Vision-based human grasp reconstruction inspired by hand postural synergies. (August 2018)

• Record Type: Journal Article
• Authors: Chattaraj, Ritwik; Khan, Siladitya; Roy, Deepon Ghose; Bepari, Bikash; Bhaumik, Subhasis
• Is Part Of: Computers & electrical engineering. Volume 70(2018)
• Page Start: 702
• Page End: 721
• Publication Date: 2018-08
• Subjects: Human hand -- Dimensionality reduction -- Principal component analysis -- Isomap -- Postural synergies -- Grasp reconstruction; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2017.10.018
• Languages: English
• ISSNs: 0045-7906

Abstract

Highlights: Human grasp reconstruction exploiting kinematic coordinations between fingers. Resolving the Location Determination Problem apparent in hand motion capture in the light of RGB + D sensors. Computation of grasp action subspace using dimensional reduction techniques. Optimal feature point selection to balance reconstruction efficacy and computational burdens. Synergistic vision based grasp pose recovery. Abstract: The human hand exhibits enormous versatility and dexterity, to a degree paralleled by few gripping assemblies. Although several hand posture animators have been discovered, vision-based trackers have retained the research focus owing to their compactness, cost-effectiveness and ease-of-installation. The present investigation explores a marker-based hand pose-tracking solution, using a Kinect depth-capture device. It exploits the inherent synergism within the finger linkages through a novel motion capture algorithm for grasp reclamation. The tracked data-set is analysed for an optimal number of condensed primitives which yielded an effectively reclaimed grasp-pose. Isomap based dimensional reduction followed by Principal Component Analysis (PCA) back-projection, drives the reconstruction of thirty-three Feix-grasps solely from Index, Thumb and three edges of a Palm marker. The derivatives were observed to contribute across grasp initiation to final posture assumption, with a scope for future investigations on their direct correlations to cortical motor impulses.