一种面向文本图像的颜色迁移算法

A color transfer algorithm for text image

Reinhard等人提出的经典图像颜色迁移算法,对普通场景的自然图像进行颜色迁移时有着显著的效果,而该算法用于文本图像迁移中,迁移后文本字体边缘存在弱化现象,视觉效果不佳。为此,该文提出了一种针对文本图像的颜色迁移算法。通过对源图像和参考图像进行引导滤波增强,突出文字本身在图像中所占比重。利用增强后的源图像和参考图像,计算颜色迁移系数,使得文本图像的颜色迁移效果更加出色。实验结果表明,该文所提算法对于文本图像具有很好的颜色迁移效果,其图像平均梯度结果优于Reinhard算法。

The classic color transfer algorithm,proposed by Reinhard,has achieved the remarkable result for the general scene image. However,it has a few limitations in treating the text image. After color transfer,the font edge of the text image is weakened,and the visual effect looks poor. Hence,a new color transfer algorithm is proposed in this paper. Firstly,the source image and reference image are enhanced based on the guide image filtering respectively in order to enhance the proportion of text itself in the image. Then the critical color transfer coefficients are calculated according to the enhanced source image and reference image,which makes the color transfer effect of text image more outstanding. The result of experiments shows that the proposed algorithm has good color transfer effect for text images,and the average gradient of images is better than Reinhard algorithm.