Global Search Optics: Automatically Exploring Optimal Solutions to Compact Computational Imaging Systems

Yao Gao,Qi Jiang,Shaohua Gao,Lei Sun,Kailun Yang,Kaiwei Wang
2024-04-30
Abstract:The popularity of mobile vision creates a demand for advanced compact computational imaging systems, which call for the development of both a lightweight optical system and an effective image reconstruction model. Recently, joint design pipelines come to the research forefront, where the two significant components are simultaneously optimized via data-driven learning to realize the optimal system design. However, the effectiveness of these designs largely depends on the initial setup of the optical system, complicated by a non-convex solution space that impedes reaching a globally optimal solution. In this work, we present Global Search Optics (GSO) to automatically design compact computational imaging systems through two parts: (i) Fused Optimization Method for Automatic Optical Design (OptiFusion), which searches for diverse initial optical systems under certain design specifications; and (ii) Efficient Physic-aware Joint Optimization (EPJO), which conducts parallel joint optimization of initial optical systems and image reconstruction networks with the consideration of physical constraints, culminating in the selection of the optimal solution. Extensive experimental results on the design of three-piece (3P) sphere computational imaging systems illustrate that the GSO serves as a transformative end-to-end lens design paradigm for superior global optimal structure searching ability, which provides compact computational imaging systems with higher imaging quality compared to traditional methods. The source code will be made publicly available at
Image and Video Processing,Computer Vision and Pattern Recognition,Robotics,Optics
What problem does this paper attempt to address?
The paper focuses on the problems encountered in designing compact computational imaging systems. Traditional optical system design and image reconstruction models are designed separately, but in recent years, combined design methods have emerged, which optimize both components through data-driven learning. However, the effectiveness of these designs largely depends on the initial optical system settings, and the non-convex solution space makes finding the global optimal solution difficult. The paper proposes the "Global Search Optics" (GSO) framework, aiming to automatically design compact computational imaging systems. GSO consists of two parts: 1) Opti-Fusion, an automatic optical design fusion optimization method, which finds diverse initial optical systems given the design specifications; 2) Efficient Physics-aware Joint Optimization (EPJO), which performs parallel joint optimization of initial optical systems and image reconstruction networks, taking into account physical constraints to select the optimal solution. Experimental results show that GSO can provide better global optimal structure search capability in designing three-piece (3P) spherical computational imaging systems, improving the imaging quality of compact computational imaging systems. In short, this paper attempts to address the effective automatic search for the global optimal solution in designing compact computational imaging systems, in order to achieve joint optimization of optical system and image reconstruction models, thereby improving system performance.