Adaptive active contour model based automatic tongue image segmentation

For about 1800 years, tongue inspection has been one of the four major diagnostic methods in Traditional Chinese Medicine (TCM). The tongue is believed to be able to reflect the health status of the human body. However, making an accurate diagnose with the tongue is not a trivial task. It usually requires enormous training on the TCM doctor before he can make a reasonable diagnosis. Recently, image processing methods have been proposed to automatically process the tongue images and make diagnosis. This study proposes a k-means clustering and adaptive active contour model based automatic tongue region segmentation algorithm. This study is the first step towards the automatic tongue diagnosis. The method was applied on a set of real tongue images. To quantitatively evaluate the segmentation results, the automatically extracted boundaries were compared to the tongue boundaries drawn by experts. An average coverage ratio of 92% was found, indicating the accuracy of the proposed algorithm.