On-Orbit Data Process Based On FPGA

At present, the data processing capacity of on-board computer system is insufficient, especially when using in processing remote sensing image. We usually acquire remote sensing image, and then we compress them and send them to the ground station for processing. The image resolution is very high, while the downstream bandwidth is limited. So we need to enhance the data processing capacity of on-board computer system, giving the remote sensing satellite more autonomy and flexibility. We can use multiple processors to accomplish a task which is called multi-core processing technology. In this way, we can improve efficiency of the task. Xilinx FPGA has a high logic capacity and anti-radiation capability, we can using it to process on-orbit data. Because of the high speed of FPGA, we can apply it to on-orbit data process.In the paper, we propose a heterogeneous multicore streamline method based on FPGA to extracting feature points of remote sensing images. The algorithm we use is SURF, which is scale and rotation invariant. We implement heterogeneous multicore system on Xilinx Virtex-4chip and verify the validity of the system. The main contents of this paper includes, the research of system on chip and multi-core system, scale and rotation invariant feature point extraction SURF algorithm and its implementation, the overall design and implementation of heterogeneous multicore on-orbit data processing system, implementation of demonstration platform, experiment and data analysis.This paper is expected to reach these goals, we propose a heterogeneous multicore streamline method based on a comprehensive study of existing on-board computer processing capabilities and multi-core processing technology. By using heterogeneous multi-core data processing technology, we solve the problem of insufficient data processing capabilities. We study SURF algorithm and implement it on heterogeneous multicore system. In this way, we form pipeline among multicore system. We conducted experiments and data analysis, testing the efficiency of heterogeneous multicore systems.