J4 ›› 2014, Vol. 41 ›› Issue (3): 152-156.doi: 10.3969/j.issn.1001-2400.2014.03.022

• 研究论文 • 上一篇    下一篇



  1. (1. 西安电子科技大学 物理与光电工程学院,陕西 西安  710071;
    2. 中国科学院 光电研究院,北京  100094)
  • 收稿日期:2013-03-01 出版日期:2014-06-20 发布日期:2014-07-10
  • 通讯作者: 赵宝玮
  • 作者简介:赵宝玮(1983-),男,西安电子科技大学博士研究生,E-mail: zbw8200980@163.com.
  • 基金资助:


Parallel RX algorithm implementation based on the FPGA and multi-DSP system

ZHAO Baowei1;XIANGLI Bin1,2;Lü Qunbo2;ZHANG Guifeng2;ZENG Xiaoru2;GUO Wenji2   

  1. (1. School of Physics and Optoelectronic Engineering, Xidian Univ., Xi'an  710071, China;
    2. Academy of Opto-electronics, Chinese Academy of Sciences, Beijing  100094, China)
  • Received:2013-03-01 Online:2014-06-20 Published:2014-07-10
  • Contact: ZHAO Baowei



关键词: 并行处理, 高光谱图像, 目标探测, RX算法, FPGA, 多数字信号处理


Hyperspectral imagery is hard to apply for real-time detection of abnormal targets due to its huge amount of computation. This paper proposes a fast processing method by parallelizing the RX detection algorithm on a hardware platform based on FPGA and multi-DSP. According to the principle and characteristics of the RX algorithm, computer simulation is completed to confirm the feasibility and characteristics of parallelization. On this basis, the parallel processing RX algorithm is implemented on this platform, where FPGA implements the SVD for dimensionality reduction in hyperspectral imagery to reduce the amount of data processing and transmission, and multi-DSP speeds up the process of abnormal target detection by parallelizing the RX algorithm. For hyperspectral data including 64 bands of images at size 280×800, the proposed system only took 4.86s to obtain the detection result. The demonstrated ability can satisfy the requirement for online processing and fast result acquisition in applications of hyperspectral remote sensing.

Key words: parallel processing, hyperspectral imagery, target detection, RX algorithm, FPGA, multi-digital signal processing


  • TP751.1