Traditional tongue diagnosis information is mainly obtained by clinical subjective judgment, lacking objective and quantitative measures. Moreover, complexity of the patient self-state also affects accuracy of the diagnostic results. In this regard, a diagnostic method under the comparative analysis of self-tongue image chromatography is proposed. That is, use image processing techniques to overlay positive films of healthy tongue images and colored negative films of unhealthy tongue images, select the information sensitive area of the tongue images, collect the data, use the conversion relationship between RGB and CIE Lab color model, obtain the discrete chromatography distribution characteristics, and combine the quantitative parameter range to diagnose the health state. Through simulation and experimental analysis, the feasibility and correctness of the method are verified. The method proposed can effectively improve the tongue image diagnosis effect and is useful for the further development of tongue diagnosis in traditional Chinese medicine.