Efficient and photorealistic polygon-based CGH through unified linear mapping

This paper presents an efficient and photorealistic polygon-based rendering algorithm for computer-generated holography (CGH) based on a unified linear mapping. Conventional fast Fourier transform (FFT)-based algorithms suffer from computational redundancy and fidelity loss due to the decoupled rasterization pipeline and the spatial-domain interpolation required to resolve the geometric shape mismatch in texture mapping. To address these limitations, we propose a unified approach comprising three related techniques. First, by introducing a judgment coordinate system, we implement a unified smooth shading rasterization strategy that synchronizes the point-in-triangle judgment with intensity interpolation, thereby eliminating redundant calculation loops. Second, to address the texture mapping bottleneck, we propose a direct spectral texture-mapping method. By establishing a linear mapping between the texture coordinate system and the parallel local coordinate system, this method maps the texture spectrum directly to the angular spectrum, thereby eliminating the spatial-domain interpolation stage and reducing the conventional two-stage interpolation to a single spectral-domain interpolation. Third, to accurately restore the concave-convex 3D perception of object surfaces, the framework integrates micro-surface normal textures with negligible computational overhead. Experimental results demonstrate that the proposed algorithm reduces rendering-stage computational time by about 80% and end-to-end model generation time by about 31% compared with the traditional smooth shading algorithm, while significantly enhancing the visual details and photorealism of the reconstructed holographic images.