SIA OpenIR  > 光电信息技术研究室
红外多光谱场景建模与仿真方法研究
其他题名Modeling and Simulation of Infrared Multispectral Scene
李波1,2
导师赵怀慈
分类号TP391.9
关键词多光谱场景 建模与仿真 仿真图像 辐射模型 光照模型
索取号TP391.9/L31/2016
页数113页
学位专业模式识别与智能系统
学位名称博士
2016-11-30
学位授予单位中国科学院沈阳自动化研究所
学位授予地点沈阳
作者部门光电信息技术研究室
摘要红外多光谱场景建模与仿真方法是仿真领域的前沿热点与难点之一。美国国防部试验资源管理中心提出多光谱测试计划,将多光谱场景建模与仿真技术作为重大需求,列入重点研发计划。多光谱红外仿真技术作为核心国防科学技术,美国等先进国家设置了障碍,对我国进行技术封锁,严重影响了我国相关技术的发展。 本文面向真实物理环境仿真重大需求,开展了红外多光谱场景建模与仿真方法研究。本文的主要研究内容包括: 1 在分析物理场景辐射基本原理的基础上,定义了包含地物对象、大气对象、太阳对象等基本元素的仿真场景六元组模型。设计了场景仿真的基本数据模型和数据结构,构建了适合于多光谱仿真的基于计算机集群和GPU的并行计算框架;在此基础上对场景的红外辐射特性进行了多光谱仿真,生成了多光谱图像序列,并与国际主流仿真软件的仿真结果进行了比对,平均误差在5%左右。 2为了提高多光谱仿真场景的逼真度,分别从红外波段多光谱光照建模和辐射影响两个方面开展了研究。首先,由顺光、逆光问题引出了探测方向对目标辐射特性的影响研究,提出了一种红外波段的光照反射模型,有效地反映了不同探测方向对红外图像的影响,实验结果验证了模型的有效性和准确性。其次,对场景中背景物对目标的辐射影响进行了研究,提出了一种背景对目标辐射特性影响仿真计算模型,实验结果验证了模型的有效性。 3 提出了一种基于卫星多光谱图像的红外多光谱大场景快速仿真方法。针对仿真建模中的材质分类问题,提出了一种基于SAM和SID的自适应混合材质分类方法,能够有效提高卫星图像中材质分类精度。结合分类结果和通过地面高程数据构建的简化三维模型,快速建立具有材质标定的大规模场景三维模型,并在此基础上推演出该场景在特定气候条件下的红外多光谱仿真图像。 4设计并实现了多光谱场景仿真软件原型系统。完成了软件系统的结构设计、功能模块划分、关键类的设计;分析并设计了系统的时序关系、交互关系,定义了系统工作流程,为将来进一步研究提供了基本的软件支持。
其他摘要Modeling and Simulation of Infrared Multispectral Scene is one of the hotspots in the forefront of scene simulation. The Multispectral Testing project proposed by TRMC DOD focus on the research of modeling and simulation methods of multispectral scene, and provides continuous support. As one of the cores of national defense science and technology, multispectral scene simulation technology is forbidden to deliviered to China by the developed countries such as USA. This paper faces the big needs of simulation of physical scene, and focus on the technology of modeling and simulation of multispectral scene. The work of this paper is mainly embodied in the following aspects: 1 On the basis of analyzing the radiation principles of the physical scene, a definiation of simulation scene is given in the form of tuple with 6 items which includes ground objects, the atmosphere environment, the sun and so on. The data models and structures for computer simulation are designed. The simulation of the radiation of the scene is done on a paraellel computation framework of computer cluster and GPUs, and a serial of multispectral images are generated. The average difference between the generated images and images from JRM is around 5%, which verified the correctness of this method. 2 In order to improve the fidelity of the simulated images, the illumation model and radiation influence in multispectrum are studied. First, the radiation effection caused by detecting direction is studied. An illumation model which can be used in IR simulation is presented, and reflects the effection caused by detecting direction on simulated images effectively. The experiment results verify the validity and accuracy of the model. Second, the radiation influence caused by nearby background object is studied. A radiation influence simulation computation model is presented, and the experiment results verify the validity of the model. 3 A new method for rapid large scale scene simulation is presented on the basis of mulitispectral satellite images. An adaptive hybrid material classification method based on SAM and SID is presented to improve the classification accuracy of materials in the satellite image. A simple 3D scene with material could be constructed by combining DED and the classification results. The multispectral IR simulation images in specific climatic conditions could be derived based on the 3D scene. 4 A prototype of multispectral scene simulation software is designed and implemented. The system structure and function modules are expounded, the key class definition, the sequence diagram, the interaction diagram and the work flow of the system are stated. The software provides a basic foundation condition for further work.
语种中文
产权排序1
文献类型学位论文
条目标识符http://ir.sia.cn/handle/173321/19451
专题光电信息技术研究室
作者单位1.中国科学院沈阳自动化研究所
2.中国科学院大学
推荐引用方式
GB/T 7714
李波. 红外多光谱场景建模与仿真方法研究[D]. 沈阳. 中国科学院沈阳自动化研究所,2016.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
红外多光谱场景建模与仿真方法研究.pdf(4248KB)学位论文 开放获取CC BY-NC-SA请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[李波]的文章
百度学术
百度学术中相似的文章
[李波]的文章
必应学术
必应学术中相似的文章
[李波]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。