自适应非局域均值滤波的超声图像去噪

Adaptive nonlocal means filter for ultrasound image denoising

目的提出一种去除超声图像噪声的新方法。方法对超声图像进行非局域搜索,找到相似的图像块进行加权平均,降低噪声。通过定义一个特征强度,区分斑点噪声和图像边界;然后将特征强度引入非局域滤波方法中,对平坦区域和边界进行自适应滤波。结果本方法可有效去除斑点噪声,提高噪声图像的峰值信噪比(PSNR)和结构相似度指数(SSIM),优于常规方法。结论自适应非局域均值滤波可有效去噪,并保护超声图像特征。

Objective To present a novel adaptive nonlocal means（NLM） filter for ultrasound image denoising.Methods Nonlocal searching for ultrasound image was performed to find more similar image blocks,which were weighted to remove speckle noise.A new feature strength index to distinguish speckle noise from boundary was defined and introduced into NLM for adaptive filtering on flat area and image boundary.Results The presented method could remove speckle noise effectively and improve peak signal to noise ratio（PSNR） and structural similarity index（SSIM） evaluation index,which were better than conventional ultrasound denoising methods.Conclusion Adaptive nonlocal means filter can remove noise and protect image features of ultrasound.