Tongue segmentation using active contour model

Abstract

Tongue is an organ of the human body for tasting sense. Healthy conditions can be known from observation of the surface tongue by an expert. Before analyzing the tongue, feature extraction process is needed to segment the tongue from image, so it is possible to develop an application that can segment the tongue image from opened mouth image. This research uses Canny Edge Detection and Active Contour method. Canny Edge Detection is used to find the edges of tongue. This method has four steps: Smoothing Gaussian Filter, Finding Gradients, Non-maximum Suppression, and Hysteresis Thresholding. After finding the tongue edge, Active Contour Model will be generating energy that can pull into edges curve that is already defined and cropping that to produce tongue image. Testing result of this research yield an accuracy rate of 75%, by which from all 40 tongue images, 30 are successfully segmented.