Objectified and quantified indicators in the diagnosis of tooth-marked tongue in traditional Chinese medicine (TCM) are lacking. Thus, an algorithm for identifying and detecting multiple objectified indicators of tooth-marked tongue was proposed to aid the clinical comprehensive detection of tooth-marked tongue in TCM. The regions of interest of the tooth marks were extracted from the obtained tongue image, and a region binarization pretreatment was performed. The Graham scan method was used to extract the characteristics of the tooth marks, and the support vector machine algorithm was used to classify and discriminate the presence or absence of tooth marks. The improved Douglas-Peucker algorithm was used to fit the edge curve of the tooth marks, and the minimum number of curves satisfying the conditions was set as the number of tooth marks. Finally, the degree of tooth marks was determined by combining the number and depth of tooth marks. The algorithm has a correct rate of 80.86% in judging the presence or absence of tooth marks, 80.00% in detecting the number of tooth marks, and 89.63% in detecting the degree of tooth marks. Experimental results show that the proposed algorithm can realize the comprehensive detection of tooth-marked tongue and provide the corresponding objective parameters to assist the clinical diagnosis and treatment of TCM.