Computer color matching of powder paints using neural networks

A recipe prediction model for color matching in powder paints production based on the BP neural networks is presented. The mapping between the colorimetric values and the recipe values in the reflective powder paints samples can be set up by the BP neural networks. The color matching experiments for typical powder paints are conducted by using such a model. The experimental results show that the mapping between the colorimetric space and the recipe space can be realized by the multi-layer BP neural networks, and the average prediction error for 64 training samples is less than 1 unit of CIELAB color difference.