Light Field Super-resolution Research Articles

Spatial–angular–epipolar transformer for light field spatial and angular super-resolution

Transformer-based light field (LF) super-resolution (SR) methods have recently achieved significant performance improvements due to global feature modeling by self-attention mechanisms. However, as a method designed for natural language processing, 4D LFs are reshaped into 1D sequences with an immense set of tokens, which results in a quadratic computational complexity cost. In this paper, a spatial–angular–epipolar swin transformer (SAEST) is proposed for spatial and angular SR (SASR), which sufficiently extracts SR information in the spatial, angular, and epipolar domains using local self-attention with shifted windows. Specifically, in SAEST, a spatial swin transformer and an angular standard transformer are firstly cascaded to extract spatial and angular SR features, separately. Then, the extracted SR feature is reshaped into the epipolar plane image pattern and fed into an epipolar swin transformer to extract the spatial–angular correlation information. Finally, several SAEST blocks are cascaded in a Unet framework to extract multi-scale SR features for SASR. Experiment results indicate that SAEST is a fast transformer-based SASR method with less running time and GPU consumption and has outstanding performance on simulated and real-world public datasets.

Light Field Super-Resolution Using Decoupled Selective Matching

Light Field Super-Resolution Using Decoupled Selective Matching

Self-Supervised Learning for Spatial-Domain Light-Field Super-Resolution Imaging

Self-Supervised Learning for Spatial-Domain Light-Field Super-Resolution Imaging

Adaptive pixel aggregation for joint spatial and angular super-resolution of light field images

Recent advancements in Light Field (LF) super-resolution (SR) have significantly enhanced the spatial size and angular sampling rate in LF imagery. Typically, most methods treat spatial and angular SR as separate tasks. However, the two tasks are closely related and can be jointly optimized by leveraging the intrinsic high-dimensional property of LF images. Existing methods for joint spatial and angular SR of LF images also suffer from the limitations of disparity-based warping operations. To address these challenges, we propose a novel one-stage approach for directly synthesizing densely-sampled high-resolution LF images from sparsely-sampled low-resolution observations. Specifically, we propose to adaptively aggregate advantageous pixels from different sub-aperture images of input LF for each novel view interpolation. This adaptive strategy enables our method to effectively incorporate the spatial and angular correlations of input views. Additionally, to enhance the parallax structure of reconstructed LF images, we propose a feature separation and interaction module to refine the intermediate features. Experimental results on public datasets demonstrate that our proposed method can achieve state-of-the-art performance both numerically and visually compared with previous methods. Our codes are available at https://github.com/GaoshengLiu/LFSR.

Cross-View Recurrence-Based Self-Supervised Super-Resolution of Light Field

Compared with external-supervised learning-based (ESLB) methods, self-supervised learning-based (SSLB) methods can overcome the domain gap problem caused by different light field (LF) acquisition conditions, which results in the performance degradation of light field super-resolution on unseen test datasets. Current SSLB methods exploit the cross-scale recurrence feature in the single view image for super-resolution, ignoring the correlation information among views. Different from previous works, we propose a cross-view recurrence-based self-supervised mapping framework to correlate complementary information among views in the down-scaled input LF. Specifically, the cross-view recurrence information consists of geometry structure features and similar structure features. The former is to provide sub-pixel information according to disparity correlations among adjacent views, and the latter is to acquire similar color and contour information among arbitrary views, which can compensate for error disparity guidance of geometry structure features in sharp variance areas. Moreover, instead of the widely used “All-to-All” strategy, we propose a “Part-to-Part” mapping strategy, which is better competent for SSLB approaches with limited training examples solely extracted from input LF. Finally, considering that self-supervised methods need to retrain from the beginning toward each test image, based on the proposed “part-to-part” strategy, an efficient end-to-end network is designed to extract these cross-view features for superior SASR performance with less training time. Experiment results demonstrate that our method outperforms other state-of-the-art ESLB methods on both large and small domain gap cases. Compared with the only SSLB method (LFZSSR [1]), our approach achieves better performance with 524 times less training time.

AIF-LFNet: All-in-Focus Light Field Super-Resolution Method Considering the Depth-Varying Defocus

As an aperture-divided computational imaging system, microlens array (MLA) -based light field (LF) imaging is playing an increasingly important role in computer vision. As the trade-off between the spatial and angular resolutions, deep learning (DL) -based image super-resolution (SR) methods have been applied to enhance the spatial resolution. However, in existing DL-based methods, the depth-varying defocus is not considered both in dataset development and algorithm design, which restricts many applications such as depth estimation and object recognition. To overcome this shortcoming, a super-resolution task that reconstructs all-in-focus high-resolution (HR) LF images from low-resolution (LR) LF images is proposed by designing a large dataset and proposing a convolutional neural network (CNN) -based SR method. The dataset is constructed by using Blender software, consisting of 150 light field images used as training data, and 15 light field images used as validation and testing data. The proposed network is designed by proposing the dilated deformable convolutional network (DCN) -based feature extraction block and the LF subaperture image (SAI) Deblur-SR block. The experimental results demonstrate that the proposed method achieves more appealing results both quantitatively and qualitatively.

Light field super-resolution using complementary-view feature attention

Light field (LF) cameras record multiple perspectives by a sparse sampling of real scenes, and these perspectives provide complementary information. This information is beneficial to LF super-resolution (LFSR). Compared with traditional single-image super-resolution, LF can exploit parallax structure and perspective correlation among different LF views. Furthermore, the performance of existing methods are limited as they fail to deeply explore the complementary information across LF views. In this paper, we propose a novel network, called the light field complementary-view feature attention network (LF-CFANet), to improve LFSR by dynamically learning the complementary information in LF views. Specifically, we design a residual complementary-view spatial and channel attention module (RCSCAM) to effectively interact with complementary information between complementary views. Moreover, RCSCAM captures the relationships between different channels, and it is able to generate informative features for reconstructing LF images while ignoring redundant information. Then, a maximum-difference information supplementary branch (MDISB) is used to supplement information from the maximum-difference angular positions based on the geometric structure of LF images. This branch also can guide the process of reconstruction. Experimental results on both synthetic and real-world datasets demonstrate the superiority of our method. The proposed LF-CFANet has a more advanced reconstruction performance that displays faithful details with higher SR accuracy than state-of-the-art methods.

MFSRNet: spatial-angular correlation retaining for light field super-resolution

MFSRNet: spatial-angular correlation retaining for light field super-resolution

MFSR: Light Field Images Spatial Super Resolution Model Integrated with Multiple Features

Light Field (LF) cameras can capture angular and spatial information simultaneously, making them suitable for a wide range of applications such as refocusing, disparity estimation, and virtual reality. However, the limited spatial resolution of the LF images hinders their applicability. In order to address this issue, we propose an end-to-end learning-based light field super-resolution (LFSR) model called MFSR, which integrates multiple features, including spatial, angular, epipolar plane images (EPI), and global features. These features are extracted separately from the LF image and then fused together to obtain a comprehensive feature using the Feature Extract Block (FE Block) iteratively. Gradient loss is added into the loss function to ensure that the MFSR has good performance for LF images with rich texture. Experimental results on synthetic and real-world datasets demonstrate that the proposed method outperforms other state-of-the-art methods, with a peak signal-to-noise ratio (PSNR) improvement of 0.208 dB and 0.274 dB on average for the 2× and 4× super-resolution tasks, and structural similarity (SSIM) of both improvements of 0.01 on average.

Neural Radiance Field-Based Light Field Super-Resolution in Angular Domain

Neural Radiance Field-Based Light Field Super-Resolution in Angular Domain

Progressive Multi-Scale Fusion Network for Light Field Super-Resolution

Light field (LF) cameras can record multi-view images from a single scene, and these images can provide spatial and angular information to improve the performance of image super-resolution (SR). However, it is a challenge to incorporate distinctive information from different LF views. At the same time, due to the inherent resolution of the image sensor, high spatial and angular resolution are trade-off problems. In this paper, we propose a progressive multi-scale fusion network (PMFN) to improve the LFSR performance. Specifically, a progressive feature fusion block (PFFB) based on an encoder-and-decoder structure is designed to implicitly align disparities and integrate complementary information between complementary views. The core module of the PFFB is a dual-branch multi-scale fusion module (DMFM), which can integrate the information from a reference view and auxiliary views to produce a fusion feature. Each DMFM consists of two parallel branches, which have different receptive fields to fuse hierarchical features from complementary views. Three DMFMs with a dense connection are used in the PFFB, which can fully exploit multi-level features to improve the SR performance. Experimental results on both synthetic and real-world datasets demonstrate that the proposed model achieves state-of-the-art performance among existing methods. Moreover, quantitative results show that our method can also generate faithful details.

A GPU-accelerated light-field super-resolution framework based on mixed noise model and weighted regularization

Light-field (LF) super-resolution (SR) plays an essential role in alleviating the current technology challenge in the acquisition of a 4D LF, which assembles both high-density angular and spatial information. Due to the algorithm complexity and data-intensive property of LF images, LFSR demands a significant computational effort and results in a long CPU processing time. This paper presents a GPU-accelerated computational framework for reconstructing high-resolution (HR) LF images under a mixed Gaussian-Impulse noise condition. The main focus is on developing a high-performance approach considering processing speed and reconstruction quality. From a statistical perspective, we derive a joint ell ^1-ell ^2 data fidelity term for penalizing the HR reconstruction error taking into account the mixed noise situation. For regularization, we employ the weighted non-local total variation approach, which allows us to effectively realize LF image prior through a proper weighting scheme. We show that the alternating direction method of the multipliers algorithm (ADMM) can be used to simplify the computation complexity and results in a high-performance parallel computation on the GPU Platform. An extensive experiment is conducted on both synthetic 4D LF dataset and natural image dataset to validate the proposed SR model’s robustness and evaluate the accelerated optimizer’s performance. The experimental results show that our approach achieves better reconstruction quality under severe mixed-noise conditions as compared to the state-of-the-art approaches. In addition, the proposed approach overcomes the limitation of the previous work in handling large-scale SR tasks. While fitting within a single off-the-shelf GPU, the proposed accelerator provides an average speedup of 2.46{times } and 1.57{times } for {times }2 and {times }3 SR tasks, respectively. In addition, a speedup of 77{times } is achieved as compared to CPU execution.

Detail-Preserving Transformer for Light Field Image Super-resolution

Recently, numerous algorithms have been developed to tackle the problem of light field super-resolution (LFSR), i.e., super-resolving low-resolution light fields to gain high-resolution views. Despite delivering encouraging results, these approaches are all convolution-based, and are naturally weak in global relation modeling of sub-aperture images necessarily to characterize the inherent structure of light fields. In this paper, we put forth a novel formulation built upon Transformers, by treating LFSR as a sequence-to-sequence reconstruction task. In particular, our model regards sub-aperture images of each vertical or horizontal angular view as a sequence, and establishes long-range geometric dependencies within each sequence via a spatial-angular locally-enhanced self-attention layer, which maintains the locality of each sub-aperture image as well. Additionally, to better recover image details, we propose a detail-preserving Transformer (termed as DPT), by leveraging gradient maps of light field to guide the sequence learning. DPT consists of two branches, with each associated with a Transformer for learning from an original or gradient image sequence. The two branches are finally fused to obtain comprehensive feature representations for reconstruction. Evaluations are conducted on a number of light field datasets, including real-world scenes and synthetic data. The proposed method achieves superior performance comparing with other state-of-the-art schemes. Our code is publicly available at: https://github.com/BITszwang/DPT.

Reivew of Light Field Image Super-Resolution

Currently, light fields play important roles in industry, including in 3D mapping, virtual reality and other fields. However, as a kind of high-latitude data, light field images are difficult to acquire and store. Thus, the study of light field super-resolution is of great importance. Compared with traditional 2D planar images, 4D light field images contain information from different angles in the scene, and thus the super-resolution of light field images needs to be performed not only in the spatial domain but also in the angular domain. In the early days of light field super-resolution research, many solutions for 2D image super-resolution, such as Gaussian models and sparse representations, were also used in light field super-resolution. With the development of deep learning, light field image super-resolution solutions based on deep-learning techniques are becoming increasingly common and are gradually replacing traditional methods. In this paper, the current research on super-resolution light field images, including traditional methods and deep-learning-based methods, are outlined and discussed separately. This paper also lists publicly available datasets and compares the performance of various methods on these datasets as well as analyses the importance of light field super-resolution research and its future development.

Texture-Enhanced Light Field Super-Resolution With Spatio-Angular Decomposition Kernels

Despite the recent progress in light field super-resolution (LFSR) achieved by convolutional neural networks, the correlation information of LF images has not been sufficiently studied and exploited due to the complexity of 4-D LF data. To cope with such high-dimensional LF data, most of the existing LFSR methods resorted to decomposing it into lower dimensions and subsequently performing optimization on the decomposed subspaces. However, these methods are inherently limited as they neglected the characteristics of the decomposition operations and only utilized a limited set of LF subspaces ending up failing to sufficiently extract spatio-angular features and leading to a performance bottleneck. To overcome these limitations, in this article, we comprehensively discover the potentials of LF decomposition and propose a novel concept of decomposition kernels. In particular, we systematically unify the decomposition operations of various subspaces into a series of such decomposition kernels, which are incorporated into our proposed decomposition kernel network (DKNet) for comprehensive spatio-angular feature extraction. The proposed DKNet is experimentally verified to achieve considerable improvements compared with state-of-the-art methods. To further improve DKNet in producing more visually pleasing LFSR results, based on the VGG network, we propose a light field VGG (LFVGG) loss to guide the texture-enhanced DKNet (TE-DKNet) to generate rich authentic textures and enhance LF images’ visual quality significantly. We also propose an indirect evaluation metric by taking advantage of LF material recognition to objectively assess the perceptual enhancement brought by the LFVGG loss.

Light Field Image Super-Resolution Based on Multilevel Structures

Light field (LF) images suffer from low spatial resolution due to the trade-off between angular and spatial resolutions. Thus, spatial super-resolution (SR) of LF images is an essential task to obtain high-quality LF images. However, the existing SR networks still have limitations, since they exploit only single-level features to use sub-pixel information in LF images. In this paper, we proposed a light field super-resolution (LFSR) network to effectively improve the spatial resolution of light field images. The proposed network takes one target image and its 8-neighboring images for references. We construct multi-level structures for the proposed network to effectively estimate and mix sub-pixel information in reference images. The proposed network is composed of a feature extractor, a feature warping module, a feature mixing module, and a upscaling module. The feature extractor provides multi-level features for SR and offsets to the feature warping module to obtain aligned features for multiple reference images. The feature mixing module mixes multiple aligned features based on the similarity between the target and reference images to obtain multi-level mixed features. Finally, the upscaling module generates a high-resolution residual image using the multi-level mixed features. Experimental results demonstrate the proposed network outperforms the state-of-the-art methods on various light field datasets. The pre-trained model and source codes are available at https://github.com/Hwa-Jong/LF_MLS.

3DVSR: 3D EPI volume-based approach for angular and spatial light field image super-resolution

Light field (LF) imaging, which captures both spatial and angular information of a scene, is undoubtedly beneficial to numerous applications. Although various techniques have been proposed for LF acquisition, achieving both angularly and spatially high-resolution LF remains a technology challenge. In this paper, a learning-based approach applied to 3D epipolar image (EPI) is proposed to reconstruct high-resolution LF. Through a 2-stage super-resolution framework, the proposed approach effectively addresses various LF super-resolution (SR) problems, i.e., spatial SR, angular SR, and angular-spatial SR. While the first stage provides flexible options to up-sample EPI volume to the desired resolution, the second stage, which consists of a novel EPI volume-based refinement network (EVRN), substantially enhances the quality of the high-resolution EPI volume. An extensive evaluation on 90 challenging synthetic and real-world light field scenes from 7 published datasets shows that the proposed approach outperforms state-of-the-art methods to a large extend for both spatial and angular super-resolution problem, i.e., an average peak signal to noise ratio improvement of more than 2.0 dB, 1.4 dB, and 3.14 dB in spatial SR ×2, spatial SR ×4, and angular SR respectively. The reconstructed 4D light field demonstrates a balanced performance distribution across all perspective images and presents superior visual quality compared to the previous works.

MPIN: a macro-pixel integration network for light field super-resolution

MPIN: a macro-pixel integration network for light field super-resolution

LFI-Augmenter: Intelligent Light Field Image Editing With Interleaved Spatial-Angular Convolution

The emerging light field images (LFIs) support 6 degrees of freedom (6DoF) user interaction, which is the key feature for future virtual reality (VR) media experiences. Compared to regular 2-D images, LFIs are characterized by particular image structure with both spatial and angular information. In practice, it is infeasible for the user to manually edit each subaperture of the LFI, respectively, and the user cannot guarantee the parallax consistency between different subapertures. To address this problem, we propose a deep-learning-based LFI editing scheme named central view augmentation propagation (CVAP), which employs interleaved spatial-angular convolutional neural networks (4-D CNN) for effective learning of both spatial and angular features from the input LFI. Moreover, for comparison purposes, we also implemented a “direct editing” scheme based on the geometry correspondence between subviews, and another benchmark method based on light field super resolution (LFSR). The experimental results show that CVAP achieved higher PSNR and overall more pleasant visual quality than direct editing and LFSR.

High-Dimensional Dense Residual Convolutional Neural Network for Light Field Reconstruction.

We consider the problem of high-dimensional light field reconstruction and develop a learning-based framework for spatial and angular super-resolution. Many current approaches either require disparity clues or restore the spatial and angular details separately. Such methods have difficulties with non-Lambertian surfaces or occlusions. In contrast, we formulate light field super-resolution (LFSR) as tensor restoration and develop a learning framework based on a two-stage restoration with 4-dimensional (4D) convolution. This allows our model to learn the features capturing the geometry information encoded in multiple adjacent views. Such geometric features vary near the occlusion regions and indicate the foreground object border. To train a feasible network, we propose a novel normalization operation based on a group of views in the feature maps, design a stage-wise loss function, and develop the multi-range training strategy to further improve the performance. Evaluations are conducted on a number of light field datasets including real-world scenes, synthetic data, and microscope light fields. The proposed method achieves superior performance and less execution time comparing with other state-of-the-art schemes.