Neural network based color decoupling technique for color fringe profilometry

• We propose a new description of the color cross talk phenomenon.
• We use a back propagation neural network to fit the coupling function.
• In simulations, the influence of the cross talk can be reduced dramatically.
• In experiments, both sinusoidal and phase quality can be improved by the proposed decoupling technique.

A color decoupling algorithm based on neural network is proposed to improve the performance of color fringe projection system which suffers from the complicated color coupling effect. We exploit the generalization and interpolation capabilities of a feed-forward backpropagation neural network to map from coupled color data to decoupled color data. The advantage of this algorithm is that the coupling effect is regarded as a nonlinear function which is a rational description of the color CCD camera׳s imaging principle. Theoretical analysis and the neural network training process are introduced comprehensively. Sufficient experiments are carried out to prove that the proposed method can effectively compensate for both the stripe sinusoidal property and the phase quality.