Color Image And Video Decolorization Based On Perceptual Contrast

Color image decolorization technology is a very important technology.On the one hand,grayscale image for newspapers,books and papers and other printing industry with its low cost to occupy most of the market,on the other hand in the art of cinematography,it also has an indispensable position because of the unique performance capabilities.More important is that color image decolorization is base technology for other image processing,such as edge detection,face recognition and advanced image processing technology often use decolorization as a pretreatment method.The traditional grayscale method is very simple,just by extracting a weighted color image RGB channels,or luminance channel and to obtain the final brightness,but this method is a fixed mapping mode for any color image,which resulting in the same color converted into the same gray value.Because gray value in color space relative to the space 256 gray values has a huge number of color values in a large number of gray values bound to the same color value,so the traditional grayscale method,although the fastest,but also the lowest quality relative.In recent years,decoloration has gradually developed from the method with grayscale image increased from the details in the process of graying on the basis of traditional local methods to ensure the progressive development of the whole figure of the same color conversion to the same global gray value method.However,these methods have distinct disadvantages,localization method for enhancing grayscale detail by highpass of original image,but may make grayscale apparent inconsistency defects.Although the global methods guarantee that same color will be mapped to the same gray value,but the details of the existing global method also can lose details in the optimization process.The decolorization method based on region perceptual contrast proposed in this paper are mainly divided into two parts,color to gray mapping function with parameters and region perceptual contrast optimization objective function.Mapping function with parameters proposed in this paper is divided into two parts,one is the main part of the brightness,calculated by the color information for each pixel in the color image,the other part is the neighborhood of the color matrix on two chrominance channels,the matrix contains chroma influence from the surrounding pixels affecting grayscale pixels.Therefore,the same color of the pixel may be due to different color information of the surrounding pixels and lead to different final gray value,not only in line with the characteristics of the human visual system,but also allows us to have a more flexible mapping function method than the global map way.This mapping function also means that our approach has totally different solution space with the global decolorization methods,which is the back to get the best candidate grayscale image.With different parameters,we can have a series of candidate grayscale by the mapping function,then use a significant degree by the regional model get grayscale visual consistency between the original color image.The proposed method use a perceptual contrast model based on the human visual system tends to see similar pixels in the color space similar to as a whole.The proposed method firstly use superpixel method to get the segmentation of the color image,then use the perceptual contrast model to evaluate the perceptual consistence between the grayscale image and the original color image.After that we can establish the objective function,and solving the optimal parameters and then get the final grayscale.In this paper,we have further more discussion about color video decolorization based on the color image decolorization problems.Unlike color images decolorization,we also need to consider the consistency in the time domain on the basis of grayscale images.On this issue,this paper analyzed and established the corresponding mathematical model of the problem in detail,and the use of optical flow technology gives specific video graying of extension methods.