Research On Key Technologies Of Low Altitude UAV Borne Multi-head And Wide-angle Composite Cameras And Their Performance Evaluation

In recent years, low altitude aerial photogrammetry based on unmanned aerial vehicles (UAV for abbreviation) has become increasingly developed, and turns into an important data acquisition technology complementing satellite remote sensing and traditional aerial photogrammetry. Its main application purpose is to acquire high resolution aerial images and to carry out large scale mapping. However, current UAV-based low altitude aerial photogrammetry focuses on light and small fixed-wing UAVs equipped with single head cameras, resulting in lower operation efficiency, less height precision, and lower platform stability. On the other hand, multi-head and wide format aerial cameras available home and abroad are relatively mature, meaning little roomfor UAVs. Therefore, the fundamental solution to the existing problems of low-altitude UAV-based aerial photogrammetry is to develop more sophisticated UAV platforms and wide-format and wide field of view multi-head aerial cameras suitable to UAV platforms(i.e., multi-head and wide-angle camera).The research on multi-head and wide-angle camera needs to take into account both application and theoretic aspects. In application, the first issue concerns how to combine multiple small format cameras into a wide format aerial camera that approximately satisfies single central projection. Secondly, there is the issue of integration of the multi-head and wide-angle camera with high precision POS system. Initial exterior orientation elements of high precision are not only the requirements of high precision aerial photogrammetry, but helpful to greatly improve the management and processing of low altitude aerial images. On the theoretic side, the constructions of aerial cameras and the process of aerial photographing should also emphasize the guidelines and theoretic basis of photometry in addition to the well-established geometric optic principles in the field of photogrammetry. Moreover, aerial cameras are located in a moving platform during aerial photography, leading to image motion affecting the performances of aerial cameras. This paper introduces some information measures pertaining to information theory to evaluate the information content of images acquired by the multi-head and wide-angle cameras, thus evaluating their performances.The main work and contributions of this thesis are as follows:(1) Several design and assembling aspects and merits of the current airborne aerial cameras are systematically summarizedas a guide to the thesis work on multi-head and wide-angle cameras. The characteristics of current airborne aerial cameras and their limitations for UAV platforms are analyzed. Then, some design and assembling aspects of these cameras, including CCD types, combination and stitching mode, exposure mode, color image generation, image motion compensation, field of view(FOV for abbreviation), and GPS/IMU integration, are summarized. This forms the basis for the assertion that the aerial camera suitable for low altitude UAV platform should be multi-head and wide-angle cameras. Finally, types of suitable cameras are discussed, and the physical structuring of the multi-head and wide-angle camera is accomplished.(2) The method and process of static geometric calibration for the multi-head and wide-angle cameras are studied. For this, the geometric imaging model is derived, which establishes the rigorous geometric relationships between sub-images, and between sub-images and virtual (composite) image. Furthermore, single camera calibration and multi-camera joint calibration are carried out to solve the orientation elements and distortion parameters of the camera. At last, the calibration results are analyzed.(3) A self-calibration model that takes into account both the relative orientation motions and the remaining radial distortions of the sub-images is proposed. Self-calibration refers to a dynamic calibration method which compensates slight changes of the primary static calibration parameters of the multi-head and wide-angle camera due to the synchronization errors and changes of relative fight heights during aerial photography. The mathematical model of self-calibration is first derived which considers the relative changes of positions and attitudes and the remaining radial distortions. A fast matching method for short baseline sub-images of the multi-head and wide-angle camera is realized, with the corresponding points matched in the overlapped areas of sub-imagesfurther utilized to accomplish self-calibration. At last, the precisions of self-calibration results are analyzed.(4) A method for color balancing of virtual images based on the principles of photometry is proposed. First, related physical quantities pertaining to photometry are introduced. And the exposure control mechanism of the aerial camerabased on photometry during aerial photographyis discussed. Second, through the analysis of radiation transmission during the imaging process, it is revealed that differences of the colors among a set of sub-images lie in the differences of their exposures. Hence, the mathematical model of the exposure differences is derived. And the exposure compensation parameters are precisely obtained via robust estimations. Finally, compared with three other traditional methods, the proposed method has demonstrated its better performance and robustness.(5) A theoretic viewpoint and a quantitative paradigm that combines the principles of photometry and information theory is proposed to direct aerial photography and evaluate data quality. At the beginning, the principles of photometry are utilized to reveal the imaging issues and to put forward a set of normative practical procedures. Second, the limitations of classical information theory applied to measure the information amounts of images are analyzed. And the quantitative method suitable for measuring information capacity of an optical imaging system is introduced. Based on the information measures, such as information entropy, signal-to-noise ratio, and informationcontent, both ground-based and flight tests are carried out to objectively evaluate the impact of spatial resolution, image motion, and flight height on the image quality. Finally, the relationships between the information quantification results according to information theory and the guidelines according to photometry are established. The experiments show that the results of information measures are consistent with the guidelines of photometry, which ascertain that information measures could be used as a quantitative basis for evaluating image quality.(6) A method for estimating the synchronization error between the POS system and the multi-head and wide-angle camera during aerial photography is proposed. The integration of a POS system (GPS/IMU) with the multi-head and wide-angle camera is an important component of the work. However, the measurements of the POS system are different from the actual values corresponding to the exposure time due to synchronization error and would further affects the precisions of converted primary exterior elements pertaining to each exposure time. For this, related coordinates are defined and the conversion from the measurements to the exterior orientation elements is discussed. Then, the calibration model of the boresight misalignments between the POS system and the camera is derived. And a strict estimation method for the synchronization error is put forward on the premise that the boresight misalignments should be a constant. The experimental results show that the precisions of calibrated boresight misalignments and angular elements conversation are remarkably improved while considering the synchronization error. Finally, through experiments of aerial photography over a test area with ground control points, the geometric precision of the multi-head and wide-angle camera is evaluated. Besides, the feasibility of aerial triangulation with spare control points aided by high precision POS system is analyzed.