Study on Conversion Method of Color Space under a Big Color Gamut

Since the equipment in computer color quantification system has different color gamut and color feature, accurate control and transmission of color information in this system is particularly difficult, and therefore the color luminance meter is selected for marking the color quantification system. Munsell color system is selected to establish the mutual conversion between RGB and L*a*b* color model for camera. The training set includes 1550 samples and the testing set includes 52 samples, which certainly will lead to the redundant problem of the hidden-layer node number, so the two-hidden- layer neural network is considered. The training program, testing program and foreseeable program is compiled respectively by Neural Network Toolbox in Matlab applications. The conversion relation under a big color gamut is expressed by four-layer BP network. Through training this network, the training error is 0.000748566, using the data of testing set to test this network and calculating the color difference between forecast value and true value, the maximum color difference is 5.6357 NBS, the minimum color difference is 0.5311 NBS, and the average color difference is 3.1744 NBS. The result shows that the network can express the color quantitatively, and it is not subjective and vague, the quantitative mensuration and control of color could be done.