Abstract: Adaptive coloring grayscale image based on multiple color images competing selection model is proposed.Firstly，this paper transforms the RGB color space to lab color space to remove the correlation in RGB color channel.Then the image information retrieval technology is introduced to select some optimal similar source color images from given color image database which match with the contents of input gray image，and then using similar competition and adaptive sampling analysis of the partial and total information to choose the best color image information and transfer it to the gray image to synthesize a new color image adopting according pixel corresponding principle.This scheme combines the automatic image retrieval technique with color transfer together，so it can make good use of color information in multiple pieces of color images to improve color performance.The experimental results demonstrate that this algorithm has good color effects and a more extensive scope of application compared with the classic method.

Key words: image coloring, color transfer, color space, content-based image retrieval

摘要： 提出一种基于相似性竞争选择的多彩色图像自适应颜色混合迁移算法。首先将彩色图像的颜色空间转换为l?琢?茁空间，然后利用亮度图像信息熵检索选择与灰度图像内容最匹配的多幅彩色图像作为彩色参照图像；然后利用相似性竞争选择进行局部和整体相结合的自适应采样分析，采用像素点对应原则，选择参考图像中最佳颜色信息迁移到灰度图像中，合成一幅崭新的彩色图像。该方法将自动图像检索技术与颜色迁移相结合，达到充分利用多幅彩色图像中彩色信息的目的。最后给出实验结果表明，该算法相对于经典算法具有更好的颜色迁移效果和更加广泛的适用范围。

关键词: 图像彩色化, 颜色迁移, 颜色空间, 图像检索