Abstract:Firstly, we construct a multi‐region disparity selection module based on angular patch, which selects specific regions for generating angular patch, achieving representative sub‐angular patch by balancing different regions. Secondly, different from traditional guided deformable convolution, it leverages colour prior information to learn the aggregation of sampling points, which enhances the deformable convolution ability by learning deformation parameters and fitting irregular windows. Finally, to achieve high‐precision LF depth estimation, the authors have developed a network architecture based on the proposed multi‐region disparity selection and guided optimisation module. Light field (LF) depth estimation is a key task with numerous practical applications. However, achieving high‐precision depth estimation in challenging scenarios, such as occlusions and detailed regions (e.g. fine structures and edges), remains a significant challenge. To address this problem, the authors propose a LF depth estimation network based on multi‐region selection and guided optimisation. Firstly, we construct a multi‐region disparity selection module based on angular patch, which selects specific regions for generating angular patch, achieving representative sub‐angular patch by balancing different regions. Secondly, different from traditional guided deformable convolution, the guided optimisation leverages colour prior information to learn the aggregation of sampling points, which enhances the deformable convolution ability by learning deformation parameters and fitting irregular windows. Finally, to achieve high‐precision LF depth estimation, the authors have developed a network architecture based on the proposed multi‐region disparity selection and guided optimisation module. Experiments demonstrate the effectiveness of network on the HCInew dataset, especially in handling occlusions and detailed regions.

Efficiently Enhancing Co-Occurring Details While Avoiding Artifacts For Light Field Display

Near Eye Light Field Display Based on Human Visual Features.

Epipolar Focus Spectrum: A Novel Light Field Representation and Application in Dense-view Reconstruction

Sheared Epipolar Focus Spectrum for Dense Light Field Reconstruction.

An efficient method for all-in-focused light field rendering

Toward Low-Computation Light Field Displays by Foveated Rendering

Unsupervised Dense Light Field Reconstruction with Occlusion Awareness

A Light Field Depth Estimation Algorithm by Edge Disparity Extraction

Light Field All-in-focus Image Fusion Based on Spatially-Guided Angular Information.

Disparity Enhancement-based Light Field Angular Super-Resolution

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

Learning a Compact Spatial-Angular Representation for Light Field

72‐2: A High Resolution Near Eye Light Field Display with Fast Reconstruction Speed

Removing Stray-Light for Wild-Field Fundus Image Fusion Based on Large Generative Models.

High precision light field image depth estimation via multi‐region attention enhanced network

Efficient Light Field Acquisition for Integral Imaging with Adaptive Viewport Optimization

Light field image coding using a residual channel attention network–based view synthesis

Apparent Resolution Enhancement for Near-Eye Light Field Display

Enhanced Light Field Reconstruction by Combining Disparity and Texture Information in PSVs Via Disparity-Guided Fusion

Light Field Editing Propagation Using 4D Convolutional Neural Networks.

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