摘要： 在超声医学图像斑点噪声处理过程中，SRAD算法易受噪声梯度的影响。为此，提出一种基于自适应加权的SRAD-高斯金字塔联合优化模型。以最小的迭代次数去除超声图像中的斑点噪声，利用优化的高斯金字塔模型对原始图像重新融合，计算8个不同扩散方向和中心的相似度，并分别赋予相应权重。实验结果表明，在保存图像纹理的同时，该模型能较好地抑制噪声、加快处理速度。

关键词: 自适应加权, 纹理细节, 图像去噪, 噪声梯度, 迭代次数, 图像融合, 相似度

Abstract: This paper proposes an optimization model of Gaussian-SRAD joint model based on adaptive weighting for the problem which has large impact in the gradient of speckle for the normal SRAD model in the ultrasound image speckle processing. It uses minimum number of iterations to reduce the speckle in ultrasound image. After the re-fusion by optimization gaussian pyramid, the proposed model gives the corresponding weight by calculating the similarity between the center point and the remaining point of eight diffusion direction. Experimental results show that the proposed model can inhibit the speckle, accelerate the processing speed and preserve the texture information at the same time.

Key words: adaptive weighting, texture detail, image denoising, gradient of speckle, iteration number, image fusion, similarity

中图分类号: 

• TP391