时频光热多源信息融合的非均匀弥散介质多宗量场反演研究

Non-contact measurement of multi-parameter fields (optical properties, thermal paramters, temperature fields etc.) in disperse media is a challenging problem in many application areas, such as bioluminescent imaging, combustion diagnostics, infrared detection, target detection and new bio-energy utilization. These multi-parameter fields need to be retrieved by inverse methods using measured thermal and/or optical information. Due to the complex interrelation between multi-parameter fields, the inverse problem of simultaneous reconstruction in multi-dimensional non-uniform disperse media possesses serious ill-posed and multi-valued features. It cannot be effectively solved by present inverse theories based on single optical or thermal information. Thus, this study proposes a solution for this big challenge, in which a hybrid intelligent optimization and gradient-based algorithm combined with regularization theory is developed to solve the inverse problem based on synthetizing the time-frequency domain thermal and optical information as the prior information. This project includes the following research aspects: An efficient and accurate forward model is established to simulate the time-frequency optical and thermal information on the boundary of disperse media. On this basis, the time-resolved optical information, amplitude and phase information of frequency modulated light and time-resolved temperature response are synthetized as the prior information. The sensitive ranges of the time-frequency thermal and optical information for the optical and thermal parameters reconstruction in dispersed media are determined. Then, new theories and algorithms are developed to overcome the ill-posed characteristic. Finally, a general multi-parameter fields reconstruction theory and method for nonuniform dispersed media are developed. It will provide a theoretical foundation and algorithm support for the developing of high technologies such as optical imaging and combustion diagnostics etc.

弥散介质内多宗量场（光学特性、热物性、温度等）非接触测量是生物光学成像、燃烧诊断、红外检测等领域面临的难题，需通过测量外部光或热信号，利用数值反演方法实现。然而多维非均匀弥散介质多宗量间存在复杂相互关联性，致使协同重建逆问题存在严重病态性和多值性，依靠现有的单一光或热信息反演理论无法有效解决。因此，本研究提出综合利用时频光热信号等多源信息，采用智能优化-梯度混合算法，结合正则化理论，旨在解决多维非均匀弥散介质内多宗量场的协同反演。内容包括：建立高效准确的弥散介质时频域光热信息正算模型；在此基础上，融合时间分辨光信息、调频光幅值相位信息及时变温度响应等多源信息，确定时频光热测量信号对弥散介质光热参数重建的敏感区间，发展克服病态、多值性的新模型和新算法；最终构建具有广泛适用性的非均匀弥散介质多宗量场协同重建理论及方法，为光学成像、燃烧诊断等高新技术的发展提供理论基础和算法支撑。

弥散介质内多宗量场（光学特性、热物性、温度等）非接触测量是生物光学成像、燃烧诊断、红外检测等领域面临的难题，需通过测量外部光或热信号，利用数值反演方法实现。然而多维非均匀弥散介质多宗量间存在复杂相互关联性，致使协同重建逆问题存在严重病态性和多值性，依靠现有的单一光或热信息反演理论无法有效解决。因此，本研究提出综合利用时频光热信号等多源信息，采用智能优化—梯度混合算法，结合正则化理论，旨在解决多维非均匀弥散介质内多宗量场的协同反演。本项目主要研究内容包括：建立了高效准确的弥散介质时频域光热信息正算模型；在此基础上，融合时间分辨光信息、调频光幅值相位信息及时变温度响应等多源信息，确定了时频光热测量信号对弥散介质光热参数重建的敏感区间，发展了克服病态、多值性的新模型和新算法；最终构建了具有广泛适用性的非均匀弥散介质多宗量场协同重建理论及方法，为光学成像、燃烧诊断等高新技术的发展提供了理论基础和算法支撑。主要成果包括发表刊物论文44篇（其中国际刊物40篇）、国际会议论文4篇，其中，SCI检索40篇，EI收录4篇，申报发明专利8项。培养博士后1人（在站），博士研究生5人（已毕业）、硕士研究生6人（已毕业）。