Leg cycling tracking by dynamic vision.

Abstract

This study describes a method of tracking of human body limbs from a monocular sequence of perspective images. These objects and the associated articulations must be modelled. The principle of the method is based on the interpretation of image features as the three-dimensional perspective projections points of the object model and an iterative process method to compute the model position in accordance with the analysed image. This attitude is filtered (Kalman filter) to predict the model position relative to the next image of the sequence. The image features are extracted locally according to the computed prediction. Tracking experiments, illustrated in this study by a leg cycling sequence, have been conducted to demonstrate the viability of the approach.