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.

Toward Real-World Light Field Depth Estimation: A Noise-Aware Paradigm Using Multi-Stereo Disparity Integration

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

A Robust Monocular Depth Estimation Framework Based on Light-Weight ERF-Pspnet for Day-Night Driving Scenes

A Light Field Depth Estimation Algorithm by Edge Disparity Extraction

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

Anti-Noise Light Field Depth Estimation Using Inline Occlusion Handling

Depth Estimation Via Light Field Camera with a Hybrid Stereo Matching Method

Learning Sub-Pixel Disparity Distribution for Light Field Depth Estimation

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

An Occlusion and Noise-aware Stereo Framework Based on Light Field Imaging for Robust Disparity Estimation

Accurate Disparity Estimation in Light Field Using Ground Control Points.

Disparity Estimation for Focused Light Field Camera Using Cost Aggregation in Micro-Images

Baseline-Adaptive Light Field Depth Estimation Based on Stereo Matching

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

Fast and Accurate Optical Flow based Depth Map Estimation from Light Fields

Depth Map Estimation with 4D Light Fields Using Confocal Stereo

Depth map enhancement based on color and depth consistency

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

Disparity Estimation for Camera Arrays Using Reliability Guided Disparity Propagation.

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

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