A preliminary research on segmentation of sublingual vegetation

A method for segmenting the sublingual vegetation from color sublingual images was studied. Firstly, the light reflecting detection process for tongue surface was proposed based on the image feature of the dorsum of tongue, and a strong detection method was specially designed for sublingual vegetation segmentation for light reflecting regions, and the inverse distance weighting interpolation was used to fill the detected regions. Then, the adaptive histogram equalization combined with the multi-threshold Otsu method was applied to coarse segmentation of the image, and the candidate vegetation regions was obtained. Finally, the final contour of sublingual vegetation was extracted by observation of the shadows produced by the protuberance of the sublingual vegetation in its neighborhood. The area overlap measure was employed to measurement of the similarity of the segmentation results and the ground truth, and the experimental results reveal that the proposed method can effectively isolate the target regions and accurately depict the boundary of sublingual vegetation to a great extent.