Modeling brightness perception and syntactical image coding

We present a study concerning the realizability of a syntactical image coding scheme for the modeling of brightness perception. This scheme is based on a multiscale Gabor analysis, a symbolic line and edge representation, additional low-pass channels, and nonlinear amplitude transfer functions. It is shown that this scheme, although not optimized, can already predict various brightness effects such as illusions. Although only studied in the 1-D static case, and without going into details concerning quantization effects, we argue that such a brightness model may be advantageous in digital image coding applications.