Research On Color Image Graying Algorithm Based On RGB Color Space

Nowadays,although color image is gradually becoming the mainstream,gray image does not die out quietly.On the contrary,in some specific areas,such as black-and-white printing or photography,medical diagnosis,target tracking and pattern recognition technology,it plays a vital role.With the continuous widening of gray image in various fields,all kinds of theories and methods about color to gray transformation acquired more attention.Traditional algorithms only choose color or luminance to map to gray level but do not consider the spatial distribution of color,which have caused the gradient structure disordered and contrast information lost seriously.Therefore,the subject about how to retain the original color images' gradient features and contrast information maximizely in the processing of color to gray really attracted much attention from researchers.Lu's decolorization algorithm is most widely used among those existed algorithms.But Lu's algorithm model only sampled the pairing pixels' color difference of neighborhood leading to not perceive the image's global structure information.In addition,some approximate treatments in transformation steps not only make a lot of time cost in the process of space transformation but also lead to the vast loss of color information.As a result,some serious problems such as color information leakage,images' gradient feature structure disorder and disharmonious edges appear in gray-scale image.Based on these shortcomings,the paper has made some improvements and verified the validity of the algorithm in this paper through simulation experiments.The paper introduces the background and significance of color to gray transformation,analyzes and describes research status at home and abroad.Then it has a brief introduction to some basic knowledge about this subject,and analyzes the existed algorithm and the main deficiencies existing in these algorithms.This paper is supported by the principal component analysis theory and Lu algorithm model which are most popular,and mainly did the job in following some respects:(1)Sampling color contrast difference based on the CIE Lab color space which many colors beyond the range of human vision can receive wavelength,it can only keep partial emotionally important color contrast to a certain extent.What's more,there are tedious steps in the process of converting the image from RGB color space which related to hardware equipment into CIE Lab color space.And it may miss a part of the color information of original color images due to the most of the approximate numbers of replacement.So the paper chose to realize algorithm based on RGB color space.We constructed a color difference model and an objective energy optimization function which directly contains three components of R,G and B of the color image,and got high quality gray images through traversing the candidate set of weight coefficients.(2)Because there is shrunken color difference sampling scope and it is difficult to capture the global color contrast,the paper proposes a pseudo random pairing method to sample color differences of paired pixels so as to construct the mapping function to grayscale in order to obtain gray-scale images with more rich levels,more satisfactory effect.(3)Lu's decolorization model only attached great importance to the contrast retain while ignored the protection of original images' inner gradient features structure,the paper improved the energy optimization function in Lu's model,and then introduced a gray variance constraint item based on the principal component analysis which has overall perception characteristics of different gradient and which can retain the original color image to the utmost extent.In the end,the experiment proves that the method can effectively protect gradient feature structure and global contrast to a large extent,and the method conforms to human's different perception to image contrast change.