舌下赘生物轮廓提取方法初步研究

A preliminary research on segmentation of sublingual vegetation

研究了一种从彩色舌下图像中提取舌下赘生物的方法。根据舌下图像的特点,方法中首先对舌面反光区域检测方法进行改进,提出了针对提取舌下赘生物的反光区域强检测方法,并运用反距离权重插值法对检测出的反光区域进行填充;然后应用自适应直方图均衡化,结合多阈值大津法进行粗分割,得到候选赘生物区域;进而,采用几何测度对不符合舌下赘生物特点的区域进行初步筛选;最后,通过观察赘生物隆起会在其临近区域产生阴影的舌质特征划定最终轮廓。实验结果表明,上述方法能有效地分离目标区域,较为准确地刻画舌下赘生物的轮廓边界。

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.