Haptic Gesture Analysis and Recognition

Youssef Chahir (1) , François Jouen (2) , Michèle Molina (3) , Bahjat Safadi (4)

1 Equipe Image - Laboratoire GREYC - UMR6072
2 Laboratoire de Psychobiologie du Développement
3 PALM - Psychologie des actions langagières et motrices
4 LIG - Laboratoire d'Informatique de Grenoble

Haptic perception properties is achieved by the execution of specific exploratory procedures "Eps". An "EP" is a stereotyped movement pattern which is dictated by the object properties that the haptic system chooses to process, both perceptually and cognitively. The aim of the present work is to devise an automatic manual testing procedure able to extract tactile information effectively. Our system has been conceived to recognize the texture and the hardness of an object through the video analysis of hand actions. Two objects properties have been tested: texture and consistency. For each property, two modalities were proposed. The texture of the object being explored could either be smooth or granular. Its consistency could either be hard or soft. In this paper, we propose an automatic approach for hand gesture analysis and recognition for understanding human action and manipulation. To enhance robustness, each hand sequence is characterized globally by a volume which is characterized by its 3D geometrical moments. Neural networks are then used using distance between vectors of features. We tested the proposed approach with different hand gestures and results showed that our method is effective, achieving a high recognition rate.

Domaines: Traitement des images [eess.IV]

Youssef Chahir, François Jouen, Michèle Molina, Bahjat Safadi. Haptic Gesture Analysis and Recognition. IEEE/RSJ Intl. Conf. on Intelligent Robots and Systems, workshop on Grasp and Task Learning by Imitation- IROS2008, 2008, Nice, France. pp.65-70. ⟨hal-00823545⟩