Model-reference neural color correction for HDTV systems based on fuzzy information criteria

The authors present a new adaptive color correction process for a high definition television (HDTV) system based on the human color perception model. A cost-effective indirect adaptive control scheme consisting of both a backpropagation neural net controller and a forward model is proposed to overcome the difficulty of dealing with both nonlinearity and model-identification of HDTV systems. The forward model is used to convert the plant output error (color difference) into control signal error for training the backpropagation controller. The values of derivative-relative parameters of the forward model are determined by Hornik´s multilayer neural net identification process. The measure of the color difference (plant output error) in human color perception space should be quantified by fuzzy information to perform the system control. The effectiveness of the method is being verified by a number of experiments