Analysis of the Influence of Color Space Selection on Color Transfer

With the development of deep learning technology, the automated grayscale image colorization methods have been widely used. As color space is very important in colorization, we analyzed 3 different color spaces, named YUV, Lab, and HSV. We used deep learning methods to evaluate the effect of different color spaces to image colorization methods. The colorization method was based on the VGG16 and the Residual Encoder model. The results demonstrated that the quality of color transfer in the YUV and Lab color spaces is better than that in the HSV color space.