Enhancing dynamic hand gesture recognition through the HMM-LRB/LR algorithm

Human-computer interaction (HCI) encompasses various interaction forms, with gestures playing a vital role in expressing meaning and thoughts. Gesture-based interfaces are widely used in smartphones, televisions, video games, and other applications. Hand gesture recognition has become a promising technique within HCI, offering a natural, nonverbal communication mode that can complement or replace verbal speech and writing. In addition, hand gestures are crucial in Augmented Reality (AR), Virtual Reality (VR), and gaming. This article introduces a novel dynamic hand gesture recognition model, the HMM-LRB/LR (Hidden Markov Model Left-Right Banded/Left-Right) approach, designed to enhance recognition accuracy using input data from the Leap Motion Controller. The proposed algorithms aim to reduce errors and improve system performance by accessing predefined parameters in the learning data. Accurate real-time results are achieved by appropriately approximating the commonly used DHG 14/28 dataset with the presented topologies. The proposed model achieves an impressive recognition accuracy of approximately 97%, significantly outperforming existing methods. Additionally, the model demonstrates high efficiency and is capable of processing gesture data in real-time scenarios. This study addresses significant shortcomings in existing methods, providing detailed descriptions of the models and their parameters, and includes a comprehensive experimental analysis. Key factors such as computational cost, flexibility, and recognition accuracy are critical to successfully integrating hand gestures into these fields. This research demonstrates significant advances in dynamic hand gesture recognition, highlighting the strengths and potential applications of the HMM-LRB/LR model.