摘要
当前,如何实现高效的主被动三维影像生成技术是遥感应用中的一个研究热点。由于激光雷达-光学CCD的三维影像生成算法具有数据量大、处理复杂的特点,为实现其高效实时的应用需求,本文利用TMS320C6678多核DSP开展了并行处理技术方面的研究。并针对两种主流并行模型(主从模型和数据流模型),分别进行了三维影像生成算法的并行设计。最后,依托开发平台进行测试分析,总结论证了两种主流模型的设计要点与适用情况,为多核DSP在实际应用中的资源高效利用提供了思路。
At present, the question of how to achieve efficient active-passive three-dimensional （3D） image generation technology is a research hotspot in remote sensing applications. The technology involves a large quantity of data and complex processing. In order to fulfill the high efficiency and real-time requirements of practical applications, this study developed parallel processing technology based on the TMS320C6678 multicore DSE The parallel design of the 3D image generation algorithm was carried out based on the two main parallel models（the Master-Slave and Data Flow models）. Finally, relying on the measurement and analysis of the development platform, the characteristics and applied scene of those two main models were summarized. These results can provide ideas for the efficient utilization of multicore DSP in practical applications.
出处
《红外技术》
CSCD
北大核心
2017年第11期983-989,共7页
Infrared Technology
基金
国家高技术研究发展计划(863计划)资助项目(2015AA123801)
中国科学院光电研究院创新项目(Y50B12A14Y)
关键词
多核DSP
三维影像生成
主从模型
数据流模型
multicore DSP, three-dimensional image generation technology, Master/Slave model, Data Flow model
