Development of tongue diagnosis algorithms using image processing technique and integration with raspberry Pi

Tongue inspection for diagnosis purposes in Traditional Chinese Medicine (TCM) has been widely studied in recent years to solve limitations while increasing application value and practicality. The automatic tongue acquisition system is one of the recently introduced systems to assist medical practitioners in diagnosis. However, most systems are not equipped with intelligent automatic diagnosis systems. In this research, the digital tongue diagnosis system consists of image segmentation using Brightness Conformable Multiplier (BCM), tongue's coating and substance recognition analysis using HSV, and color classification using Support Vector Machine (SVM) with K-means Clustering techniques are embedded on Raspberry-pi to perform an automated diagnosis and classify tongue images into a different group that associated with other imbalance condition or disease. Segmentation algorithm used threshold brightness of Hue, Saturation and Value (HSV) colour space that the upper threshold brightness recognized as tongue body will automatically be adjusted by BCM depending on the condition. At the same time, k-rneans clustering is used to cluster tongue images into four clusters known as background, transitional pixel, deep red pixel and red/light red pixel. Red/light red pixel clusters proceed with SVM to classify between red pixel and light red pixel using the maximum color distance identifier. The segmentation algorithm produced a result that distinguished the tongue body from the perioral area, while the color classification algorithm clustered four clusters that then classify deep red, red and light red tongue colors.