基于偏微分方程的低剂量CT投影降噪算法  

DENOISING ALGORITHM FOR LOW-DOSE CT PROJECTION BASED ON PARTIAL DIFFERENTIAL EQUATION

作　　者：唐瑜[1] 崔学英[1] 张权[1] 刘祎[1] 桂志国[1] 

机构地区：[1]中北大学电子测试技术国防重点实验室,山西太原030051

出　　处：《计算机应用与软件》2014年第3期179-182,213,共5页Computer Applications and Software

基　　金：国家自然科学基金项目(61071192;61271357);山西省自然科学基金项目(2009011020-2)

摘　　要：低剂量CT图像由于采用了较低的X射线放射剂量,大大降低了辐射对人体的危害,但由此带来的问题是投影数据受噪声污染严重,从而导致了重建图像质量的降低。针对上述问题,在基于偏微分方程的基础上,提出一种改进的投影域降噪算法。该算法在各向异性扩散方程的基础上,利用局部熵可以有效地反映图像局部特征的特点,来控制扩散的程度。实验结果表明,新的算法在提高重建图像信噪比的同时更好地保持了图像的细节和边缘信息。Because of using lower X-ray dose, the harm of radiation in low-dose CT image on the human body）s greatly reduced. But the problem it brings out is that the projection data are polluted seriously by the noise, thus leads to the reductions in quality of the reconstructed image. In view of the above problems, based on partial differential equations we present an improved projectiond domain denoising algorithm. On the basis of the anisotropic diffusion equation, the algorithm can effectively reflect the feature of local characteristics of the image by using the local entropy to control the degree of diffusion. Experimental results show that the new algorithm better preserves the detail and edge information of the reconstructed image while increasing the SNR of reconstructed image.

关 键 词：低剂量CT 偏微分方程 局部熵 投影域 图像降噪 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]