POS数据支持下的面阵航测影像GPU几何纠正方法  

A GPU Geometric Correction Method of Area-array Aerial Image Based on POS Data

作　　者：马远征 方勇[2] 

机构地区：[1]长安大学,陕西西安710061 [2]西安测绘研究所,陕西西安710054

出　　处：《测绘科学与工程》2017年第6期53-58,共6页Geomatics Science and Engineering

摘　　要：针对航空摄影数据高效处理需求，研究基于GPU架构、POS数据支持下的面阵航测影像快速几何纠正方法。根据POS提供的影像外方位元素，建立无控制条件下影像严密对地定位关系；基于CUDA编程框架，设计面阵影像GPU几何纠正算法并编程实现，完成面阵航测影像的快速几何纠正。利用某新型国产面阵航测相机进行试验，结果表明：POS支持下无控制定位精度优于0．25个像素，单张影像纠正时间优于400ms，效率较CPU算法提升20倍以上。In order to meet the efficiency requirement of aerial photography data processing, a correction method for air- borne photogrammetrie image based on GPU architecture and POS data is proposed in this paper. According to the image exterior elements provided by POS, a strict image positioning relation without ground control point is established. Based on CUDA pro- gramming framework, a GPU geometric correction algorithm for area-array aerial image is designed and programmed, and the fast geometric correction of area-array aerial images is completed. The efficiency and accuracy of the proposed method and algorithm are verified by experiment using a new domestic format array camera. The result shows that with the support of POS, the positio- ning accuracy without control points is better than 0.25 pixels, the single image geometric correction time is better than 400 milli- seconds, and the efficiency improves by 20 times compared with that of CPU.

关 键 词：快速几何纠正 面阵航测相机 定位定向系统 计算统一设备架构 

分 类 号：P231[天文地球—摄影测量与遥感]