非合作安检太赫兹快速高分辨率成像研究

For the security inspection at the high-throughput public places, this research project focuses on the novel techniques and approaches of terahertz imaging. It aims to make the solutions of the key scientific issues and technical challenges for improving comprehensive ability of high-throughput security inspection. Based on the multi-input and multi-output (MIMO) system, the main contents to establish the physical basis of the project include the system of the non-cooperation terahertz security imaging, the optimized scheme of the sparse array, the transceiver array design and integration, the way of both the transmitting and the receiving of the broadband frequency modulation (FM) signal and so on. By studying both the new mechanisms and methods for MIMO technique, such as the algorithms of synthetic aperture imaging, the reconstruction theory of sparse targets, the motion compensation method for non-cooperative targets and so on, it builds the algorithm foundation for imaging with high frame rate, large depth of field and adaptive-focusing capability. Furthermore, by studying the terahertz characteristics of the target appeared at the security areas, it builds the theoretical and experimental foundation for the multi-dimensional and multi-polarization information fusion, and then provides the method basis for the target recognition of high-throughput security. Based on the above mentioned, this project will carry out the principle experiment research about terahertz three-dimensional fast imaging for non-cooperative targets. System-level verification of new devices, imaging mechanisms and algorithms will also be taken into account. Finally, plenty of comprehensive innovations on devices, systems, signal and image processing, target characteristics and identifications, etc., would be formed. We believe in that the development of our project will enhance the national basic research and technological innovation in the field of non-cooperative security inspection and terahertz science.

本项目围绕重要社会场所高通量安检，开展太赫兹成像新体制与新方法研究，解决提升高通量安检综合能力的关键科学问题与技术挑战。通过研究MIMO体制下，太赫兹非合作安检系统方法、稀疏阵列拓扑优化方法、阵列设计与集成方法、宽带捷变调频信号产生与接收方法等为高通量安检的实现建立物理基础。通过研究MIMO高效合成孔径算法、稀疏信息图像重建理论、非合作目标运动补偿方法等新机理和新方法，为高帧率、大景深、自适应聚焦成像建立算法基础。通过系统研究安检目标的太赫兹特性，为多维度、多极化信息融合奠定理论与实验基础，进而为高通量安检的目标识别提供方法基础。基于以上理论、方法与技术，项目将开展非合作目标太赫兹三维快速成像的原理实验研究，对新型器件、成像体制和算法进行系统级验证，形成贯穿太赫兹器件、系统方法、信号与图像处理、目标特性与识别等方面的综合创新成果，增强我国在非合作安检及太赫兹领域的基础研究和科技创新实力。

本项目聚焦重要社会场所高通量安检，开展太赫兹成像新体制与新方法研究，解决提升高通量安检综合能力的关键科学问题与技术挑战。围绕非合作安检太赫兹快速高分辨率成像的技术难题，包括：（1）待测目标运动状态高度不确定，（2）太赫兹波相位提取同时性要求高，（3）太赫兹回波数据量大，（4）三维图像重建算法复杂、计算量大，（5）敏感隐藏目标识别实时性高等，提出并验证了一系列关键技术，实现了两套关键技术验证平台，对关键技术解决方案进行了成功演示。项目在理论和实验上解决的关键技术包括：（1）创新性的提出合成孔径与机械扫描相结合的三维成像新体制，在横向采用准光学柱面镜形成线形扫描波束进行一维机械扫描，在纵向采用合成孔径成像体制，解决了柱面镜的准光学设计、待测目标运动补偿等，（2）采用稀疏多进多出（MIMO）方案，对收发阵列拓扑结构进行优化，进一步提高合成孔径成像速度，（3）系统深入分析非合作安检场景的纵向稀疏特性，优化压缩感知算法，结合GPU加速技术，提高三维图像重建的精度和速度，（4）采用深度学习技术，自动对复杂敏感目标进行高准确度识别。在此基础上，研制成功两套太赫兹雷达成像演示系统，对太赫兹三维实时成像关键技术进行了试验研制。本项目的知识产权成果包括：发表42篇学术论文，授权发明专利13项，申请发明专利12项，登记软件著作权4项；在人才培养方面培养博士6人，硕士12人。本项目实现了项目预期目标，实现了各项技术指标。本项目的研究成果有效促进了我国在非合作目标太赫兹三维实时成像方面的进展。