【作者】 周国香；

【导师】 卢秀山；

【作者基本信息】 山东科技大学 ， 摄影测量与遥感， 2009， 硕士

【摘要】 由于受航空平台及传感器的限制,传统航空摄影测量在获取小面积、大比例尺数据方面,存在着成本高、性价比差、高程精度达不到规范要求等问题。而无人飞艇低空遥感系统具有低成本、机动灵活等优点,能够低空低速飞行,在小区域内快速获取高质量遥感影像,可以作为国家航空遥感监测体系的重要补充。由于受无人飞艇平台体积、重量等方面的限制,目前无人飞艇遥感系统大多采用带稳定平台的单相机作为遥感设备。由于单个CCD面阵尺寸较小,其影像的地面覆盖范围和基高比都比较小,致使航测内、外业工作量增加,高程精度较低。因此,采用多相机拼接技术,增大像幅尺寸,成为目前摄影测量界研究的热点。本文对组合相机的拼接方案进行了研究,分析了多面阵组合相机拼接需要考虑的各项参数,包括相机间距、相机倾角、影像重叠度、虚拟像幅等。并结合Cannon 5D数码相机的各项参数具体分析了相机间距对拼接误差的影响,相机倾角对图像变形的影响,以及相机倾角与影像重叠度的虚拟像幅之间的关系。结论表明:相机间距越大,导致的拼接误差越大;像片倾斜导致GSD随之变化,并且倾角越大,GSD变化就越明显;相机倾角越大,重叠度越小,虚拟像幅尺寸越大,但虚拟有效像幅的尺寸越小。数码相机属于非量测相机,其内方位元素未知,且镜头存在较大的畸变差,使用之前必须进行精确的检校;组合相机固定好之后,相机之间的“相对外方位元素”也就确定下来,必须将其标定,作为初值参与运算。本文介绍了基于空间后方交会原理的数码相机检校方案,推导了双拼相机检校的数学模型,并对实现了对单个相机和双拼相机的检校。结果表明利用大型室外检校场对数码相机进行检校,可以获得很高的检校精度。本文对多面阵组合相机虚拟影像的拼接技术进行了重点研究,其具体流程可以概括为:首先对各子影像进行畸变差改正,借助组合相机检校获取的“相对外方位元素”初始值对各子影像进行水平纠正,然后对虚拟片重叠区域内的影像进行特征匹配寻找同名点,借助同名点完成子影像间的相对定向,求得精确的拼接参数,最后利用这些拼接参数完成虚拟影像的拼接。本论文最后通过实验验证了利用无人飞艇搭载单相机系统可以完成1:2000的航摄任务,而搭载双拼相机系统可以完成1:1000的航摄任务。更多还原

【Abstract】 Due to the restriction of aeronautical platform and sensors, the traditional aerial photography has many shortcomings in gagining small area, large scale data such as high-cost, low performance-to-price ratio, the elevation accuracy can not achieve to the precision requirements etc. With the merits of low cost, flexible and mobile etc, low attitude UAVRS-F (Unmanned Aerial Vehicle Remote Sensing) can quickly acquire high-quality remote images in small area region. It can be used as one important complement of national aviation remote sensing system.Owing to the restriction of platform’s volume and weight, UAVRS-F can only use single camera system with stable platform presently. Single digital camera’s coverage and base-height ratio is small, which increase the interior and field workload, moreever, the elevation accuracy is rather low. So, using Multi-Format mosaic technology, increasing the format size has become a research focus recently.In this paper the camera-combined schemes was studied. Several mosaic parameters were analyzed, including camera space, camera tilt angle, image overlap degree, virtual image etc. Combined with the parameters of Cannon 5D digital camera, the paper analyzed concretely the influence on mosaic error by camera space, the image deformation law resulting from camera tilt angle, the relationship among camera tilt angle, overlap degree and virtual image. Several conclusions were obtained such as: the camera space is larger, the mosaic error is larger, photoes tilting leads to the Ground Sampled Distance (GSD) is different; and the tilt angle is bigger, the image deformation is more obvious; the camera tilt angle is larger, the overlap degree is smaller, the virtual image format is larger, but the effect virtual image format is smaller.Digital cameas belong to non-metric cameras, its inner orientation elements are unknown and the lens’ distortion is large. So, the digital cameras must be accurately calibrated before used. After the combined-camera having been fixed, the relative exterior orientation elements are also be determined, which need to be calibrated and be used as the initial values to participate computation. The paper introduced the digital camera calibration scheme based on resection principle, deduced the mathematic model for Two-Format cameras, and realized the calibration for single camera and combined camera. The results showed that digital camera calibration susing large outdoor calibration field can reach very high accuracy.The paper put emphasis on the study of virtual image mosaic technology for Multi-format combined cameras. The concrete flow include: the distortion calibration of single photos, the photos’ transformation from oblique plane to virtual horizontal plane by using the relative elements of exterior orientation, searching homonymy points on the images in overlap areas by feature points matching to complete relative orientation. Through the former case, the precise mosaic parameters were gained, then using these parameters to complete the mosaic of virtual image.At last, through experiments the paper proved that single camera system can complete 1:2000 scale aerial photographic tasks, and the Two-Format digital camera system can achieve 1:1000 scale aerial photographic tasks.更多还原