Optical degradation correction of manufacturing-perturbed glass-plastic hybrid lens systems via a joint hardware-software optimization framework.

Glass-plastic hybrid lens systems are increasingly critical in various optical applications due to their unique advantages and growing demands. Due to limitations in manufacturing processes and costs, the yield rate of glass-plastic hybrid lens systems in mass production struggles to match that of mature all-plastic ones. In this work, we propose a pioneering joint hardware-software optimization framework designed for correcting optical degradation in manufacturing-perturbed glass-plastic hybrid lens systems. Our framework begins with the establishment of a differentiable imaging simulation system that is capable of simulating various manufacturing errors. This system facilitates the preliminary estimation of manufacturing deviations across individual lenses without precise measurements. Subsequently, from the perspective of the hardware assembly process, we integrate active alignment of the glass aspherical lens to mitigate degradation caused by these deviations. Moreover, we introduce a novel and lightweight degradation correction network as post-processing software to address residual optical degradation without fine-tuning for each manufacturing-perturbed lens system, significantly reducing deployment costs for mobile devices. Extensive experiments validate the efficacy of our joint hardware-software optimization framework, showing substantial improvements in imaging quality and enhanced yield rates in mass production. Overall, our framework establishes a new paradigm for optical degradation correction in glass-plastic hybrid lens systems by synergizing the front-end lens assembly process with the back-end degradation correction method. This new paradigm represents an inaugural effort within the optical engineering domain.