摘要： 为解决传统的彩色化算法中效果不佳、速度慢等问题，提出一种基于遗传算法的图像彩色化方法。该方法先产生初始种群，将源图像块和目标图像块的亮度特征以及Tamura纹理特征构成适应度函数，初始种群经过选择、交叉和变异等操作逐代进化后，将源图像块的颜色转移到匹配的目标图像块中，得到最终的彩色化图像。实验结果表明，与其他彩色化方法相比，该方法能在缩短运行时间的同时，有效提高彩色化质量。

关键词: 彩色化, 颜色转移, 遗传算法, 适应度函数, 纹理特征

Abstract: A novel image colorization algorithm based on genetic algorithm is proposed aiming at overcoming unsatisfactory and inefficiency in the traditional colorization algorithm. Initialization of the chromosomes is introduced in the proposed method and then the luminance feature and Tamura texture feature of object block image and source block image are adopted as the fitness function, with which basic principles of nature evolution, selection, crossover and mutation are applied. The colorized image is effectively acquired by transferring the color from source block images to matched object block images. Simulation results show that the proposed algorithm outperforms previous method in speeding up the colorization process and meanwhile improving the image quality.

Key words: colorization, color transfer, Genetic Algorithm(GA), fitness function, texture feature

中图分类号: 

• TP301.6