SAR Autofocusing via Phase Error Function Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The Phase-Gradient Algorithm for autofocusing SAR images is inaccurate when point targets are not available and positional errors of reflectors cause coordinate errors, making real-time processing challenging due to high computational requirements.
Innovation Solution
A method using the range Doppler algorithm for autofocusing SAR raw data, estimating azimuth signals from spatially distributed point targets, calculating local and entire phase error functions to correct azimuth signals, and employing a refined algorithm that selects reflectors based on energy content and ranking criteria, with deramping and Fourier transformations to improve phase error estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the Phase-Gradient Algorithm uses strong reflectors consisting of many joint point targets to estimate the phase error function, then the algorithm can operate without requiring precise point targets, but the estimation accuracy deteriorates and coordinate errors increase
Solution Approach 1:
The patent segments the reflector into multiple individual point targets and processes their azimuth signals separately through deramping and Fourier transformation. Instead of treating the reflector as a single strong target, the method divides it into constituent point targets, estimates phase error for each, and then combines them. This segmentation allows precise phase error estimation from individual point targets while maintaining the ability to operate without requiring pre-identified precise point targets in the scene.
2Measurement precision
If the Phase-Gradient Algorithm is applied iteratively to improve image quality, then the estimation accuracy improves, but the computational complexity increases making real-time processing difficult
Solution Approach 1:
The patent performs preliminary range compression and motion compensation using INS and GPS data before applying the phase error estimation. By pre-processing the data to correct obvious motion errors and compress the range dimension, the algorithm reduces the computational burden of subsequent phase error estimation. This preliminary action enables accurate single-pass phase error estimation without requiring multiple iterative applications, thus achieving real-time processing capability.
3Ease of operation
If positional errors of reflectors are used in the Phase-Gradient Algorithm, then the processing can proceed without precise target locations, but azimuth dependent coordinate errors are introduced in the SAR image
Solution Approach 1:
The patent uses the estimated phase error function as feedback to correct the azimuth signals before azimuth compression. By estimating the phase error from the raw data itself and then applying this correction, the system creates a feedback loop that compensates for positional errors of reflectors. This feedback mechanism allows the algorithm to proceed with imprecise reflector locations while correcting the resulting coordinate errors in the final SAR image, maintaining both ease of operation and manufacturing precision.
Data Source
Figure 1
Figure 2
AI summary
An algorithm for estimating the phase error function of SAR raw data is described, which exploits reflectors consisting of several neighbouring point targets. In a first step the azimuth signal of a reflector, which appears as highlight in its near environment, is extracted and used to estimate the azimuth signal of a single point target. In the second step a local phase error function is determined for each extracted reflector from the azimuth signal of the point target. The entire phase error function of a SAR image strip is then constructed from the local phase error functions using a weighted superposition technique. Compared to the Phase-Gradient Algorithm, the quality criterion for evaluating the sharpness of SAR images is improved by 15%. The standard deviation of coordinates errors of objects in the SAR image is reduced from 94 to 3.8 pel.