UNSUPERVISED STATISTICAL NEURAL NETWORKS FOR MODEL-BASED OBJECT RECOGNITION

Statistical neural networks executing soft decision algorithms have be en shown to be very effective in many classification problems. A neura l network architecture is developed here that can perform unsupervised joint segmentation and labeling of objects in images. We propose the semi-parametric hierarchical mixture density (HMD) model as a tool for capturing the diversity of real world images and pose the object reco gnition problem as a maximum likelihood (ML) estimation of the HMD par ameters. We apply the expectation-maximization (EM) algorithm for this purpose and utilize ideas and techniques from statistical physics to cast the problem as the minimization of a free energy function. We the n proceed to regularize the solution thus obtained by adding smoothing terms to the objective function. The resulting recursive scheme for e stimating the posterior probabilities of an object's presence in an im age corresponds to an unsupervised feedback neural network architectur e. We present here the results of experiments involving recognition of traffic signs in natural scenes using this technique.