Hand-gesture recognition: Comparative study of global, semi-local and local approaches

Asepsis preservation in operating rooms is essential for limiting patient infection by hospital-acquired diseases. For health reasons, surgeons may not be directly in contact with sterile equipment surrounding them, and must instead rely on assistants to interact with these in their place. Gesture-based Human-Computer Interfaces constitute an interesting option for allowing surgeons to control such equipment without breaking asepsis-preservation rules. In this paper, we describe the results obtained by a preparatory study to such a system. Methods representative of the three main approaches in object description (i.e. local, semi-local and global approaches) are compared in terms of hand posture recognition performance when picture backgrounds are not removed. A picture database presenting variations in the postures' angles and illumination, as well as plain and altered backgrounds was created to this end. We study the effects of incomplete training databases, user-dependency and altered backgrounds on recognition rates. Obtained results give semi-local approaches such as Dalai et al.'s Histograms of Oriented Gradients more promising than local or global ones. However, performance degradation when considering non-uniform picture backgrounds makes pre-treatments aiming at segmenting the hand from the background appear to be necessary.