基于小波变换与SRAD融合的医学超声图像斑点噪声抑制
2024,32(12):184-190
摘要：为了提升传统图像去噪算法的准确性和有效性，结合医学图像特点，针对传统各向异性扩散斑点降噪(SRAD)算法对图像边缘和细节信息保持能力不足的问题，提出了基于小波分析的各向异性扩散斑点降噪算法(Wavelet-SRAD)。通过对超声信号进行多尺度分解，在低频信号部分引入SRAD算法进行各向异性多尺度滤波，在高频信号部分采用软阈值收缩算法，并采用梯度算子和拉普拉斯算子区分噪声或边缘引起的灰度变化，以提升图像去噪效果。经Matlab仿真实验，将改进算法与传统中值滤波、高斯滤波、均值滤波和SRAD滤波算法进行对比，计算均值、标准差、斑点指数和等效视数来分析去噪结果，并采用灰度直方图从能量角度对去噪效果进行可视化。实验结果表明，与传统图像去噪算法相比，改进的Wavelet-SRAD算法能够更准确、有效地去除医学超声图像中的斑点噪声，并保持良好的组织纹理和边缘细节信息，表现出优越的滤波去噪性能。因此，改进的Wavelet-SRAD滤波去噪算法是一种有效的医学超声图像斑点噪声抑制算法。
关键词：小波变换;软阈值;各向异性扩散;斑点噪声;降噪
Speckle Noise Suppression of Ultrasonic Medical Image Based on Wavelet Transform and SRAD
Abstract：In order to improve the accuracy and effectiveness of traditional image denoising algorithms, this paper combines the characteristics of medical images and addresses the problem of insufficient edge and detail information preservation capability in the traditional anisotropic diffusion speckle reducing algorithm (SRAD). A wavelet analysis-based anisotropic diffusion speckle reducing algorithm (Wavelet-SRAD) is proposed. Using Matlab for simulation experiments, the proposed algorithm is compared with traditional median filtering, Gaussian filtering, mean filtering, and SRAD filtering algorithms. Mean, standard deviation, speckle index, and equivalent visual number are calculated to analyze the denoising results. The denoising effect is visualized from the energy perspective using a grayscale histogram. The experimental results show that compared with traditional image filtering denoising algorithms, the improved Wavelet-SRAD algorithm can more accurately and effectively remove speckle noise in medical ultrasound images, and has good preservation capability for tissue texture and edge detail information, demonstrating superior filtering and denoising performance. Therefore, the improved Wavelet-SRAD filtering denoising algorithm is an effective method for suppressing speckle noise in medical ultrasound images.
Key words：wavelet transform; soft threshold; anisotropic diffusion; speckle noise; noise reduction
收稿日期：2023-10-10
基金项目：
