Lip Image Feature Extraction Utilizing Snake’s Control Points for Lip Reading Applications

Abstract

Snake is an active contour model that catches and locks image edges, then localizes them accurately. The simplest Snake consists of a set of control points that are connected by straight lines to form a closed loop. This paper discusses the application of Snake to find the visual feature of lip shapes. In most previous papers, visual feature of lip shapes is represented by Snake’s contour. In this paper, the feature of lip shapes is represented by six control points on lip Snake’s contours. By simply utilizing six control points representing one lip Snake’s contour, it is expected to reduce the burden on pattern recognition stage. To demonstrate the performance of this method, some analysis has been conducted on the effect of lip conditions and illumination. The results shows that the overall lip feature extraction using the proposed method is better for lips that have more contrast to the surrounding skin, optimum room illumination that gives the best result is in the range of 330-340 lux.