Optical Flow Enables Hand Tracking With EyeGlove Low‐Cost Cameras in Confined Environments
Year: 2026
Type: article
Abstract: In order to mitigate the challenges associated with exploration and inspection within confined environments, a cost‐effective (<£150) hand‐wearable system (i.e., EyeGlove) is proposed to help operators track their hand movements when performing camera‐based inspection in confined environments by leveraging the inherent dexterity of the human hands as manipulators. The EyeGlove system integrates two low‐cost cameras and two sets of contact pads (working with magnets), thereby creating a stereo camera system characterized by disjointed camera configurations. When wearing the EyeGlove system, operators harness the manipulation capabilities of their hands to pose the cameras, enabling real‐time in situ visual inspection within confined spaces. While the construction of the measurement function has been reported, we are now enhancing the facility by enabling hand tracking for in situ maintenance and service. This provides a comprehensive solution encompassing inspection, measurement, and tracking capabilities, setting the EyeGlove system apart from other systems. Building upon the unique design of the EyeGlove system, which features disjointed camera configurations, this paper proposes a novel hand tracking method tailored to the low‐cost EyeGlove camera, based on sparse optical flow techniques, i.e., the Lucas–Kanade method. The proposed tracking method utilizes optical flow‐based keypoint/feature match techniques in both stereo match and frame match to achieve robust hand movement tracking, which is specifically customized for the low‐cost cameras of the EyeGlove system. Finally, validation experiments have been conducted to demonstrate the tracking performance of the EyeGlove system in confined environments with various lighting conditions.
Source: Advanced Intelligent Systems
Authors: Erhui Sun, Xin Dong, Dragos Axinte, Andy Norton
Institutions: University of Nottingham, Rolls-Royce (United Kingdom)
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $3,970
Funder: UK Research and Innovation
Awards: G2373528398, G2922548090
OpenAlex ID: W7147326978
