Abstract: There were inherent speckle noises in the medical ultrasound image. The image denoising was necessary for the lower image quality. The balanced orthogonal multiwavelet had both orthogonal and symmetry. It had higher resolution and better denoising effect than the scalar wavelet. Therefore, the ultrasound medical image was discomposed to multiwavelet domain. Then its coefficients were shrunk by fuzzy clustering with semi-soft threshold for image de-noising. The result indicated that the method was more effective than hard threshold and soft threshold. The speckle noise in the original image was removed efficiently and at the same time the image details were reserved.

Key words: image denoising, balanced orthogonal multiwavelet, fuzzy clustering, semi-soft threshold

摘要： 医学超声图像存在特有的斑点噪声，它大大降低了超声图像的质量，因此必须进行降噪处理。平衡正交多小波同时满足正交性和对称性，具有比单小波分解更加精确、去噪效果更好的特点。故对超声图像进行平衡正交多小波分解，然后利用模糊聚类与半软阈值相结合方法对小波系数进行萎缩处理，实现降噪目的。结果表明该方法优于硬阈值和软阈值方法，可以有效地降低原图像的斑点噪声并保留图像细节。

关键词: 图像降噪, 平衡正交多小波, 模糊聚类, 半软阈值