The Art Illumination Transfer Algorithm Of Portrait

The illumination of art is a comprehensive art,which combines painting,art,photography in a body,is the overall art by multi converging.People can see the world image,because the object surface to reflect light to the human eye and then become a visual image.Not only the light for all things life,but also gives light to the artistic life,the meaning of art as air is to human beings.Without light there is no form of object,without light there is no gorgeous color,without light there is no concept of time and space.The art illumination can determine the atmosphere of the scene,shaping the style of the picture;you can also portray the character image,the performance of the characters.Therefore,this paper chooses the portrait of the face art light and shadow as the research object,and carries on the research of the light and shadow transfer algorithm.Portrait art light transfer mainly divided into three parts:the first part will convert the image color space;the second part,layer decomposition;third part,light transmission.This paper mainly studies the two parts of the layer decomposition and light transmission.First,The RTV(Relative Total Variation,RTV)in the process of smoothing the image,the need to constantly adjust the smoothing parameters in order to achieve the purpose of smoothing,and the parameters of the whole is very cumbersome and time-consuming.On this basis,this paper through the gradient statistics dynamically set the smoothing parameter of RTV model was improved,reducing the tedious process of the original RTV model to adjust the smoothing parameter,at the same time,it can realize the effect of preserving the edge and smoothing the image,and extract the fuzzy layer.Second,in the process of light transmission,the original face structure of the target image always changes to a certain extent,which will affect the final effect.In order to solve this problem,this paper uses guided filtering algorithm based on the similar structure,by looking for the similar structure of the reference image and the target image,In addition,the guidance of similar structure is added in the procedure of guiding filtering,thus the more objective function is obtained,so the structure of the edge information of the target image is well preserved and the purpose of light transmission is realized.Finally,this paper uses four kinds of classical illumination distribution style to carry out the experiment,thus we verify the effectiveness of the algorithm.In addition,we compared with other algorithms and analyzed the advantages and disadvantages of this algorithm.