Color Editing Technology Based On Interactive Analysis

With the development of color imaging equipment,color image is widely used in people's lives,and the research of itis gaining more and more attention.The captured images in daily life usuallyare unsatisfactory,due to the restrictions of shooting equipment,environment and other factors,such as the color of the image captured in cloudy day is not as bright as in sunny day,and the color matching in the image does not meet the requirements.To this end,a variety of color editing technology is presented.O ne of them based on the palette is concerned because of its low input requirements and easy operation.To simplify user interaction and maintain image details,this paper proposes a new image color editing technique.The contributions of this paper are as follows.1? Providing a different palette that have the same numbers and colors as visual image,which ensures that the main colors exist and are unique in the palette.Since the main colors cannot be obtained directly from once clustering,especially the large proportion of the different main colors in the image,a double clustering method with Lab and HSV color spaces is used to solve the problem.First,a k-means algorithm based on Lab feature is used to cluster the image color into a large number of classes(16 is used in the paper),and all the main colors appear in this clustering,which will weaken the proportion of different main colors while bringing the problem that a main color usually corresponds to multiple classes,so a double clustering is needed.The number of class in second clustering is determined by user.In the experiment,we observed that the hue in HSV space between visually high-similar colors is close,therefore we convert the color of the first clustering to the HSV space,and then get the final expected main color by hue-based clustering.2? Providing a detail keeping scheme.When spread thedominant color editor to the whole images by RBF interpolation,during the diffusion,only color feature is considered without location feature and other factors,some details of the ima ge is still not well maintained,to solve this problem,this paper propose a method for image processing based on rolling guide filtering technology to optimize the incremental editing.The basic idea is to keep the details of the image by maintaining the gradient of the pixels,while the color increment of adjacent pixels is the same,the gradient is unchanged before and aft er editing.Smoothingthe image by using the rolling guide filtering technique,so that the pixels with similar color and location should have the same color increment as possible,soasto preserve the image details.The experimental results show that this method is not only easy to operate,but also has good image quality as well as wide applicability.