Research On Uniform Color Space Based On Fuzzy BP Neural Network

This paper researches on the uniform color space .A uniform color space means where there are the same differences in vision, there should have the same distances in the space. According to the human's analysis of physiological frame of retina and the course of color vision info process and on the basis of Munsell renovation data, this paper establishes the mixed Neural Network, which is formed a new uniform color space by the trained samples. The FBP Neural Network that is made utilizes the theory of fuzzy competition to simulate the high blur during the course of color formed. At the same time, it uses BP Neural Network to simulate the theory of tri-color apperception and four-color transmit. At last, this article gains a compact and effective color differences formula. By the method of color differences comparison, the new color space, CIE1976Lab and CIE1994 color differences formula are compared simultaneously. The result is that the uniformity of FBP color space is better than the rest. On the one hand, the new establishment of FBP color space further promotes the uniformity of surface color system. On the other hand, it offers a direction for the uniformity of color space.