片颜色迁移及重构

MEDICAL IMAGE COLOUR TRANSFER AND 3D RECONSTRUCTION BASED ON IMPROVED GENETIC ALGORITHM

为了得到有效和符合真实人体器官组织的彩色仿真图片,通过灰色MRI切片和真彩图像的亮度及纹理的适应度的搜索取优而实现灰度MRI切片的颜色迁移。着重研究将改进遗传算法与点邻域亮度纹理分布和局部颜色结构分布相结合,并应用于Welsh图像彩色化算法和比较,探索用于遗传算法的种群选择、交叉和变异、编码方式和码值渐变自适应等调整技术和方法。将点邻域分布和彩色局部分布的彩色特征通过穷举法、随机法和遗传算法进行最优彩色点搜索分析比较。最后得到的彩色化的MRI切片的三维重构模型能多层次清晰地反映器官组织的分布和构造。

In order to gain the colourful simulated image of human organs and tissues which are valid and in line with the real, we realise the colour transferring of a gray MR! slice by searching and picking up the optimal fitness of a gray MRI slice to the brightness and textures of a true colour image. In this paper we particularly study the combination of the improved genetic algorithm with the distribution of brightness and textures within pixel neighbourhood and the local colour structure, and apply our study to Welsh image colourisation algorithm and comparison, probe the adjustment technologies and ways of population selection, crossover and mutation, coding mode and code value gradual varying adaptiveness, all are used in genetic algorithm. By using exhaust algorithm, random algorithm and genetic algorithm, colour characteristics with pixels neighbourhood distribution and with local colour distribution are analysed and compared for best colour points search. The reconstructed 3D model of colourised MRI slices eventually derived can clearly and in multi-level reflect the distribution and construction of human organs and tissues.