Quality enhancement of computer-generated hologram without contents-dependent optimization by considering aberrations in the optical system

In this paper, we propose a content-independent method for enhancing the quality of computer-generated holograms (CGHs) by incorporating optical aberrations. Traditional approaches often rely on simulation-based optimization or deep learning to reduce speckle noise, but these methods can be computationally expensive, object-dependent, and sensitive to discrepancies between simulations and real-world optical setups. In contrast, the proposed method optimizes a multi-layer hologram synthesis function to achieve object-independent quality enhancement. By training the system only for a uniform amplitude target image and applying the model to arbitrary images, we significantly reduce computational time. Moreover, optimization accounts for environmental factors in the actual optical system through a camera feedback mechanism. The effectiveness of the proposed approach is demonstrated through a series of optical experiments.