航空影像快速处理系统之关键技术研究

发布时间：2018-03-16 14:34

本文选题：快速处理　切入点：POS系统　出处：《北京建筑大学》2013年硕士论文　论文类型：学位论文

【摘要】：自然灾害和公共安全事故灾难的频繁发生、灾害损失的不断增长，引起了党中央、国务院以及各部门对防灾抗灾工作的重视。在多次灾害应急及灾后重建工作中，测绘部门在其中起到了基础地理信息保障的重要作用。由历年来国内发生的重大自然灾害（如汶川地震、舟曲泥石流等）的灾后救援和应急策略可知，第一手的灾情数据信息对于及时的救援和救灾策略制定，有着至关重要的作用。而测绘部门往往可第一时间提供灾前基础地理信息数据，但无法在第一时间获取灾区灾害后的航空影像、第一时间制作完成灾区影像图。存在问题主要表现在灾时空域紧张以及应急测绘航摄数据快速处理装备不足两个方面：一是测绘航摄用飞机无优先空域权，在生命高于一切的抢险救灾过程中，空中救援运输是首要任务，必须重点保障，从而导致测绘部门的航摄飞机难以第一时间获准起飞执行任务；二是由于现阶段使用的航摄遥感软件大多是后处理软件需要人工参与，飞行过程中无法实时解算处理，必须待飞机落地停稳后利用大量的地面控制点才能解算处理，显然在分秒必争的应急时刻，迫切需要实现机上航空数据实时处理，飞机落地即可获得灾区高分辨率航空影像图。为此，需建立一套搭载在通用飞机平台上的应急测绘航空数据实时获取与机上快速处理系统。基于此，为了满足应急测绘航空数据机上快速处理的应用要求，实现飞机落地时即可获得灾区高分辨率遥感影像图，本文研究了基于三维地形影像控制数据库（即影像数据库与地形数据库）的机上快速处理流程，利用已有的数据库，形成基于三维地形影像控制数据库的机上航空遥感数据正射纠正、灾区影像拼接及镶嵌后影像几何纠正等的快速处理工艺，减少了人工参与的过程，解决了航空遥感数据应急获取与处理的时间效率问题，建立一套应急测绘航空影像快速处理系统来满足抗灾救灾等应急保障的需求。本文主要研究成果如下：（1）重点分析了传统航空影像处理的流程方法，同时针对航空影像和“应急响应”的特点，提出了应用已有的三维地形影像控制数据库的快速处理流程，解决了传统作业方法时间长、步骤多和效率低以及不能实时在机上处理的缺点，并且通过该方法能快速生成灾区高分辨率整体影像图，，来满足灾后应急宏观把控的需求。（2）通过对正射纠正的方法研究，在缺乏控制信息的情况下提出了应用POS数据和地形数据库中DEM数据来对航空影像进行快速正射纠正，省去了传统空三加密的过程，减少了整个处理的时间，实现并生成带地理坐标的正射影像，为后续的基于坐标的几何镶嵌提供影像资料。（3）为了实现影像的快速拼接，采用了基于坐标的拼接方式，同时在实验中，针对一些POS精度不高的情况，提出了基于坐标微调的拼接方法，利用匹配相邻影像之间的同名点对来计算坐标偏移量，解决部分区域POS精度不高的问题，进而快速准确的得到灾区宏观的镶嵌影像，满足应急救灾需求。（4）运用控制点影像库内多源控制点影像对几何镶嵌后的整体影像进行平差纠正，分析了控制点影像数据库及内部数据的特点以及影像的特征提取、特征匹配算法和流程，同时实现了多源遥感影像间的匹配。本文采用二次多项式的方法，利用匹配好的控制点数据对镶嵌后的影像进行整体的平差和纠正，解决影像由于POS数据和DEM数据精度问题造成的位置偏差，提高了镶嵌后整体影像的精度。
[Abstract]:Frequent natural disasters and accidents of public security disaster, increasing disaster losses, caused by the Party Central Committee, the State Council and the Department of disaster prevention and disaster relief work attention. In times of disaster emergency and post disaster reconstruction work, surveying and mapping departments to the important role of basic geographic information security in them.
A major natural disaster in China over the years (such as the Wenchuan earthquake, Zhouqu mudslides) the disaster relief and emergency strategy that the disaster data first hand information for timely rescue and relief strategy plays a very important role. While surveying and mapping departments often provide pre disaster basic geographic information data for the first time, but unable to get the aerial image after the disaster disaster area in the first time, the first time made disaster image maps. The problems mainly in the spatial domain and nervous disaster emergency surveying and mapping aerial data fast processing equipment less than two: one is the aerophotographic mapping plane without priority right in the airspace, the life is higher than all the disaster relief process air rescue, transportation is the primary task, we must focus on security, resulting in surveying and mapping departments of aerial aircraft to the first time allowed off the task; two is due to At this stage the use of aerial remote sensing software is mostly postprocessing software need artificial participation during the flight can not be real-time solution treatment, must be stopped after the plane landed a large number of ground control points to solution treatment, apparently in emergency time the urgent need to achieve machine count every minute and second, the aviation data real-time processing, the aircraft can be landed obtained high resolution aerial images of disaster. Therefore, to establish a set of emergency mapping data in the general aviation carrying aircraft on the platform of real-time processing and machine system.
Based on this, in order to meet the application requirements of emergency mapping aviation data fast processing machine, realize the plane figure of high resolution remote sensing images can be obtained when the landing area, this paper studies the three-dimensional terrain database based on image control (i.e. image database and terrain database) fast processing machine, using the existing database, forming a three-dimensional terrain image control based on the database on the aerial remote sensing data correction, fast processing and image mosaic mosaic area after image geometric correction, reduce the artificial participation, solve the aerial remote sensing data acquisition and processing efficiency of the emergency time, to meet the needs of disaster relief and other emergency security system to establish a set of emergency rapid processing surveying and mapping aerial images.
The main research results in this paper are as follows:
(1) focuses on the analysis of the process of traditional aviation image processing, at the same time for aerial images and "emergency response" features, put forward a fast process of 3D terrain image control of the existing database application, to solve the traditional operation method for long time, many steps and low efficiency and not real-time processing in machine faults and by this method can quickly generate a high resolution image of the whole area, to meet the post disaster emergency macro-control needs.
(2) through the research on the method of ortho rectification, in the absence of control information is presented under the condition of the application of POS data and DEM data in terrain database for fast orthorectification for aerial image, the traditional process of eliminating the empty three encryption, reduce the processing time, the generation of orthoimage belt the geographical coordinates, for subsequent geometric mosaic based on coordinate image data.
(3) in order to achieve rapid mosaic images, using the splicing method based on coordinate, at the same time in the experiment, some POS accuracy is not high, the mosaic method based on coordinate adjustment, using the same point matching between adjacent images to calculate the coordinates of the offset, solve part of POS accuracy the problem, then quickly and accurately get hit mosaic image macro, meet emergency relief needs.
(4) the use of multi-source control point image database in the control point image on the overall image geometric mosaic after adjustment correction, analyzed the control point image database and the internal features of the data and image feature extraction, feature matching algorithm and the flow, and realize the matching between multi-source remote sensing images. This paper uses the two polynomial on the whole, the mosaic image using control point matching good data adjustment and correction, solve the position deviation due to the POS image data and DEM data accuracy caused the problem, improve the precision of the whole image mosaic.

【学位授予单位】：北京建筑大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P231

【参考文献】