Autofocus for SAR Images via Multi-Dimensional Entropy Minimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current autofocus techniques for synthetic aperture radar images are limited in estimating and correcting high-order phase errors, leading to image smearing and degradation, especially when phase errors vary spatially or when targets are closely located, resulting in reduced signal-to-noise ratio and inaccurate phase continuity between image blocks.
Innovation Solution
A method that optimizes an array of coefficients for a polynomial representing phase correction to minimize entropy values in the image, using an entropy calculation to identify and correct phase errors, thereby focusing the image through inverse fast Fourier transforms and phase corrections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If map drift and phase difference methods are used for autofocus, then the method is less sensitive to scene contents, but the maximal order of phase error that can be estimated is limited to around five
Solution Approach 1:
The patent transitions from one-dimensional search algorithms to multi-dimensional optimization by optimizing an array of coefficients simultaneously. This allows estimation of higher-order phase errors (beyond fifth order) by treating the coefficient optimization as a multi-dimensional problem rather than sequential one-dimensional adjustments, thereby resolving the limitation on phase error estimation order while maintaining robustness to scene contents.
Solution Approach 2:
The patent changes the optimization approach from adjusting individual parameters sequentially to optimizing an array of coefficients together using entropy minimization. This parameter transformation enables the system to estimate higher-order phase errors by representing phase correction as a polynomial with multiple coefficients that are optimized simultaneously, overcoming the previous limitation of maximal order five estimation.
2Measurement precision
If full collection array data is divided into small data blocks for processing, then the processing can handle higher order phase error, but the signal-to-noise ratio is reduced
Solution Approach 1:
The patent merges the processing of multiple data blocks by optimizing an array of coefficients that represent phase correction across the entire dataset. Instead of treating each small data block independently (which reduces signal-to-noise ratio), the methodology combines information from all blocks through joint coefficient optimization, thereby maintaining high signal-to-noise ratio while enabling higher-order phase error correction.
Solution Approach 2:
The patent creates a universal optimization framework that processes multiple data blocks simultaneously using a single set of polynomial coefficients. This multi-functional approach allows the same coefficient array to correct phase errors across all data blocks, enabling higher-order phase error estimation without the need to process each block separately, thus preserving signal-to-noise ratio.
3Ease of operation
If phase gradient algorithm is used, then the method is simple and efficient, but distinguishing targets closely located in the azimuth from isolated targets cannot be perfect
Solution Approach 1:
The patent applies local quality by optimizing polynomial coefficients that represent local phase characteristics in different regions of the data. By allowing different coefficients to capture local phase variations, the method can distinguish closely located targets in the azimuth direction while maintaining algorithmic simplicity through the unified polynomial framework, overcoming the limitation of phase gradient algorithms.
4Adaptability or versatility
If small image blocks are used for phase error estimation, then the method can handle spatially varying phase error, but phase continuity between divided small image blocks is difficult to achieve
Solution Approach 1:
The patent merges the phase correction across multiple image blocks by using a unified polynomial model with a shared array of coefficients. This approach maintains phase continuity between blocks while still allowing the model to adapt to spatially varying phase errors, as the polynomial can represent global phase trends that connect local variations smoothly across block boundaries.
Data Source
AI summary
A computer implemented method, apparatus, and computer usable program code for focusing an image. In one advantageous embodiment, a method is used to focus an image. Optimization is performed to identify an array of coefficients for a polynomial representing a phase correction in a manner that minimizes an entropy value generated by an entropy calculation for the image. The array of coefficients is applied to the polynomial to obtain a desired phase correction. A phase error in the image is corrected using the desired phase correction to focus the image.