研究目的
To study the feasibility of onboard SAR image focusing using FPGA-based systems to reduce data storage and transmission requirements, enabling near real-time processing for spaceborne SAR systems.
研究成果
The FLIP2 demonstrates feasibility for near real-time onboard SAR image focusing with acceptable image quality (e.g., RMSE <3% for amplitude, high coherence for phase) and faster processing than CPU-based methods, though hardware resource usage and speed need optimization for full real-time application in future SAR systems.
研究不足
The FLIP2 is an evaluation model using commercial off-the-shelf products and not designed for actual onboard use; it requires multiple FPGAs due to resource constraints (e.g., BRAM shortage); processing speed is 1/4 of real-time needs for ALOS-2/PALSAR-2; image quality degradation may occur in darker areas.
1：Experimental Design and Method Selection:
The study uses an FPGA-based hardware model (FLIP2) implementing the Chirp-Scaling Algorithm (CSA) for SAR image focusing, compared to a CPU-based simulation using MATLAB with the same algorithm.
2：Sample Selection and Data Sources:
Raw signal data from ALOS-2/PALSAR-2, specifically 3m-resolution images of Miyake-jima island, are used.
3：List of Experimental Equipment and Materials:
FLIP2 evaluation model with Xilinx FPGAs (e.g., XCKU060), DRAMs, FPGA evaluation boards, serial ATA cables, Host PC with USB interfaces, and Intel Xeon E5-2697 v4 CPU for simulation.
4：Experimental Procedures and Operational Workflow:
Raw SAR data is sent from Host PC to FLIP2 via USB; FLIP2 processes data using CSA with 32-bit block floating point quantization; output SLC images are compared to simulation images for quality assessment.
5：Data Analysis Methods:
Quantitative evaluation using RMSE, NRMSE, PSNR, dynamic range, and interferometric coherence calculations on selected image areas (land surface, sea surface, high-intensity pixels, shadow).
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
