Tongueprint images of healthy populations can be differentiated from those of unhealthy populations by features of tongueprints (tongueprint is fissile texture on the tongue) according to observation by naked eyes, classification, and statistical analysis on tongueprints for large tongue images of healthy and unhealthy populations. Tongueprint binary image is gotten by an existed method, and then tongue images are classified as no-tongueprint and tongueprint images by our approaches. In terms of obtained statistical results on tongueprints, a series of computerized methods, for example computing length ratio of long to short axis for optimal fitting ellipse of tongueprint binary regions, getting location and amount of extremity and cross point of tongueprint skeletons by computing pixel connective numeral to determine pixel type, straight line segment approach, support vector machine (SVM) classifier and so on, are employed to recognize tongueprint images of healthy and unhealthy populations. Applying our method to the large database of tongue images, we achieve promising experimental results.