MODELING FEATURE SIGNALS FOR VISION-BASED HAND POSTURE CLASSIFICATION

摘要

This paper proposes methodology for inference of features of static hand posture images of sign language vocabulary. Image processing, image analysis and signal processing procedures are used to derive set of signals, modeling hand region properties in the image: shape and appearance of distribution of hand region intensity levels. Hand postures are described with the set of curves exibiting distinguishing shape characteristics which makes them usefull for classification purposes.