Light Field Depth Estimation with Information Entropy-Based Pre-Computed Occlusion Masks

Light field cameras capture synchronized spatial-angular information of lights, offering a new paradigm for 3D visual perception. Depth estimation, as a fundamental task in light field analysis, underpins the critical applications such as 3D reconstruction and visual odometry. However, occlusion-induced estimation errors remain a persistent challenge. This paper proposes an occlusion-aware depth estimation framework featuring two novel modules: an entropy-based occlusion mask pre-computation method and a viewpoint screening-driven depth estimation algorithm. In the investigation, first, the light field occlusion in the polar plane diagram is modeled by analyzing the information entropy of micro-images array, and a local entropy extremum-based occlusion mask pre-computation approach is constructed, thus overcoming the limitations of conventional techniques in characterizing occluded regions. Subsequently, viewpoint screening is employed to eliminate the interference from occluded perspectives, thus effectively reducing estimation errors and decreasing the proportion of disparity outliers exceeding 0.03 pixels compared to other methods. The core contribution of this paper lies in establishing a theoretical connection between information entropy and light field occlusion, enabling a robust occlusion-aware depth estimation framework based on the information entropy of micro-images array. Experimental results on light field benchmark dataset demonstrate that the proposed method achieves superior performance in mean absolute error and 25th error percentile. Comparative ablation studies confirm the efficacy of the entropy-driven occlusion mask, thus highlighting the critical role of information entropy theory in the framework of light field depth estimation.