Gyeongsu Jin, Seung‐Yeol Lee
Computer-generated holography (CGH) is a 3D imaging technique that faithfully reproduces visual perception cues for humans. Recently, various applications have been developed by integrating cascaded phase patterns with CGH. However, numerous CGH studies employing cascaded structures have been limited to the reconstruction of 2D planar images. In this paper, we propose a cascaded structure capable of reconstructing 3D objects through a stochastic gradient algorithm. To achieve high-quality reconstruction, we adopt a non-hogel-based light-field to CGH synthesis method, which enables holograms with high resolution in both spatial and angular domains. We validate the effectiveness of the proposed model through applications such as encryption/decryption and hologram channel multiplexing.