A Research Of Image Recoloring Based On Emotion-palette Combinations

Image recoloring technology plays a very important role in computer image processing and computer graphics,and is widely used in various engineering fields such as literature,film science,and aesthetics.In recent years,a variety of image recoloring algorithms have been derived from different perspectives.Among them,the image recoloring algorithm based on the pixel palette has a relatively simple editing mode and an intuitive interactive interface,which has attracted the attention of many computer graphics and digital image processing researchers.However,the traditional palette-based irmage recoloring algorithm requires troublesome pixel-level modification,which is extremely inconvenient and easily lead to wrong operation.In order to solve this problem,this paper propose an image recoloring algorithm based on emotion-palette combinations,using a predefined combination of palettes associated with a given mood to quickly modify the original image style.For an input image,the optimized k-means algorithm is used to create a set of the most representative color palettes.Then,system search for the best color combinations from a predefined database according to the color palette characteristics of the input image.Finally,using the selected emotion-palette we conduct an image recoloring.Using the palette-based photo recoloring and acceleration method can reduce the time to complete image recoloring,we use Lab color space to carry on the color transfer to the image and handle the color transfer problem in both brightness and color spaces,thus the color correctness and image quality can be guaranteed after recoloring.For the one-to-one correspondence between color and emotional words,we quantify the color space and use the Gaussian Mixture Model(GMM)to train a reasonable mapping relationship between emotional semantics and color combinations,making the calculated emotional palette is more closely related to the emotional state described by natural language.For the color and content characteristics of different images,the Euclidean distance measurement function is used to select the emotion-palette that is suitable for the image.The experimental results show that our system can meet users' various recolor goals for the image.