DISCRETE, SPATIOTEMPORAL, WAVELET MULTIRESOLUTION ANALYSIS METHOD FORCOMPUTING OPTICAL-FLOW

A wavelet-based system for computing localized velocity fields associa ted with time-sequential imagery is described. The approach combines t he mathematical rigor of the multiresolution wavelet analysis with wel l-known spatiotemporal frequency flow computation principles. The foun dation of the approach consists of a unique, nonhomogeneous multiresol ution wavelet filter bank designed to extract moving objects in a 3-D image sequence based on their location, size, and speed. The filter ba nk is generated by an unconventional 3-D subband coding scheme that ge nerates 20 orientation-tuned filters at each spatial and temporal reso lution. The frequency responses of the wavelet filter bank are combine d using a least-squares method to assign a velocity vector to each spa tial location in an image sequence. Several examples are provided to d emonstrate the flow computation abilities of the wavelet vector motion sensor.