Auxiliary Illuminant-based Research On Objectivity Of Tongue Diagnosis

Tongue Diagnosis is one of the most important diagnostic methods in Traditional Chinese Medicine(TCM).It mainly depends on the doctor's eyesight and text description.However,the diagnosis results are often influenced by doctor's experience,patients' mood and environmental changes.Therefore,it is difficult to quantify results description,get accurate diagnosis and save tongue images.These problems prevent further improvement and academic communication of tongue diagnosis.Thus,quantitive and objective tongue diagnosis is required.In order to achieve automatic acquisition,segmentation,separation of tongue body and coating,recognition and localization of tongue spots and prickles in tongue images,this paper designed an auxiliary illuminant system in terms of tongue diagnosis needs as well as different illuminant sensitivity of tongue body and coating.This auxiliary illuminant system includes image acquisition system and tongue image analysis system.Specifically,the image acquisition system is composed of a relative standard illuminant,auxiliary illuminant,“camera obscura”,digital camera,infrared distance sensor,Microprogrammed Control Unit(MCU)and computer.This acquisition system can take two pictures automatically: a color image and a monochromatic light image,according to tongue extension level.Besides,based on C and C# programing,this acquisition system can achieve a variety of control functions.For example,MCU can realize AD sampling control of infrared distance sensors,and switching between auxiliary illuminant and relative standard illuminant;the computer can communicate with MCU by serial ports and control digital camera.Moreover,according to demands of tongue diagnosis objectivity,the tongue images analysis system can realize segmentation,separation of tongue body and coating,as well as spots and prickles identification,by image processing and development tool,such as OpenCV.The specific mechanism as the following.First,for tongue images segmentation,this system applies different initial contour extraction methods to proceed Snakes algorithm under relative standard illuminant and auxiliary illuminant.Then,it applies image registration and K-means clustering to separate tongue body from coating and to compare this two image segmentation results.Finally,it adopts image preprocessing,edge detection,contour extraction,the axis determining,and zone dividing to complete tongue spots and pricks identification and zone counting.In sum,through image registration and K-means clustering,the auxiliary illuminant system in this paper can overcome the separation problems of tongue body and coating,the accuracy rate could reach up to 86.7%.What's more,it could achieve extraction of tongue spots and pricks as well as automatic division of tongue zone.Third,the accuracy rate of tongue spots and prickles counting in different zones could reach over 70%,especially on tongue tip and root.