Radial basis function (RBF) neural network was applied to study the conversion of color values between different color space by using the color space conversion of RGB to CIE L*a*b* as example. RBF neural network model of color space conversion was established by using basic data sets of sample. The precision of model conversion was improved by adding sample data and using color subspace dynamic planning method. The result showed that the conversion speed and precision of this method is better than dynamic subspace automatic divided BP neural network color space conversion method.