Introduction
Fast algorithms of transferring color to grayscale images are widely used in image processing. It transfers color information to a grayscale image, then the grayscale image has the similar color distribution with the color one. It has been one of the most practical techniques.
Many methods have been developed for color transfer [1], [2] and [3]. Welsh [4] proposed an algorithm due to color component color component. However, this method suffers from the ambiguity in image details and the background error. Pitie [5] improved it with N-dimensional probability density function. However this method suffers from time-consuming. Guoying Zhao [6] adjusted the slope of the source image data to preserve some details. Irony [7] or Zhenhua Li [8] proposed colorizing algorithms based on texture match or wavelet. However all of the above methods need a large number of complex algorithms and caused the background error.
To achieve better results, some new approaches based on manual intervention have been proposed. Li and Hao [9] proposed a learning-based colorizing algorithm, which will easily cause color distortion. Levin [10] proposed a colorizing algorithm based on chromaticity statistics. Qu [11] and Luan [12] improved these methods using constraint of texture similarity, which significantly reduced the complex texture error diffusion. Additionally, Guofei Hu [13] and Horiuchi [14] proposed adaptive colorizing algorithm and probability relaxation algorithm. However these methods are complex and targeted.
In this study, we propose a new algorithm based on sorting pixels comparison. The proposed comparison of rearranged pixels color transfer method is compared with conventional color transfer methods.
The rest of the paper is structured as follows. In Section 2, works related to rearranged pixels, including in lab color space. In Section 3, the low-complex equalization method is presented. Experimental results are demonstrated in Section 4. And Section 5 presents our discussions and conclusions.