A robust algorithm for hand-forearm segmentation

Abstract

Despite lots of work in gesture recognition, very little work is done in hand-forearm segmentation. Most of the previous work have overlooked this step by imposing restriction on users. This work presents a novel algorithm for hand-forearm segmentation. The proposed algorithm is inspired and based on the observation of human hand anatomy. Experiments are performed with RGB-D sensor to show the robustness of our algorithm. Additionally, we have compared the performance of our algorithm with the state of the art methods on HGR1 database. The comparative result shows that the presented method performs better than the other state of the art methods. The proposed algorithm is invariant to rotation, scale & translation changes. The advantage of our approach is that users do not require to wear any band/full sleeve shirt. © 2018 Association for Computing Machinery.