Human limb tracking using simulated annealing for a low-cost rehabilitation system

Abstract

In this work, an algorithm for human limb movement tracking is presented. This proposal considers the problem as an optimisation problem and attempts to solve it using an implementation of simulated annealing aided by detected skin features. The parameters of the presented algorithm are set using a general-purpose framework for automatic parameter selection. Application on small video sequences shows good results with selected parameters. Also, some drawbacks of the methods to be considered in practical use are discussed. This technique has been designed for use in the development of a real-time interaction interface in video games for rehabilitation of patients with motor disabilities, and as an offline pose analysis framework to evaluate individual progress in the therapy of those patients.