Detecting pauses in continuous sign language

Sign language (SL) segmentation breaks a continuous SL sentence into its basic lexical units by detecting word boundaries. In Speech signal, silence period between two words are detected for a reliable speech parsing. Similarly, for robust SL recognition, majority of direct segmentation approaches exploit these inter-sign pauses in a stream of hand gestures to demarcate the word boundaries. The delayed absolute difference (DAD) signature of the hand positions provides means for analyzing the segmentation features like pauses, repetitions and directional variations in unique tool. In this paper, a novel scheme for detecting pause features is presented.