基于CUDA的三维成像载荷快速融合处理方法研究
发布时间：2018-11-01 14:05
【摘要】：三维成像载荷是一种能够同时采集激光雷达数据以及CCD影像,并快速融合生成三维立体影像的新型载荷,其所获取的激光点云具有规则格网分布的特点,适合采用并行的方式进行处理。结合计算统一设备架构(CUDA)对三维成像载荷快速融合处理算法核心部分进行了针对性的并行方案设计,实现了三维影像的高程值内插与平面坐标解算的快速处理,与基于CPU的串行执行模式相比,基于CUDA的三维成像载荷快速融合处理方法可显著提高计算效率,并行计算加速比可达约12倍,可支持成像区域三维影像的实时生成。
[Abstract]:Three-dimensional imaging load is a new type of load which can simultaneously collect lidar data and CCD images and quickly fuse to generate three-dimensional images. The laser point cloud acquired by the payload has the characteristics of regular grid distribution. It is suitable for parallel processing. Combined with the unified computing equipment architecture (CUDA), a parallel scheme is designed for the fast fusion algorithm of 3D imaging load, which realizes the fast processing of elevation interpolation and plane coordinate solution of 3D image. Compared with the serial execution mode based on CPU, the fast fusion method of 3D imaging load based on CUDA can significantly improve the computational efficiency, and the speedup of parallel computation can reach about 12 times, which can support the real-time generation of 3D images in the imaging region.
【作者单位】： 中国科学院光电研究院中国科学院定量遥感信息技术重点实验室;中国科学院大学;
【基金】：国家863计划项目“主被动一体化迅捷多维成像技术”(2015AA123801)
【分类号】：P237
本文编号：2304136
[Abstract]:Three-dimensional imaging load is a new type of load which can simultaneously collect lidar data and CCD images and quickly fuse to generate three-dimensional images. The laser point cloud acquired by the payload has the characteristics of regular grid distribution. It is suitable for parallel processing. Combined with the unified computing equipment architecture (CUDA), a parallel scheme is designed for the fast fusion algorithm of 3D imaging load, which realizes the fast processing of elevation interpolation and plane coordinate solution of 3D image. Compared with the serial execution mode based on CPU, the fast fusion method of 3D imaging load based on CUDA can significantly improve the computational efficiency, and the speedup of parallel computation can reach about 12 times, which can support the real-time generation of 3D images in the imaging region.
【作者单位】： 中国科学院光电研究院中国科学院定量遥感信息技术重点实验室;中国科学院大学;
【基金】：国家863计划项目“主被动一体化迅捷多维成像技术”(2015AA123801)
【分类号】：P237
【相似文献】
相关期刊论文 前2条
1 李树楷,刘彤,尤红建;机载三维成像系统[J];地球信息科学;2000年01期
2 ;[J];;年期
，本文编号：2304136
本文链接：https://www.wllwen.com/kejilunwen/dizhicehuilunwen/2304136.html
