PROBABILISTIC VISUAL LEARNING FOR OBJECT REPRESENTATION

We present an unsupervised technique for visual learning, which is bas ed on density estimation in high-dimensional spaces using an eigenspac e decomposition. Two types of density estimates are derived for modeli ng the training data: a multivariate Gaussian (for unimodal distributi ons) and a Mixture-of-Gaussians model (for multimodal distributions). These probability densities are then used to formulate a maximum-likel ihood estimation framework for visual search and target detection for automatic object recognition and coding. Our learning technique is app lied to the probabilistic visual modeling, detection, recognition, and coding of human faces and nonrigid objects, such as hands.