Detecting Diabetes Mellitus from Tongue Image Hybrid Features and Neural Netowrk Classifier

Metabolic disease like Diabetes mellitus (DM) can be diagnosed using tongue image processing. The abnormalities of tongue can be diagnosed by pre-processing stage feature extraction stage and post processing stage. The pre-processing stage denoise the image and crop the particular part. The diagnostic features DM can be extracted as color, texture and shape. For color feature we will go for color histogram and for texture feature we will go for DRLBP pattern. Histogram features and texture features are calculated from HSV image. Thirteen different features like measurement of distances, areas, perimeters, eccentricity, major axis length, minor axis length etc. Neural network classifier was used to differentiate images with DM symptoms