Research On Color-to-gray Conversion Based On Normalized Cross Correlation

Digital media is closely related to our lives,and the images have a wide range of applications as a basic display of digital media.Although the color images are the most primary and important information carriers,the grayscale images are not forgotten or ignored yet.They have the important applications and research values in many fields,such as medical research and pattern recognition.The color-to-gray conversion is a process of reducing dimensions.Since the information loss is inevitable,so it is necessary to retain the feature information of color images into the grayscale image as much as possible.It is also necessary to keep the information matching between beforeand after-conversion possibly,such as brightness,contrast,structure,etc.The image matching is not only an important target for the conversion,but also an important evaluation content for the conversion quality.In this work,a linear projection global mapping method based on the image matching is introduced.Actually,the problem of finding the optimal linear projection direction,in essence,is finding the maximum value of images matching between before-and after-conversion and retaining the maximum structural information of color image at the same time.The grayscale method proposed in this paper introduce the popular Normalized Cross Correlation(NCC)algorithm as a criterion for the image matching.The main idea is obtaining the candidate grayscale images by discretizing the linear projection weight of RGB image,and then taking the candidate image with the largest NCC coefficient for color image as the final grayscale result.The experiment demonstrates that the method proposed in this paper could achieve very good objective and subjective evaluation compared with several newer researches.After several optimization methods are tried,there are many effective optimizational methods as follows: the accumulation of absolute value of similarity coefficient,the expansion of search space,the dynamic determination for cumulative weight of NCC coefficient.The final comprehensive optimization grayscale method can effectively enhance the subjective visual effect of the grayscale results,and can also keep a high objective evaluation.