optics

To address the requirement for high-resolution, low-distortion imaging in long-range military target detection, this study proposes a cooperative optimization method for relay lenses tailored to optical links, achieving simultaneous low aberration and high resolution across a wide field of view (FOV). Leveraging ZEMAX, we perform comprehensive optical design and simulation, evaluating modulation transfer function (MTF), spot diagrams, Seidel aberrations, field curvature, and distortion. The optimized relay lens achieves an MTF > 0.6 at 50 lp/mm across the entire FOV and geometric distortion < 0.8%, significantly outperforming conventional separately optimized designs. This approach overcomes the fundamental trade-off between large FOV and high resolution in military relay systems, enabling robust, real-time target acquisition and tracking. The work provides a reliable optical foundation for next-generation high-precision electro-optical targeting systems.

Optical multi-lens system alignment accuracy critically affects imaging performance; however, conventional alignment methods relying on specialized metrology equipment suffer from low efficiency and poor scalability. This paper proposes a marker-free, purely optical misalignment diagnosis method based on deep learning. It employs a dual-modality input—integrating ray-tracing–generated spot patterns with physically simulated grayscale images—and introduces two complementary inverse neural network architectures to jointly model 5-DOF (for six-mirror systems) or 4-DOF (for two-mirror systems) misalignment errors. The approach requires no prior structural parameters or mechanical sensors, enabling end-to-end error prediction from a single optical measurement. Experiments demonstrate high accuracy: mean absolute error (MAE) of 0.031 mm for lateral translation and 0.011° for tilt in the six-mirror configuration. Both two- and six-mirror systems confirm strong robustness and cross-configuration generalization capability, significantly enhancing automation and efficiency in quality inspection for precision imaging systems.

Existing end-to-end computational imaging optimization methods are computationally prohibitive for jointly modeling diffraction and aberrations in compound optical systems, often neglecting wave-optical effects or off-axis aberrations—thereby compromising design robustness. This work introduces a differentiable wave-optical simulator that, for the first time, enables efficient and accurate joint modeling of diffraction and aberrations within an end-to-end framework, facilitating co-optimization of optical hardware and downstream reconstruction or classification algorithms. Our method leverages differentiable wavefront propagation, physics-driven light-field simulation, and gradient-based parameter updates via backpropagation. Experiments demonstrate that designs optimized solely under geometric optics degrade by 12–28% under realistic wave-optical conditions; in contrast, our co-designed systems achieve significantly enhanced robustness and generalization across both reconstruction and classification tasks.

To address the design-to-manufacturing performance gap in deep optical systems caused by fabrication and assembly tolerances, this paper proposes the first end-to-end tolerance-aware joint optimization framework. Methodologically, it explicitly incorporates multiple tolerance types into the deep optical design pipeline—integrating wavefront propagation modeling, aberration parameterization, differentiable optical simulation, and stochastic tolerance sampling—and enhances robustness via physics-informed and data-driven co-optimized backpropagation. Experiments demonstrate that the framework significantly narrows the performance gap between simulation and real hardware: imaging quality stability and downstream vision algorithm robustness are substantially improved, with quantitative error reduced by over 40%. The source code and visualization results are publicly released.

Joint optimization in compound-lens computational imaging systems suffers from heavy reliance on manually initialized optical designs, hindering simultaneous achievement of global optimality and physical realizability. Method: We propose Quasi-Global Synthetic Optimization (QGSO), a novel optical design paradigm comprising two stages: (i) OptiFusion automatically discovers diverse initial optical configurations; (ii) EPJO enables multi-initialization parallel, physics-constrained end-to-end joint optimization, integrating differentiable optical modeling, neural rendering, and gradient-driven co-updating of optics and computation. Contribution/Results: QGSO significantly outperforms conventional stepwise design and existing joint optimization approaches across multiple imaging tasks—achieving substantial PSNR and SSIM gains while eliminating manual initialization bottlenecks. The implementation is open-sourced, facilitating reproducible research in intelligent, physics-informed optical design.

This work addresses the challenge of simultaneously and precisely controlling both the angular and spatial distributions of light within a single, compact optical element, thereby eliminating the need for complex multi-lens systems. The authors propose a novel single-lens design methodology based on the co-optimization of two freeform surfaces, which, for the first time, enables joint manipulation of angular and spatial light fields to achieve target irradiance distributions on two distinct receiver planes concurrently. Built upon an extended caustic design framework, the approach integrates dual-plane-constrained freeform surface modeling with numerical optimization, significantly enhancing system compactness and optical efficiency. Simulation results demonstrate that the proposed method achieves stable, high-fidelity light-field control, delivering performance comparable to that of conventional multi-component optical systems.

Existing optical lens simulation methods struggle to balance photorealism and real-time performance, often neglecting complex effects such as chromatic aberration and lens flare. This work proposes a precomputed lens transmission mapping model that, for the first time, jointly models wavelength-dependent aberrations and Fresnel transmittance within a unified framework, thereby avoiding per-wavelength fitting. By treating wavelength as an explicit input to a neural regression model, the method predicts Fresnel intensity and incorporates a binary occlusion mask to focus computation on valid rays, enabling efficient and high-fidelity simulation of static rotationally symmetric lens systems. Experiments demonstrate that the approach significantly improves accuracy over polynomial baselines while achieving an order-of-magnitude speedup compared to brute-force ray tracing, making it well-suited for lens simulation scenarios demanding both efficiency and visual realism.

This work addresses the long-standing reliance on expert intuition in optical design and the inability of conventional methods to automate the process effectively. While existing large language models possess general optical knowledge, they fail to generate physically realizable lens systems. To bridge this gap, we propose a physics-driven agent framework that integrates optical priors with the reasoning capabilities of large language models, enabling automatic generation of functional lenses without task-specific training. Our key innovations include an optical lexicographic reward mechanism—enforcing structural validity, physical feasibility, and ray-tracing accuracy—a hybrid training objective, and a DrGRPO-based policy optimization strategy. We further introduce the OptiDesignQA dataset and an end-to-end optimization pipeline. Experiments demonstrate that our approach significantly outperforms both traditional automated design algorithms and current large-model methods, successfully producing high-fidelity, physically consistent lens systems across multiple benchmarks.

This study addresses the challenge of accurately recovering incident wavefronts from a single intensity measurement by optimizing pupil design in optical systems. The work proposes the first quantitative metric for pupil asymmetry tailored to wavefront estimation and systematically investigates its impact on wavefront recoverability. Through extensive simulations and experimental validation, the authors establish a positive correlation between pupil asymmetry and wavefront reconstruction performance. The results demonstrate that highly asymmetric pupils significantly enhance estimation accuracy while effectively balancing photon throughput and robustness to noise. These findings yield a general design principle applicable to computational imaging, adaptive optics, and related fields requiring precise wavefront sensing from limited measurements.