A Real-time Preprocessing Method for Mixed Heterogeneous Meteorological Remote Sensing Images in Orbit

To solve the real-time generation of level 1 data for meteorological satellites in orbit and improve user experience, this paper proposes a mixed heterogeneous meteorological satellite in orbit real-time preprocessing method. This method receives 0-level data from meteorological satellite payloads and processes it in real-time in orbit to generate 1-level positioning and calibration result data. The positioning and calibration operation of Level 1 products is a repeated operation of the load detector pixel by pixel. This method adopts the hardware acceleration idea to transplant the repeated simple addition, subtraction, multiplication, and division operations into FPGA, realizing real-time processing of remote sensing data in orbit. Before and after hardware acceleration, the processing speed of the FY-3 spectral imaging instrument payload level 1 product increased by 4.5 times, and the positioning accuracy was better than one pixel compared to the standard coastline. The hybrid heterogeneous core processor adopts ARM+FPGA architecture, which consumes 17% of FPGA resources for hardware acceleration compared to CPU processing. The time indicators can meet the real-time processing requirements of various meteorological satellite payloads.