Algoirthm Research On Automatic Tongue Image Segmentation In Traditional Chinese Medicine

Object: The segmentation of the tongue, there are many segmentationmethods such as grayscale threshold method, region growing, edge detection,watershed algorithms and so on.Each has their own advantages anddisadvantages. Gray level threshold images for more category segmentationis difficult; the region growing method automatically selects the seed is moredifficult;edge detection method of color pixel, and the watershed algorithmfor segmentation, This paper aims to improve the traditional segmentationmethods and the successful introduction of the level set method in imagesegmentation of the tongue body, trying to find a suitable algorithm forautomatic segmentation of the Tongue body image, for clinical diagnosis andTongue ofmodernization to provide the necessary reference.Method: Fully understanding the traditional segmentation methods,particularly the watershed algorithm, the method of active contour and levelset method base on the paper using Matlab software after pretreatment of thetongue body image using the active contour method, watershed algorithm andlevel set method.At last this paper proposes a new combination of methodsabout a watershed deal with good results using the level set method, and theformation of a user interface to a lot of tongue image segmentation purpose.Result: Theoretical results: the proposed algorithm, the accuracy ishigher than watershed algorithm, the active contour method and the level setmethod, but the computing time is longer; The experimental results: founda better algorithm, consistent with the theoretical results to achieve aautomatic segmentation of the tongue, and with good usability, and laid asolid foundation for further work of tongue diagnosis.Conclusion: Of this paper to get a comprehensive image segmentation methodof a watershed and level set method is the combination of the tongue, in largemeasure to solve the speed and accuracy of the original algorithm, making thesegmentation of the tongue with very high accuracyand good robustness.