• •    

信息存储技术学术会议2016+58+基于多核DSP的星载并行遥感图像压缩系统设计与实现

唐国斐1,周海芳2,谭庆平3   

  1. 1. 国防科学技术大学
    2. 国防科技大学计算机学院
    3. 国防科学技术大学计算机学院
  • 收稿日期:2016-12-09 修回日期:2016-12-21 发布日期:2016-12-21
  • 通讯作者: 唐国斐

Design And Implementation Of Spaceborne Parallel Remote Sensing Image Compression System Based On Multicore DSP

  • Received:2016-12-09 Revised:2016-12-21 Online:2016-12-21
  • Contact: GUOFEI TANG

摘要: 摘 要: 随着星载遥感技术的不断发展,产生的遥感数据也变得日渐庞大。目前有限的通信传输带宽远不能满足遥感图像数据传输的需求。因此研究面向星载应用的图像压缩技术对空间应用技术的发展有着十分重要的意义。采用传统单核DSP(Digital Signal Processor,数字信号处理器)难以满足性能需求,而采用FPGA(Field-Programmable Gate Array,现场可编程门阵列)则难以满足功耗需求,近年来随着硬件技术发展,多核DSP技术已经成熟,且在弹载场景已有比较成熟的多核DSP图像压缩解决方案,可供星载应用参考。课题来源于实际星载项目,目的是基于多核DSP,即TI公司的C6678多核浮点DSP平台,构建一个并行图像压缩系统,并充分利用多核DSP的硬件资源。并行模型的设计以及DSP内部存储的规划是研究的重点。考虑星载遥感图像压缩对压缩质量、压缩速度等多方面指标都有着较高的要求,系统采用JPEG2000标准进行图像压缩,并且采用了主核负责外部通信与内部任务分配,从核执行JPEG2000图像压缩的设计方案。 测试结果表明,该系统运行稳定可靠,且整体压缩性能优秀,能够满足对星载遥感图像压缩系统的性能要求。

关键词: 关键词: 图像压缩, JPEG2000, 并行, 多核, DSP, C6678

Abstract: Abstract: With the continuous development of spaceborne remote sensing technology, the remote sensing data has become increasingly large. At present, the limited bandwidth of communication transmission can not meet the demand of remote sensing image data transmission. Therefore, the research of image compression technology for spaceborne applications is of great significance to the development of space application technology. It is difficult to meet the performance requirements by adopting traditional single-core DSP (Digital Signal Processor), and it is difficult to meet the power demand by adopting Field-Programmable Gate Array (FPGA). In recent years, with the development of hardware technology , Multi-core DSP technology has matured, and the scene in the bomb has been more mature multi-core DSP image compression solution for space-borne application reference. Based on the multi-core DSP, TI's C6678 multi-core floating-point DSP platform, this paper constructs a parallel image compression system and makes full use of the hardware resources of multi-core DSP. The design of parallel model and the planning of DSP internal storage are the focus of the research. Considering the compression of spaceborne remote sensing image has a higher demand for compression quality, compression speed and other indicators, the system uses the JPEG2000 standard as the image compression standard, Using the main core responsible for external communication and internal task allocation, slave core implementation JPEG2000 image compression Of the design. Test results show that the system is stable and reliable, and the overall compression performance is excellent, to meet the satellite remote sensing image compression system performance requirements.

Key words: Keywords: image compression, JPEG2000, parallel, Multicore, DSP(Digital Signal Processing), C6678

中图分类号: