Underwater imaging is often affected by light attenuation and scattering in water, leading to degraded visual quality such as color distortion, reduced contrast, and noise. Existing underwater image enhancement (UIE) methods often lack generalization capabilities, making them unable to adapt to various underwater images captured in different aquatic environments and lighting conditions. To address these challenges, a UIE method based on conditional denoising diffusion probabilistic model (DDPM) is proposed (DiffWater), which leverages the advantages of DDPM, and trains a stable and well-converged model capable of generating high-quality and diverse samples. Considering the multiple distortion issues in underwater imaging, unconditional DDPM may not achieve satisfactory enhancement and restoration results. Therefore, DiffWater utilizes the degraded underwater image with added color compensation as a conditional guide, through which the DiffWater achieves highquality restoration of degraded underwater images. Particularly, the proposed DiffWater introduces a color compensation method that performs channel-wise color compensation in the RGB color space, tailored to different water conditions and lighting scenarios, and utilizes this condition to guide the denoising process. In the experimental section, the proposed DiffWater method is tested on four real underwater image datasets and compared against existing methods. Experimental results demonstrate that DiffWater outperforms existing comparison methods in terms of enhancement quality and effectiveness, exhibiting stronger generalization capabilities and robustness

Due to the limitation of commercial light field camera hardware devices, the imaging field of view is quite narrow. Numerous Light Field Image (LFI) stitching algorithms have been developed to expand the field of view. However, it is highly challenging to compare the performance of LFI stitching algorithms in a fair manner. Currently, due to the absence of a comprehensive benchmark database for subjective rating and a reliable objective quality metric, it is fairly difficult to comprehensively and accurately compare the actual performance of existing LFI stitching algorithms. In this study, we dedicate our efforts to the development of quality metrics for stitched Wide field of view LFI (WLFI) from subjective and objective assessment aspects. Specifically, we build the first stitched WLFI database, which provides the stitched WLFIs generated by eight representative LFI stitching algorithms, along with their corresponding subjective rating scores. Secondly, an effective blind stitched WLFI quality metric is developed to accurately assess the visual quality degradation. Extensive experiments conducted over our established WLFI database demonstrate that the proposed metric achieves higher consistency with subjective ratings than the competing quality metrics.

Light field (LF) camera can simultaneously capture the intensity and direction information of light rays, which has been widely concerned. However, limited by the size of the imaging sensor, the captured LF image (LFI) has a trade-off between spatial and angular resolutions. To this end, this paper proposes a new LF super-resolution method using frequency domain analysis and semantic prior, which designs a two-stage learning framework to enhance the spatial and angular resolutions of LFI. Specifically, the proposed method first decomposes the spatial and angular information to explore the 4D structure of LFI by using frequency domain transformation, and formulates the LF super-resolution as a frequency restoration process. Then, the decomposed frequency components are recovered in a progressive restoration manner, with new cascaded 2D and 3D convolutional neural networks. To further improve the quality of the reconstructed LFI, especially at the object boundary, the semantic prior is incorporated into the designed network to enhance its representation ability. Finally, the super-resolved LFI is reconstructed by inverse frequency domain transformation. Experimental results show that the proposed method can effectively generate high-resolution LFI, and outperforms other state-of-the-art methods in terms of both subjective visual perception and objective quality evaluation. Moreover, the proposed method can enhance the performance of LF applications such as depth estimation.

Light field (LF) imaging expands traditional imaging techniques by simultaneously capturing the intensity and direction information of light rays, and promotes many visual applications. However, owing to the inherent trade-off between the spatial and angular dimensions, LF images acquired by LF cameras usually suffer from low spatial resolution. Many current approaches increase the spatial resolution by exploring the four-dimensional (4D) structure of the LF images, but they have difficulties in recovering fine textures at a large upscaling factor. To address this challenge, this paper proposes a new deep learning-based LF spatial super-resolution method using heterogeneous imaging (LFSSR-HI). The designed heterogeneous imaging system uses an extra high-resolution (HR) traditional camera to capture the abundant spatial information in addition to the LF camera imaging, where the auxiliary information from the HR camera is utilized to super-resolve the LF image. Specifically, an LF feature alignment module is constructed to learn the correspondence between the 4D LF image and the 2D HR image to realize information alignment. Subsequently, a multi-level spatial-angular feature enhancement module is designed to gradually embed the aligned HR information into the rough LF features. Finally, the enhanced LF features are reconstructed into a super-resolved LF image using a simple feature decoder. To improve the flexibility of the proposed method, a pyramid reconstruction strategy is leveraged to generate multi-scale super-resolution results in one forward inference. The experimental results show that the proposed LFSSR-HI method achieves significant advantages over the state-of-the-art methods in both qualitative and quantitative comparisons. Furthermore, the proposed method preserves more accurate angular consistency.

Light field imaging captures both the intensity and directional information of light rays, providing users with more immersive visual experience. However, during the processes of imaging, processing, coding and reconstruction, light field images (LFIs) may encounter various distortions that degrade their visual quality. Compared to two-dimensional image quality assessment, light field image quality assessment (LFIQA) needs to consider not only the image quality in the spatial domain but also the quality degradation in the angular domain. To effectively model the factors related to visual perception and LFI quality, this paper proposes a multi-scale attention feature fusion based blind LFIQA metric, named MAFBLiF. The proposed metric consists of the following parts: MLI-Patch generation, spatial-angular feature separation module, spatial-angular feature extraction backbone network, pyramid feature alignment module and patch attention module. These modules are specifically designed to extract spatial and angular information of LFIs, and capture multi-level information and regions of interest. Furthermore, a pooling scheme guided by the LFI’s gradient information and saliency is proposed, which integrates the quality of all MLI-patches into the overall quality of the input LFI. Finally, to demonstrate the effectiveness of the proposed metric, extensive experiments are conducted on three representative LFI quality evaluation datasets. The experimental results show that the proposed metric outperforms other state-of-the-art image quality assessment metrics. The code will be publicly available at https://github.com/oldblackfish/MAFBLiF.

It is possible to generate stereo high dynamic range (HDR) images/videos by using a pair of cameras with different exposure parameters. In this paper, a learning-based stereo HDR imaging (SHDRI) method with three modules is proposed. In the proposed method, we construct three convolutional neural network (CNN) modules that perform specific tasks, including exposure calibration CNN (EC-CNN) module, hole-filling CNN (HF-CNN) module and HDR fusion CNN (HDRF-CNN) module, to combine with traditional image processing methods to model SHDRI pipeline. To avoid ambiguity, we assume that the left-view image is under-exposed and the right-view image is over-exposed. Specifically, the EC-CNN module is first constructed to convert stereo multi-exposure images into the same exposure to facilitate subsequent stereo matching. Then, based on the estimated disparity map, the right-view image is forward-warped to generate the initial left-view over-exposure image. After that, extra exposure information is utilized to guide hole-filling. Finally, the HDRF-CNN module is constructed and employed to extract fusion features to fuse the hole-filled left-view over-exposure image with the original left-view under-exposure image into the left-view HDR image. Right-view HDR images can be generated in the same way. In addition, we propose an effective two-phase training strategy to overcome the lack of a sufficient large stereo multi-exposure dataset. The experimental results demonstrate that the proposed method can generate stereo HDR images with high visual quality. Furthermore, the proposed method achieves better performance in comparison with the latest SHDRI method.