主成分分析先验约束总变分正则化CT图像重建方法  

Total variation regularization CT image reconstruction method based on pincipal component analysis prior constraints

作　　者：刘立[1] 张惠慧[1] 王建[1] 

机构地区：[1]天津大学电子信息工程学院,天津300072

出　　处：《计算机应用》2013年第A02期187-189,共3页journal of Computer Applications

基　　金：国家自然科学基金资助项目(61002030)

摘　　要：总变分正则化计算机断层扫描(CT)图像重建方法适合于有限角度图像重建问题,虽然它可以克服重建问题的不适定性,但在控制迭代次数的情况下,重建图像中的边缘区域出现模糊趋势且恢复的细节不完整。为了重建出拥有更好边缘的图像,提出了一种主成分分析(PCA)先验约束总变分正则化CT图像重建方法,并给出数学模型。该方法运用PCA进行CT图像集特征提取,将特征转化为距离约束作为先验知识加入图像重建过程。实验结果表明,在迭代次数相同的情况下,该方法重建的图像噪声更小、边缘更清晰。Total variation regularization CT image reconstruction method was suitable for limited angle image reconstruction problem, it can overcome the ill-posedness of reconstruction, but in the controlled iterations cases, the edge of reconstructed image tends to blur apparently, and the details are incomplete. In order to reconstruct an image with better edge, a total variation regularization CT image reconstruction method based on Principal Component Analysis （PCA） prior constraints was proposed, and the mathematical model was given. The feature extraction of a CT image sets used PCA algorithm in this method, then the features could be transformed into distance constraints as prior knowledge during image reconstruction. Experiments show that, in the case of the same number of iterations, the reconstructed image noise is smaller, and the edge is sharper.

关 键 词：图像重建 总变分 主成分分析 正则化 凸集投影 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]