中文摘要：

针对口腔CT图像软组织部位分辨率低，细节不清晰的问题，提出一种基于数学形态学的图像处理方法，较好地避免了目前常用的直方图均衡、反锐化掩膜以及小波阈值增强等方法在图像处理过程中产生的细节丢失、噪声放大等缺陷。该方法通过对图像进行小波变换预处理，增强表达软组织信息的低频小波系数并重构图像，并解决了增强信息的同时带来的噪声放大问题。同时引入形态学开闭级联滤波器，增强特征信息，进一步去除噪声平滑图像。实验结果证明了该方法的有效性，有助于临床口腔软组织病灶部位的区分和诊断。

英文摘要：

Aiming at the problem of oral cavity CT image that its soft tissue part has low resolution and is unclear in details, we proposed a mathematical morphology-based image processing algorithm, which well avoids the defects of current common methods, including histogram equalisation, unsharp mask and wavelet threshold enhancement, generated in image processing such as the lost of details and the noise amplification, etc. By carrying out preprocessing of wavelet transform on the image, the algorithm enhances the expression of low frequency part of soft tissue information and reconstructs the image, and solves the noise amplification problem brought forward by information enhancement. At the same time, the morphological opening and closing cascade filter is introduced for feature information strengthening and further noise elimination and images smoothing. Experimental results prove the effectiveness of the algorithm, it is helpful for the clinic differentiation and diagnosis of the lesion of oral soft tissues.