旋翼无人机单相机倾斜摄影应急遥感关键技术研究

With the capability of capturing high-resolution and multi-views geodetic images at same time, the Oblique Photogrammetry System (OPS) is attractive to many geographic industries including surveying and mapping, resource and environment research, Disaster reduction and emergency plan, Homeland Security, etc. although it is generally believed that the integration of the OPS with Unmanned Aerial Vehicle (UAV) platform can effectively meet the urgent collection requirement of fine spatial information in a up-to-date, timeliness way, the employment of OPS in Low-altitude flying platform is subjected to several challenges in practice, which covers: a) Due to limited payload and space for onboard equipment, current OPS has to be re-designed to achieve lighter weight and smaller size; b) Due to extremely expensive price and potential security risk from UAV, current OPS is far away from practice for UAV users though UAV has obvious advantages on platform construction, maintenance and operation. C) Other issues like geo-reference encoding, real-time Photogrammetry measurement, robust and easy-to-operate OPS calibration system for complex application environment, etc..This project intends to develop low-cost oblique photogrammetry system onboard Unmanned Rotorcraft toward emergency remote sensing applications and two aspects are involved: a)the development of light-weight, low-cost oblique photogrammetry system with an overall consideration on multi-vision imaging model, available high-speed Scanning imaging technologies and characteristic of Unmanned Rotorcraft platform like flying hover, payload, speed, etc.; b) the development of parallel image geo-referencing system to effectively and efficiently endow 2D oblique images with uniform 3D geodetic coordinate under the FPGA+DSP computing framework. To this end, adaptive relative affine transformation stimulation is introduced into ASIFT algorithm to enhance its performance on oblique photogrammetry wide-baseline stereo matching as well as computation accelerations both in algorithm level and hardware level are implemented for improved ASIFT algorithm and traditional aerial triangulation algorithm, respectively. The project will promote the oblique photogrammetry system to be widely used in disasters dynamic monitoring, evaluation, emergency rescue and other various geographic information applications like supervision of land and resources, digital city and so on. Meanwhile, this project provides a solid foundation of theory and technology for the development of real-time photogrammetry and massive satellite remote sensing original star data real-time processing.

倾斜摄影技术被认为是下一代航空摄影系统发展的方向。本项目率先将高分辨率倾斜摄影技术与时效性、针对性强且受环境限制少的旋翼无人机遥感平台相结合，就以下两方面内容展开研究：第一，突破低空倾斜航空摄影系统轻、小型化关键技术，基于八旋翼无人机平台构建低成本、轻载荷的单相机倾斜摄影系统原型，实现高清倾斜影像的低空探测与快速获取；第二，突破多角度交向摄影条件下的倾斜影像立体匹配、大规模空中三角测量等并行处理关键技术，构建FPGA+DSP架构下的嵌入式倾斜影像地理编码并行系统原型，实现倾斜影像的高效地理编码，满足应急遥感实时处理迫切需要，本项目将有力推动、促进倾斜摄影这一新对地观测技术在我国地质灾害动态监测、评估、应急救援以及国土资源监察、数字城市等地理信息行业的广泛应用，并为当前摄影测量与遥感发展重要方向-实时摄影测量以及海量卫星遥感原始数据星上处理，提供坚实的理论和技术基础。

本项目积极探索将倾斜摄影技术应用于低空无人机平台及其应急遥感应用开展的关键科学、技术问题，以满足地理信息应用行业对高分辨率遥感影像、地面精细空间信息获取现势性、及时性日益增长的迫切要求.吸取自动化控制、计算机视觉、数字摄影测量、FPGA并行运算以及嵌入式系统等多个领域最新研究成果，本项目就轻小型倾斜摄影系统构建、FPGA倾斜影像高效地理编码、倾斜遥感影像处理方法发展三方面主要内容展开研究:a)轻小型倾斜摄影系统构建方面，突破倾斜摄影系统轻、小型化关键技术，构建适合无人低空遥感平台的低空倾斜摄影系统原型，实现高清倾斜影像的低空探测与快速获取，同时吸收全景成像技术的最新成果，引入具有超大视野的全景相机实现目标区域空间影像信息的完整采集，以简化应急遥感系统设计并降低成本，并发展了针对不同类型商用全景相机的标定技术及全景影像处理方法，以满足后续影像地理编码及应用处理需要；b)FPGA倾斜影像地理编码方面，突破倾斜影像立体匹配、大规模空中三角测量等并行处理关键技术,构建FPGA+DSP架构下的倾斜影像地理编码并行系统原型，实现大范围倾斜影像的高效地理编码。以之为基础，基于华为海思芯片建立“ARM+GPU”架构下的嵌入式无人机上遥感影像实时拼接系统，既满足无人机航拍规划、影像质量在线控制等应用对影像在线、实时处理需求，又为各种低空倾斜摄影应急应用奠定基础；c)倾斜遥感影像处理方法发展方面，结合FPGA倾斜影像地理编码要求对影像特征检测、匹配、分割方法进行了重新设计与发展创新,结合无人机应急遥感应用要求对多影像几何纠正、拼接及其在线处理方法进行了重新设计与发展创新,结合立体影像三维建模需求对点云语义分割处理、计算全息三维建模、显示方法进行了重新设计与发展创新。本项目将有力推动、促进倾斜摄影这一新对地观测技术在我国地质灾害动态监测、评估、应急救援以及国土资源监察、数字城市等地理信息行业的广泛应用，并为当前摄影测量与遥感发展重要方向-实时摄影测量以及海量卫星遥感原始数据星上处理，提供坚实的理论和技术基础.