Abstract:

The measurement of radiative properties for disperse media has a broad application in high temperature flame combustion diagnosis,nondestructive testing,biomedical imaging and other related fields. Traditional direct measurement methods are mostly to measure the equivalent properties of the specimen,while precise measurements usually need to be obtained by inversion method. Transport-theroy-based reconstruction algorithm can effectively overcome this problem,however,it has a high computational burden and often slowly converging. Therefore,methods that have high accurate and low computation burden are desirable. In this paper,time-domain radiative transfer equation was applied to simulate the photon propagation of the pulse laser in disperse media and then to obtain abundant transmission and reflection signals of media boundary. Sequential quadratic programming algorithm was introduced to inverse the distribution of absorption and scattering coefficient of two -dimensional disperse media. The generalized Gaussian Markov random fields (GGMRF) model was adopted as regularization term to overcome the ill-posed nature of the inverse problem. Simultaneously,to reduce the computation time,the openMP parallel technology is applied to the part of the codes. The reconstruction results show that the SQP algorithm is a promising method to well reconstruct the distribution of absorption and scattering coefficient of two-dimensional disperse media.

Key words: sequential quadratic programming, properties reconstruction, time-domain radiative transfer, algorithm, radiation, numerical simulation

摘要：

介质的辐射物性的参数测量在高温火焰燃烧诊断、无损检测、生物医学成像等领域有着广泛的应用。传统的直接测量方法得到的大多是测量试件的等效物性，精确的测量通常需要通过反演的方法得到。利用时域辐射传输方程模拟脉冲激光在弥散介质内的传输，求解介质边界的透反射信号。采用序列二次规划（sequential quadratic programming，SQP）算法求解反问题，重建了二维弥散介质内的吸收系数和散射系数分布。利用广义的高斯-马尔科夫随机场模型构建正则化项，加入目标函数中以克服反问题的病态特性。同时使用OpenMP并行计算技术对部分代码做了并行化处理，从而大大减少了重建时间。重建结果表明基于SQP算法能够很好地重建出吸收系数和散射系数的分布。

关键词: 序列二次规划, 物性重建, 时域辐射传输, 算法, 辐射, 数值模拟