Efficient light field acquisition for integral imaging with adaptive viewport optimization

Abstract

Light field displays reconstruct 3D scenes through integral imaging. However, inefficient light ray acquisition degrades the visual experience, while the fixed position of the exit pupil limits viewer mobility. In this paper, we propose a novel light field acquisition method employing parallax mapping techniques, coupled with adaptive viewport optimization based on eye tracking data. The parallax mapping relationship between camera pose variation and pixel offset facilitates partitioned rendering in integral image generation, and layer stacking is conducted to incorporate multiple depth cues. An eye tracking module is integrated to detect binocular landmarks and adaptively optimize screen segmentation, thus shifting the viewport to accommodate eye positions. Experimental results demonstrate correct refocusing cues and occlusion relationships, showing robustness in displaying complex scenes. The viewing zone has been expanded by at least twice, and the dynamic display performance meets real-time visual requirements.