Beyond Joint Demosaicking and Denoising: An Image Processing Pipeline for a Pixel-bin Image Sensor

Sharif, S. M. A.; Biswas, Mithun

doi:10.1109/cvprw53098.2021.00032

Cited by 37 publications

(10 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, the inference time dramatically increased as network iterations increased. CNN-based joint demosaicking and denoising methods [29]- [33] attempted to solve the demosaicking problem by training millions of images [28], combining traditional algorithms [32], using density maps and green color guidance [33] or using a generative adversarial network (GAN) along with perceptual optimization to perform joint demosaicking and denoising for Bayer and non-Bayer CFA [37]. However, the existing CNN-based joint demosaicking and denoising methods [29]- [33] did not propose a network structure that effectively learned and restored the high-frequency details of images.…”

Section: B the Joint Demosaicking And Denoising Methodsmentioning

confidence: 99%

Multilevel Feature Extraction Using Wavelet Attention for Deep Joint Demosaicking and Denoising

2022

View full text Add to dashboard Cite

Demosaicking and denoising are critical for deciding on digital camera performance. However, conventional joint demosaicking and denoising methods use traditional hand-crafted filters for preprocessing and image restoration. Therefore, it is susceptible to noise and can produce many artifacts for images with numerous edges. This paper proposes an end-to-end multi-level wavelet attention convolutional neural network (CNN) that improves image restoration performance by reducing false color artifacts, over-smoothing, and blurring during demosaicing and denoising. In detail, this paper proposes a CNN-based preprocessing method that learns the mosaic image's inter-color correlation, improving the edge's color restoration performance. Furthermore, this paper presents a multi-level feature extraction convolutional block using the Haar wavelet-based discrete wavelet transform (DWT) to remove noise from images and restore colors better at the same time. Therefore, it preserves the information of input image and feature maps, learns the correlation between global and local features, improves image restoration performance, and suppresses phenomena such as over-smoothing that tend to occur in DWT-based denoising. The proposed method is an end-to-end network structure with CNN-based preprocessing methods. In experimental results, the proposed method improves PSNR by 3.67 dB and 0.35 dB on average compared with the traditional methods and the CNN-based methods, respectively, for the dataset with many high-frequency components.

show abstract

Section: B the Joint Demosaicking And Denoising Methodsmentioning

confidence: 99%

Multilevel Feature Extraction Using Wavelet Attention for Deep Joint Demosaicking and Denoising

2022

View full text Add to dashboard Cite

show abstract

“…Therefore, we included images capture with different smartphone cameras in our LDR dataset. Typically, due to the shortcoming of smaller sensor size [13,12,34], smartphone OEMs shipped their devices with the ability to produce HDR images. However, such default HDR settings do not fit well with our target applications.…”

Section: Methods Generalizationmentioning

confidence: 99%

“…We combined a perceptual colour loss (PCL) [34] along with the L1 loss to optimize stage-II. Here, the PCL aims to guide the network to avoid any colour degradation while mapping the given 8-bit images into a 16-bit HDR image [34]. The PCL can be derived as follows:…”

Section: Optimizationmentioning

confidence: 99%

A Two-stage Deep Network for High Dynamic Range Image Reconstruction

Sharif¹,

Biswas²,

Kim³

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Mapping a single exposure low dynamic range (LDR) image into a high dynamic range (HDR) is considered among the most strenuous image to image translation tasks due to exposure-related missing information. This study tackles the challenges of single-shot LDR to HDR mapping by proposing a novel two-stage deep network. Notably, our proposed method aims to reconstruct an HDR image without knowing hardware information, including camera response function (CRF) and exposure settings. Therefore, we aim to perform image enhancement task like denoising, exposure correction, etc., in the first stage. Additionally, the second stage of our deep network learns tone mapping and bit-expansion from a convex set of data samples. The qualitative and quantitative comparisons demonstrate that the proposed method can outperform the existing LDR to HDR works with a marginal difference. Apart from that, we collected an LDR image dataset incorporating different camera systems. The evaluation with our collected realworld LDR images illustrates that the proposed method can reconstruct plausible HDR images without presenting any visual artefacts. Code available : https://github. com/sharif-apu/twostageHDR_NTIRE21.

show abstract

“…Existing methods that utilise convolutional layers for downsampling come with the inherent drawback of information loss [29], attributed to the reduction of feature size from filters and downsampling. Given its capability to preserve input information, pixel-shuffle down-sample is widely used in image denoising [30], image deraining [31] and image super-resolution [32]. It periodically rearranges the elements of the input into an output scaled by the sample stride.…”

Section: Places2mentioning

confidence: 99%

HINT: High-Quality INpainting Transformer With Mask-Aware Encoding and Enhanced Attention

Chen,

Atapour-Abarghouei,

Shum

2024

IEEE Trans. Multimedia

View full text Add to dashboard Cite

Beyond Joint Demosaicking and Denoising: An Image Processing Pipeline for a Pixel-bin Image Sensor

Cited by 37 publications

References 25 publications

Multilevel Feature Extraction Using Wavelet Attention for Deep Joint Demosaicking and Denoising

Multilevel Feature Extraction Using Wavelet Attention for Deep Joint Demosaicking and Denoising

A Two-stage Deep Network for High Dynamic Range Image Reconstruction

HINT: High-Quality INpainting Transformer With Mask-Aware Encoding and Enhanced Attention

Contact Info

Product

Resources

About