最小交叉熵图像重建算法

Image reconstruction based on minimum cross-entropy

CT技术通过扫描和图像重建算法,获取被检物场断层图像。由于具有非侵入性、可视化等特点,该技术在工业领域获得广泛应用。为了提高CT系统重建图像的分辨率,提出一种信息扩充策略,并以此为基础采用两种最小交叉熵算法——MAP和SMART,对多相流CT系统进行图像重建。与传统ART算法相比,最小交叉熵算法有效提高了重建图像的分辨率,减少重构图像伪影。仿真和实验结果表明,基于信息扩充的SMART算法不仅改进了重建图像质量,而且提高了实时性。

During past few decades, CT was widely used in industrial area because of its characteristics of non-invasiveness and visibility. Using the principle of radiation attenuation measurement along different directions through the investigated object with special reconstruction algorithm, cross-sectional information of the scanned object can be obtained. Recently, CT technology is applied to multi-phase flow measurement to detect flow regimes. In order to improve the resolution of CT image reconstructed by conventional algorithms, a novel information extension method is proposed. Two minimum cross entropy methods, MAP and SMART, are presented for the image reconstruction of multi-phase flow CT system in this paper. The selection of weighting parameters and prior information is discussed. Both simulation and experiment results show that the SMART method with information extension not only improves the quality of reconstructed image, but also enhances the real time performance.