Study On The Algorithms Of Example-based Color Transfer

Example-based color transfer adjusts colors of target images based on color characteristics of the example ones. It can borrow the color characteristics of the sample image and maintain its shape features at the same time. It is an emerging technology and one of the hottest areas in computer vision and image processing fields.All of the algorithms are operated in 1αβcolor space, which can minimize correlation among the three channels and makes it possible to change the color value of each channel. Luminance remapping and Gaussian pyramid decomposition were used. In pixel characteristics using its own pixel brightness values, has also taken into consideration the characteristics of their neighborhood in addition, so that guaranteed the result in the color transfer effects and rate. Programming the typical color transfer algorithms with Visual C++ and through a large amount of experimental testing, analysis of the existing transmission color theory and characteristics of the algorithm, the algorithm summarized the advantages and disadvantages as well as their application.Based on the above-mentioned work, a new color transfer algorithm based on SOFM was proposed. It used the multidimensional eigenvector as pixels' feature. The algorithm use pixel's multi-dimensional neighborhood features as a pixel characteristics. In the training step, Gaussian pyramid decomposition was used in order to reduce the size of the sample image without loosing color information. After training, relationship between pixel characteristics and color value was established. The algorithm improved color transfer the accuracy of the matching process, improved color transfer effects. Through comparative analysis of the results shows that the algorithm has good results and some practicality for further research has laid the foundation.