Video Sequence Coding by Neuro-Fuzzy Image Segmentation and Polynomial Modeling of Motion Fields