Abstract:The Rich spatial and angular information in light field images enables accurate depth estimation, which is a crucial aspect of environmental perception. However, the abundance of light field information also leads to high computational costs and memory pressure. Typically, selectively pruning some light field information can significantly improve computational efficiency but at the expense of reduced depth estimation accuracy in the pruned model, especially in low-texture regions and occluded areas where angular diversity is reduced. In this study, we propose a lightweight disparity estimation model that balances speed and accuracy and enhances depth estimation accuracy in textureless regions. We combined cost matching methods based on absolute difference and correlation to construct cost volumes, improving both accuracy and robustness. Additionally, we developed a multi-scale disparity cost fusion architecture, employing 3D convolutions and a UNet-like structure to handle matching costs at different depth scales. This method effectively integrates information across scales, utilizing the UNet structure for efficient fusion and completion of cost volumes, thus yielding more precise depth maps. Extensive testing shows that our method achieves computational efficiency on par with the most efficient existing methods, yet with double the accuracy. Moreover, our approach achieves comparable accuracy to the current highest-accuracy methods but with an order of magnitude improvement in computational performance.

Depth Estimation Algorithm for Light Field Data by Epipolar Image Analysis and Region Interpolation

A Learning-Based Method Using Epipolar Geometry for Light Field Depth Estimation

A Light Field Depth Estimation Algorithm by Edge Disparity Extraction

Light Field Depth Estimation Exploiting Linear Structure in EPI

Light-field-depth-estimation Network Based on Epipolar Geometry and Image Segmentation.

Light Field Depth Estimation via Epipolar Plane Image Analysis and Locally Linear Embedding

Three Dimensional Surface Reconstruction Based on Light Field Imaging

Three Dimensional Reconstruction Using a Lenslet Light Field Camera

Depth Estimation by Analyzing Intensity Distribution for Light-Field Cameras.

EPI-Neighborhood Distribution Based Light Field Depth Estimation

Iterative Occlusion-Aware Light Field Depth Estimation using 4D Geometrical Cues

Defocusing Response Function Optimization in Depth Estimation of Boundary Complex Light Field Image

Depth Map Estimation with 4D Light Fields Using Confocal Stereo

Depth Estimation from Light Field Using Graph-Based Structure-Aware Analysis.

Depth Fused from Intensity Range and Blur Estimation for Light-Field Cameras

Depth Extraction of Low-Texture Region Using A Light Field Camera

Efficiency–Accuracy Trade-Off in Light Field Estimation with Cost Volume Construction and Aggregation

Fast Response Aggregation for Depth Estimation Using Light Field Camera

Depth Extraction from a Light Field Camera Using Weighted Median Filtering.

Occlusion-Model Guided Anti-Occlusion Depth Estimation in Light Field

Anti-Noise Light Field Depth Estimation Using Inline Occlusion Handling