摘要
针对双通道、高码速率的高分辨率光学遥感卫星吉林一号A星下传数据的实时快视需求,提出了一种基于CUDA和OpenGL互操作的遥感影像快视方法,采用CUDA C语言利用GPU并行地从原始码流中提取出像素数据并存入像素缓冲对象,之后不必将像素数据传送回主存,而是基于互操作接口将显存中像素缓冲对象作为纹理数据源直接递送给OpenGL纹理绘制接口完成影像快视显示,整个过程不需要任何数据移动或复制。实践表明,与由CPU完成数据提取和图像绘制相比,该方法可快速完成数据并行提取,提高影像快视效率,并释放CPU。
To the need of data real-time quick-look for high resolution optical remote sensing JL-1A satellite with the characteristics of dual-channel and high-code-rate,this article proposed an approach of remote sensing image quick-look based on the interoperability between CUDA and OpenGL. Pixel data was extracted by GPU paralelly using CUDA C language and wrote in pixel buffer objet,and later the pixel data in video memory was not sent back to the main memory,but took as texture data source and delivered to OpenGL texture API to render directly based on the interoperability interface without data movement or copy. Practice showed that this approach could accomplish the task of pixel data extraction fast,enhance the efficiency of image quick-look,and release CPU.
作者
段云龙
张致齐
方留杨
常鹏飞
张岭军
DUAN Yun-long;ZHANG Zhi-qi;FANG Liu-yang;CHANG Pengf-ei(The 271,1 Research Institute of China Electronics Technology Group Corporation,Zhengzhou 450047,China;State Key Laboratory of Information Engineering in Surveying,Mapping and Remote Sensing,Wuhan University,Wuhan 430079,China;Broadvision Engineering Consultants,Kunming 650041,China)
出处
《中国电子科学研究院学报》
北大核心
2020年第1期36-42,共7页
Journal of China Academy of Electronics and Information Technology
基金
国家自然科学基金青年基金(41601476)。
