为了改善无人机侦察视频质量，针对目前无人机摄像、照相数据特点，提出了一种基于前方交会的无人机航摄影像定位方法。首先，对所有无人机侦察视频帧和航片分别进行特征匹配，得到影像间关系；然后，提出基于前方交会与抗差估计的无人机航摄影像定位方法，来实现航摄影像的定位；最后，对侦察视频帧进行高频补偿与凸集投影迭代优化，得到重建后侦察视频。实验结果表明，基于前方交会与抗差估计的无人机航摄影像定位方法加强了无人机视频与航片的对应性，凸集投影迭代优化法增强了重建的边缘保持能力。该方法增强了重建图像的一致性与保真度，特别是对图像边缘细节部分等效果极为明显，且处理速度更快。
Super-resolution reconstruction for unmanned aerial vehicle video based on aerial photography locating (SRRUAVVAPL) is proposed to improve the quality of UAV reconnaissance video. Firstly, feature matching of all the video frames is performed to figure out the relations among the video frames, and feature matching of all the adjacent frames are also performed to figure out the relations among the adjacent frames. Then an unmanned aerial vehicle target locating scheme based on space forward intersection and robust estimation (UAVTLSFIRE) is put forward to realize aerial photography locatiing. Lastly, video frames are improved by high-frequency compensation and POCS iterative optimization, and the reconstructed video is obtained. Experimental results show the proposed method is an effective super-resolution reconstruction for unmanned aerial vehicle video.
[1] 王志强, 程红, 杨桄, 等. 全局图像配准的目标快速定位方法[J]. 红外与激光工程, 2015, 44(增刊1):225-229.
[2] Lowe D G. Distinctive image features from Scale-Invariant keypoints[J]. International Journal of Computer Vision, 2004, 60(2):91-110.
[3] Fischler M, Bolles R. Random sample consensus:A paradigm for model fitting with applications to image analysis and automated cartography[J]. Communication of the ACM, 1981, 24(6):381-395.
[4] 曾凡洋, 钟若飞, 宋杨, 等. 车载全景影像核线匹配和空间前方交会[J]. 遥感学报, 2014, 18(6):1230-1236.
[5] 吴迪军, 张萌萌, 潘飞, 等. 多片前方交会法无人机测图技术[J]. 测绘科学, 2018(10):1-10.
[6] 李忠美, 边少锋, 瞿勇. 多像空间前方交会的抗差总体最小二乘估计[J]. 测绘学报, 2017, 46(5):593-604.
[7] 时文俊, 郭从洲, 童晓冲, 等. 辐射保真的红外遥感图像超分辨率重建[J]. 西安电子科技大学学报, 2019, 46(2):107-113.
[8] 吴世豪, 罗小华, 张建炜, 等. 基于FPGA的新边缘指导插值算法硬件实现[J]. 浙江大学学报(工学版), 2018, 52(11):2226-2232.
[9] 郭桐宇, 宋伟东. 基于高频增强曲线与迭代反射投影的遥感影像超分辨率重建[J]. 测绘工程, 2018, 27(1):64-67.
[10] 许丽娜, 何鲁晓. 基于凸集投影的高分四号卫星影像超分辨率重建[J]. 测绘学报, 2017, 46(8):1026-1033.
[11] 李展, 陈清亮, 彭青玉, 等. 基于MAP的单帧字符图像超分辨率重建[J]. 电子学报, 2015, 43(1):191-197.
[12] Yuan Y, Zheng X, Lu X. Hyperspectral image superresolution by transfer learning[J]. IEEE Journal of Selected Topics in Applied Earth Observations & Remote Sensing, 2017(99):1-12.
[13] 王素玉, 卓力, 沈兰荪, 等. 一种简单有效的视频序列超分辨率复原方法[J]. 北京工业大学学报, 2009, 35(6):742-747.
[14] 杨蕊, 刘朝晖, 折文集. 遥感面阵凝视图像并行超分辨重建方法[J]. 红外与激光工程, 2019, 48(1):288-295.
[15] 雷俊锋, 王赫, 朱力, 等. 基于分数阶微积分的图像超分辨率重建[J]. 系统工程与电子技术, 2017, 39(12):2850-2856.
[16] 张岩, 李建增, 李德良, 等. 快速自适应鲁棒性尺度不变的特征检测子[J]. 系统工程与电子技术, 2017, 45(6):1406-1413.
[17] 张岩, 李建增, 李德良, 等. 鲁棒性交叠的标准特征描述子[J]. 北京理工大学学报, 2017, 37(8):842-847.
[18] Bian J W, Lin W Y, Matsushita Y, et al. GMS:Gridbased motion statistics for fast, ultra-robust feature correspondence[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Honolulu, USA:IEEE Computer Society, 2017:4181-4190.
[19] 张岩, 孙世宇, 胡永江, 等. 基于特征距离与内点的随机抽样一致性算法[J]. 电子与信息学报, 2018, 40(4):928-935.
