For digital color devices, complex relationship exists between their input image pixel values and the output color chromatic values, gamut mapping between them as well. The linear models are difficult to describe this relationship effectively. In the study, based on BP neural network, a gamut mapping model with micro space sections is proposed. The gamut mapping model maps color from monitor gamut to printer gamut. Compared with mature gamut mapping method CARISMA, the experiment shows that, for the perfect nonlinear performance and adaptive ability of BP neural network, the gamut mapping model based on BP neural network decreases the color error between the monitor and the printer significantly and reaches high accuracy.