CLASSIFICATION AND SEGMENTATION OF VECTOR FLOW-FIELDS USING A NEURAL-NETWORK

The main peal of this paper is to describe a neural algorithm for clas sification and segmentation of vector flow fields. We propose to use t he coefficients of their projection into an appropriate linear space a s a feature vector for classification, The projection onto a suitable set of basis vectors is computed by satisfying global optimization cri teria. Once the whole flow held is partitioned into a large number of small patches, two processes are performed. In the former, each small patch is classified using the associated projection coefficients estim ated by using a least-square-error (LSE) technique implemented on a ne ural network. In the latter, segmentation into larger homogeneous regi ons is performed using a region growing method. Two application contex ts are considered: analysis of oriented textures and 3D motion. The Li e group theory is used to identify the basis vectors suitable for defi ning the vector space describing the patterns of interest. In particul ar, the pi-ejection onto the image plane of the 3D infinitesimal gener ators of the 3D Euclidean group have proved to provide an effective de scription for the considered vector flow fields.