Research And Application Of Image Color Transfer

Color transfer is a main field in image processing,which plays an important role in improving the effect of render or changing texture styles.The main problem need to be solved is the mapping between the two color distributions.This paper by changing the weight of different colors to weak the effects of noise,and gives the optimal transport to get the final color mapping.The search contents include:(1)Cause the existing color transfer algorithms has a weakness for noise sensitivity,this paper proposes a method of the transfer images based on information entropy and optimal transport mapping.This method uses the information entropy to define the similarity between the two colors spectral by using the color clustering between the reference image and the target image,and optimization the weight of different colors through the L-BFGS algorithm.Finally,the optimal transport theory is used to calculate the color mapping between the input image and the output image.(2)The regularization of graph structure.Base on the clustering information of color clustering adjacency,the topology of adjacency graph is established.And optimizing the optimal transport graph to regularization of the discrete operator and optimal transport of topological structure.(3)Make the medical images colored by color transfer and segmented the image through threshold segmentation method.Result shows the feasibility of color transfer in medical images areas.It is proposed that the optimal transport theory and information entropy or optimal transport with regularization for color transfer are not sensitive to noise,the operation speed is very fast,and have good performance in image restoration,medical image coloring,defog and medical image segment.Finally,it should be pointed out that those algorithms are not only suitable for single image,but also can be used for color transfer between videos.