Patent
GAN (Generative Adversarial Nets)-based CFA (Color Filer Array) image demosaicing joint denoising method
TLDR
In this article, a GAN-based CFA (Color Filer Array) image demosaicing joint denoising method was proposed, which consisted of the following steps: (1) a training sample setis acquired; (2) the GAN is built; (3) parameters of a 9-layer convolution neural network are updated; (4) parameters for a 39-layer CNN are updated, and (5) whether the updating times of the CNN reach 200 times or not.Abstract:
The invention discloses a GAN (Generative Adversarial Nets)-based CFA (Color Filer Array) image demosaicing joint denoising method The method comprises the following steps: (1) a training sample setis acquired; (2) the GAN is built; (3) parameters of a 9-layer convolution neural network are updated; (4) parameters of a 39-layer convolution neural network are updated; (5) whether the updating times of the 39-layer convolution neural network and the 9-layer convolution neural network reach 200 times is judged, if yes, a step (6) is executed, or otherwise, the step (3) is executed; (6) a nonlinear mapping relationship is built; and (7) an image after demosaicing and denoising is acquired The color information of a noisy CFA image obtained by a digital camera can be well recovered, the noise introduced during an image acquisition process of the digital camera can be effectively suppressed, the appearance of unnatural colors is reduced, and the visual effects of a color image are improvedread more
Citations
More filters
Patent
Real image denoising method based on generative adversarial network noise modeling
TL;DR: In this article, a real image denoising method based on generative adversarial network noise modeling is proposed, which aims to solve the problem that real image noise cannot be effectively removed due to difficulty in explicit modeling by utilizing a deep convolutional neural network.
Patent
Convolutional neural network generation method, image demosaicing method and related device
TL;DR: In this paper, a convolutional neural network generation method for performing demosaicing processing on an image, an image demosaice method, a related device, a computing device and a medium.
Patent
Image enhancement method and device, electronic device and storage medium
TL;DR: In this paper, an image enhancement method and device, an electronic device and a storage medium, and the method comprises the steps: carrying out the denoising and super-resolution processing of a first format image with a first resolution, and obtaining a first-format image with second resolution; performing demosaicing processing on the first-formatted image with the second resolution, according to the image processing method and image processing device.
Patent
High-quality color image demosaicing method based on convolutional neural network
TL;DR: In this paper, a high-quality color image demosaicing method based on a convolutional neural network is proposed, which belongs to the field of image signal processing, under the condition that the calculated amount is not greatly increased, the self-similarity and redundant information of the image are fully utilized, a color image denoising network is designed, and the color image reconstruction network is used for representing end-to-end mapping from the mosaic image to the colour image.
Patent
Image processing method and apparatus
Yang Xiaowei,Guofeng Zhang,He Yi +2 more
TL;DR: In this article, an image processing method is proposed to obtain multiple video frames from a same scene at different angles and determining a depth map of each video frame according to corresponding pixels among the multiple videos frames.
References
More filters
Patent
Producing higher-quality samples of natural images
TL;DR: In this paper, a plurality of generative adversarial networks (GANs) are applied to a particular level k of a Laplacian pyramid, where each GAN may comprise a generative model G k and a discriminative model D k.
Patent
Image generating method and apparatus
Huang Jinchi,Chen Junhong +1 more
TL;DR: In this paper, a pre-trained deep learning network model is used to generate a processed image, the model is obtained by taking a sample image and an image obtained by degradation processing of the sample image as an output sample and an input sample respectively.
Posted Content
Joint Demosaicing and Denoising with Perceptual Optimization on a Generative Adversarial Network.
TL;DR: This work has developed a GAN-based approach toward image demosacing in which a discriminator network with both perceptual and adversarial loss functions are used for quality assurance and proposes to optimize the perceptual quality of reconstructed images by the proposed GAN in an end-to-end manner.
Patent
Generative and adversarial network-based polarimetric SAR image classification method
TL;DR: In this paper, a generative and adversarial network-based polarimetric SAR image classification method was proposed to solve the problem of low classification precision caused by unreasonable selection of polarIMetric image features in the prior art.