End-to-end infrared imaging system design with diffractive optical element

The hybrid refractive-diffractive optics system exhibits strong capabilities in achromatic and non-thermalized design, as well as information encoding. This paper introduces an innovative end-to-end design scheme for refractive-diffractive hybrid imaging optical systems, which optimizes both optical and neural network parameters simultaneously. An all-ray differentiable ray-tracing model is proposed to integrate lens and diffractive optical elements into a unified design framework, thereby maintaining precision by avoiding wave-to-ray conversion losses. The neural network is constructed based on the imaging characteristics of infrared hybrid optical systems and enhances aberration correction. The proposed method is applied to the design of a single-lens short-wave infrared imaging system, outperforming traditional discrete designs and demonstrating significant potential for infrared optical system applications.