NEURAL-NET DESIGN OF MACRO GABOR WAVELET FILTERS FOR DISTORTION-INVARIANT OBJECT DETECTION IN CLUTTER

We consider the detection of multiple classes of objects in clutter wi th 3-D object distortions and contrast differences present. We use a c orrelator because shift invariance is necessary to locate and recogniz e one object whose position is not known and to handle multiple object s in the same scene. The detection filter used is a linear combination of the real part of different Gabor filters, which we refer to as a m acro Gabor filter (MGF). A new analysis of the parameters for the init ial set of Gabor functions in the MGF is given, and a new neural netwo rk algorithm to refine these initial filter parameters and to determin e the combination coefficients to produce the final MGF detection filt er is detailed. Initial detection results are given. Use of this gener al neural network technique to design correlation filters for other ap plications seems very attractive.