A unified angular MTF evaluation framework for sparse-aperture optical system optimization

Technical Summary: A Unified Angular MTF Evaluation Framework for Sparse-Aperture Optical System Optimization

Problem Statement
Optical sparse aperture (OSA) imaging systems utilize multiple small sub-apertures to achieve the diffraction-limited resolution of a large monolithic aperture while significantly reducing volume, weight, and cost. While various aperture configurations exist—such as traditional annular, Golay-6, and Y-shaped arrays—existing research lacks a systematic performance analysis of these typical structures under unified evaluation criteria. Most prior studies rely on inconsistent or single evaluation metrics, often focusing on specific directions or frequency points, which fails to comprehensively capture the directional uniformity and mid-frequency response capabilities essential for high-quality imaging.

Methodology
The paper proposes a unified evaluation framework based on four Modulation Transfer Function (MTF) metrics designed to assess both overall average performance and worst-case directional stability. The methodology involves:

1. Unified MTF Metrics:

   • Angularly Averaged MTF Area ($A_{avg}$): Measures the overall frequency transmission capability across the normalized frequency range $[0, 1]$.
   • Mid-Frequency Angular Mean ($M_{avg}$): Calculates the mean MTF value within the critical mid-frequency band (0.2–0.8), reflecting the retention of main structure and texture information.
   • Worst-Direction MTF Area ($A_{worst}$): Evaluates the system's performance in its most unfavorable angular direction.
   • Worst-Direction Mid-Frequency Mean ($M_{worst}$): Assesses detail retention in the worst-case direction.

2. Sampling Strategy:

   • The framework employs a polar coordinate sampling approach with 360 angular samples and 600 mid-frequency samples (0.2–0.8).
   • Comparisons are conducted under fixed geometric constraints: identical circumscribed circle radius (ensuring the same cutoff frequency) and identical filling factors (approx. 23.5% for the initial comparison).

3. Optimization Process:

   • Re-evaluation: Traditional structures (Annular, Golay-6, Y-shaped) are benchmarked against the four metrics.
   • Geometric Optimization: The Golay-6 structure is optimized by fixing the three outer sub-apertures and allowing the three inner sub-apertures to undergo radial translation and synchronous rotation. A Particle Swarm Optimization (PSO) algorithm is used to maximize the four metrics.
   • Size Optimization: Based on the resulting optimal geometry (Y-shaped), a second optimization phase adjusts the radii of the inner and outer sub-apertures while maintaining the Y-shape and fixed filling factor. A phased swarm intelligence search strategy (coarse global search followed by fine local refinement) is employed.

Key Results

• Traditional Structure Comparison: Under unified constraints, the Y-shaped array outperformed both the Golay-6 and Annular structures across all four metrics. The Annular structure exhibited unstable mid-frequency responses and local zeros in certain directions, leading to the poorest performance.
• Geometric Convergence: Optimizing the Golay-6 structure caused the configuration to naturally converge toward a Y-shaped (three-arm) geometry. The optimal parameters were found to be a radial translation of $d=0.1398$ and a rotation angle of $\theta=79.14^\circ$. This confirmed the inherent geometric advantage of the three-arm structure in balancing frequency response.
• Aperture Size Optimization: Further optimization of the Y-shaped structure by redistributing the radii of inner and outer sub-apertures yielded the best overall performance. The final optimized structure achieved:
  • $A_{avg} = 0.2138$
  • $M_{avg} = 0.1471$
  • $A_{worst} = 0.1914$
  • $M_{worst} = 0.1198$
• Image Quality Verification: Simulations using 100 airplane images from the NWPU-RESISC45 dataset demonstrated the practical impact of the frequency-domain improvements. The optimized structure achieved an average PSNR of 28.33 dB and an average SSIM of 0.861.
  • This represents an improvement of 0.53 dB in PSNR and 0.019 in SSIM over the best-performing traditional Y-shaped configuration.
  • Compared to the traditional Annular structure, the improvements were 1.61 dB in PSNR and 0.046 in SSIM.

Significance and Claims
The paper claims that the proposed unified angular MTF evaluation framework provides a quantitative basis for overcoming the one-sidedness of single-direction or single-frequency-point evaluations. By systematically analyzing and optimizing aperture configurations, the study demonstrates that:

1. The Y-shaped three-arm geometry is the optimal basic configuration for six-aperture sparse systems under the proposed metrics.
2. Optimization should not be limited to aperture positioning; the redistribution of aperture sizes (inner vs. outer radii) is critical for enhancing frequency-domain coverage uniformity and mid-frequency transmission.
3. The proposed optimization method effectively translates frequency-domain improvements into tangible image quality enhancements, as evidenced by the superior PSNR and SSIM scores in image degradation simulations.

The authors conclude that this framework and the resulting optimized structure offer a robust solution for improving the imaging quality of sparse-aperture optical systems without increasing system volume or light transmission area.