A Survey of the Constraints Encountered in Dynamic Vision-Based Sign Language Hand Gesture Recognition

Abstract

Vision-based hand gesture recognition has received attention in the recent past and much research is being conducted on the topic. However, achieving a robust real time vision-based sign language hand gesture recognition system is still a challenge, because of various limitations (The term limitation in this study is used interchangeably to mean constraint or challenge in respect to the problems that can or are encountered in the process of implementing a vision-based hand gesture recognition system.). These limitations include multiple context and interpretations of gestures as well as well as complex non-rigid characteristics of the hand. This paper exposes the constraints encountered in the image acquisition via camera, image segmentation and tacking, feature extraction and gesture classification phase of vision-based sign language hand gesture recognition. It also highlights the various algorithms that have been used to address the problems. This paper will be useful to new as well as experienced researchers in this field. The paper is envisaged to act as a reference point for new researchers in vision-based hand gesture recognition in the journey towards achieving a robust system that is able to recognize full sign language.