材料期刊网

针对传统的各向异性扩散算法中扩散系数函数的平滑效果不好,扩散过程中扩散门限 K 的选取依靠经验确定,扩散过程对图像细节保护不足的问题,提出了一种改进的各向异性扩散算法.介绍了几种当前比较典型的各向异性扩散去噪算法；在典型算法分析的基础上提出了一种基于自适应中值滤波的改进扩散模型；根据扩散系数应满足的3个条件及经典的扩散系数函数,提出了改进的扩散模型中的改进扩散系数函数；提出了一种扩散门限 K 的自适应选取的方法.通过在改进的扩散模型中使用改进的扩散系数函数并结合扩散门限 K 的自适应选取,对超声图像进行去噪.实验结果表明,所提算法优于 PM 模型、Catte 模型、王常虹算法等,去噪后图像的 FOM 值比 PM 模型高出3.34％,PSNR 值比PM 模型高出0.2506.该算法在去除散斑噪声的同时有效保护了图像的细节及边缘,有助于医务人员对患病区域的准确诊断.

In the traditional anisotropic diffusion algorithm,for the result of the smoothing effect of diffusion coefficient function is not good,the value of diffusion threshold K relying on the experience and the protection of image details is inadequate in the diffusion process.To solve above problems,an improved anisotropic diffusion algorithm is put forward.First,several current typical anisotropic dif-fusion denoising algorithms are introduced.Then,an improved diffusion model which based on adap-tive median filtering is put forward on the basis of typical algorithm analysis.And then,an improved diffusion coefficient function in improved diffusion model is put forward according to three conditions which the diffusion coefficient function should satisfy and the classical diffusion coefficient function. Finally,a method of adaptive selection of the diffusion threshold K is put forward.The algorithm re-moves noise in ultrasound images by using the improved diffusion coefficient function in the modified diffusion model and combining adaptive selection of the diffusion threshold.The experimental results show that the proposed algorithm outperforms the PM model,Catte model ,the Wang Changhong al-gorithm etc.The FOM value of the denoised image is 3.34% higher than that of the PM model and PSNR value is 0.250 6 higher than the PM model.The algorithm effectively removes speckle noise and at the same time protects image details and edges effectively and it contributes to accurate diag-nose of the diseased area for medical personnel.