Hand Gesture Trajectory Estimation Using Keypoints Combination of Brisk and Minimum Eigenvalue Techniques for Human Computer Interaction Applications

Abstract

In last years, hand gesture trajectory tracking has gained the interest a sizable body of researchers. However, in 2D vision based approaches, hand gesture trajectory estimation can be a significant challenging issue, when it comes to locate hand position in the total scene, In particular when hand practices non-linear motion, scale changes, rotation, translation and postures variation under noisy environment and different lighting conditions. In such challenges, most hand tracking techniques degrades to estimate the accurate position of hand. Hence, to increase the accuracy of moving hand position estimation, this paper proposes a method uses corner keypoints of BRISK and minimum eigenvalue techniques extracted from last segmented region of hand to create searching windows and estimate hand region position on current frame image of video sequences. The experimental outcomes revealed that the proposed algorithm can accomplish correct approximation to hand position.

A Decade of Progress in Human Motion Recognition: A Comprehensive Survey From 2010 to 2020

“…The study's contribution is using HMM for gesture recognition, which could improve recognition accuracy in dynamic gesture recognition scenarios. Xu et al [73] presented an online dynamic gesture recognition system for HRI. The system extracts depth and motion features and uses an incremental online learning algorithm for recognition.…”