改进的基于各向异性扩散的B超图像降噪方法

散斑噪声是超声成像机制引起的固有形态,它对超声图像质量以及医学诊断的准确性有很大的影响。针对传统Perona-Malik(PM)各向异性扩散模型通过四方向扩散无法提供充分的信息且处理效果不明显的问题,提出八方向选择式扩散模型；针对传统算法对强噪声则失效、扩散门限参数K依据经验选取的不足以及迭代终止条件不明确,提出改进的扩散系数计算方法,应用自适应选取的扩散门限参数K,且提出适合改进算法适用的迭代终止准则。经过多组仿真实验,证明本文算法较传统模型可以更好的控制扩散过程,并且提高了算法鲁棒性和效率,综合峰值信噪比(PSNR)、边缘保持度(FOM)等指标,表明该算法相比同类算法有更好的降噪和边缘保持效果。

Speckle Denoising for Medical B-scan Ultrasound ImageBased on Anisotropic Diffusion

Speckle noise is caused by the natural form of ultrasound imaging system, and it has a great influence on the quality of ultrasound image and the accuracy of medical diagnostic. For traditional Perona-Malik (PM) by four directions adequate information and and the treatment effect is not obvious, The paper puts forward an eight selected model based on anisotropic diffusion of ultrasound image. For Perona-Malik(PM) anisotropic diffusion model experienced failure when dealing with strong noise, choosing parameter K of diffusion threshold relies on experience and doing not make sure the number of iteration termination, the paper improves the calculation method of diffusion coefficient, uses adaptive diffusion threshold and proposes the suitable iteration termination criterion. Compared with other similar methods, It not noly can controll the diffusion process compared to the conventional diffusion process, and improves robustness and efficiency of the algorithm. experiments show that the proposed algorithm has betterperformance for noise removal (PSNR) and edge-preservation (FOM).