基于多层小波分解与稳定分布的超声图像散粒噪声抑制新方法

医学超声图像在应用中遇到的一个重要问题是如何消除图像中由于散射现象的相干本质而引起的多径乘性散粒噪声。对数超声图像的二维小波系数服从具有尖峰和拖尾的边缘分布的非高斯分布。α稳定分布可以用来描述这类重拖尾非高斯尖峰脉冲信号和噪声。本研究利用一种散粒噪声模型，通过对对数超声图像的多层小波分解的高频系数的分析与稳定分布建模，提出了一种新的基于闽值的二维小波分解系数的检测分类方法，得到一种基于多层小波分解与稳定分栉模型的超声图像散粒噪声的抑制新方法。仿真结果表明，该方法比传统的基于高斯假设下的阈值去噪方法性能更好。

New Noise-removal Method Based on Multiresolution Wavelet Decomposition and Alpha Stable Model

Uhrasonic image quality is generally affected by muhiplieative speckle noises caused by the coherent nature of the scattering phenomenon. Speckle filtering is thus a critical pre-processing step in medical ultrasound imagery. In this paper, speckle noise model is used to analyze the coefficients of 2-D multiresolution wavelet decomposition of logarithmically transformed images using alpha stable distribution. Based on above analysis, we propose a new function of classifying the coefficients through which a new noise-removal method based on multiresolution wavelet decomposition and alpha stable model is established. The simulation results show that the proposed method is more robusl than the method based on Gaussian assumption.