Noise suppression in low-light images through joint denoising and demosaicing

Chatterjee, Priyam; Joshi, Neel; Kang, Sing Bing; Matsushita, Yasuyuki

doi:10.1109/cvpr.2011.5995371

Cited by 46 publications

(28 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There is a line of work that shows how to systematically perform denoising before demosaicing [47,44,48]. In the work by Menon and Calvagno [45], as well as Zhang et al [49], the authors argue and show that joint demosaicing and denoising will yield better results.…”

Section: B Noise In Demosaicingmentioning

confidence: 99%

“…More specialized algorithms, such as the one introduced by Danielyan et al [44], extend denoising methods to the demosaicing setting; in this case the block matching 3D denoising method, by directly modelling crosschannel correlations. In the model by Chatterjee [48] denoising precedes demosaicing. Their model first identifies similar patches on the noisy observations, and uses the same patches to construct the denoised images from the noisy inputs, followed by demosaicing.…”

Section: B Noise In Demosaicingmentioning

confidence: 99%

“…Empirically, we observed monotonic performance improvements up to a certain number of cascade stages, after which the performance started declining again due to overfitting. For simultaneous demosaicing and denoising task we use the same model setup as for demosaicing, that is, we do not need to separate denoising from demosaicing [46,47,49,48]. The RTF model learns the characteristics of imaging noise from the training data and does not require separate noise variance estimation prior to its use.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Joint Demosaicing and Denoising via Learned Nonparametric Random Fields

Khashabi

Nowozin

Jancsary

et al. 2014

IEEE Trans. on Image Process.

103

View full text Add to dashboard Cite

Abstract-We introduce a machine learning approach to demosaicing, the reconstruction of color images from incomplete color filter array samples. There are two challenges to overcome by a demosaicing method: first, it needs to model and respect the statistics of natural images in order to reconstruct natural looking images; second, it needs to be able to perform well in the presence of noise. To facilitate an objective assessment of current methods we introduce a public ground truth data set of natural images suitable for research in image demosaicing and denoising. We then use this large data set to develop a machine learning approach to demosaicing. Our proposed method addresses both demosaicing challenges by learning a statistical model of images and noise from hundreds of natural images. The resulting model performs simultaneous demosaicing and denoising. We show that the machine learning approach has a number of benefits: 1. the model is trained to directly optimize a user-specified performance measure such as peak signal-to-noise ratio (PSNR) or structural similarity (SSIM), 2. we can handle novel color filter array layouts by retraining the model on such layouts, 3. it outperforms the previous state-of-the-art, in some setups by 0.7dB PSNR, faithfully reconstructing edges, textures, and smooth areas. Our results demonstrate that in demosaicing and related imaging applications, discriminatively trained machine learning models have the potential for peak performance at comparatively low engineering effort.

show abstract

Section: B Noise In Demosaicingmentioning

confidence: 99%

Section: B Noise In Demosaicingmentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

Joint Demosaicing and Denoising via Learned Nonparametric Random Fields

Khashabi

Nowozin

Jancsary

et al. 2014

IEEE Trans. on Image Process.

103

View full text Add to dashboard Cite

show abstract

“…Therefore, recently there has been a lot of interest in joint denoising and demosaicing. [2][3][4][5] While a number of authors integrate denoising in the image domain (e.g., refs 2,4 ), a second line of research 3,6,7 focuses on performing demosaicing directly in the wavelet domain of the CFA mosaic image. The CFA mosaic image is a superposition of the individual CFA component images and contains both chrominance and luminance information, either non-modulated (chrominance and luminance) or modulated (chrominance).…”

Section: Introductionmentioning

confidence: 99%

Bayesian demosaicing using gaussian scale mixture priors with local adaptivity in the dual tree complex wavelet packet transform domain

Goossens

Aelterman

Luong

et al. 2013

Computational Imaging XI

View full text Add to dashboard Cite

In digital cameras and mobile phones, there is an ongoing trend to increase the image resolution, decrease the sensor size and to use lower exposure times. Because smaller sensors inherently lead to more noise and a worse spatial resolution, digital post-processing techniques are required to resolve many of the artifacts. Color filter arrays (CFAs), which use alternating patterns of color filters, are very popular because of price and power consumption reasons. However, color filter arrays require the use of a post-processing technique such as demosaicing to recover full resolution RGB images. Recently, there has been some interest in techniques that jointly perform the demosaicing and denoising. This has the advantage that the demosaicing and denoising can be performed optimally (e.g. in the MSE sense) for the considered noise model, while avoiding artifacts introduced when using demosaicing and denoising sequentially.In this paper, we will continue the research line of the wavelet-based demosaicing techniques. These approaches are computationally simple and very suited for combination with denoising. Therefore, we will derive Bayesian Minimum Squared Error (MMSE) joint demosaicing and denoising rules in the complex wavelet packet domain, taking local adaptivity into account. As an image model, we will use Gaussian Scale Mixtures, thereby taking advantage of the directionality of the complex wavelets. Our results show that this technique is well capable of reconstructing fine details in the image, while removing all of the noise, at a relatively low computational cost. In particular, the complete reconstruction (including color correction, white balancing etc) of a 12 megapixel RAW image takes 3.5 sec on a recent mid-range GPU.

show abstract

“…Digital image processing has many advantages over analog image processing; it allows a much wider range of algorithms to be applied to the input data and can avoid problems such as the build-up of noise and signal distortion during processing [1]. The captured images typically have low signal-to-noise ratio, and the demosaicing step further corrupts the image, [2]. In most digital images, pixel values aren't proportional to the light levels in the scene.…”

Section: Introductionmentioning

confidence: 99%

Properties of Color Images Pictured on Sunlight and Different Wave Lengths

Mohammed¹

2013

IJCA

View full text Add to dashboard Cite

That the amount of the clarity in the color attribute in the single-image depends on the intensity of illumination, as well as the type of light used and the angle of its fall .This research we adopted in the light of the sun and different intensities on throughout the day and the angle of light on the picture, the purpose of Study changes in the color information received from camera. The proposed methods to address these data to get the clearest picture, It was also we working to shed different wavelengths between 400 to 750 nm and the wavelength of 255 nm and 366 nm for the purpose of studying the characteristics of the image. And we can looking for the disappearance of some band of color (RGB) colorimetric with small Concentration and its appearance at other times due to changes in the angle and intensity of light on the image, so the color shows the characteristic change color in the picture and the emergence of a difference in the captured image with time. The adoption of the mediator and the rate calculated the amount of the standard deviation of the color band in all the captured images. In order to find out which color packets disappear or appear based on light intensity , It was also we deducted three areas of each color of the primary colors in the image ( red , Green , blue) and calculate the amount of the standard deviation for it to know the times that appear or disappear some of the color packets . General TermsThe general terms we used general classification of image analysis by different time taken image for get better depend on the sources light. KeywordsExposure of camera, sun light, mean of image, measure of image quality.

show abstract

Noise suppression in low-light images through joint denoising and demosaicing

Cited by 46 publications

References 23 publications

Joint Demosaicing and Denoising via Learned Nonparametric Random Fields

Joint Demosaicing and Denoising via Learned Nonparametric Random Fields

Bayesian demosaicing using gaussian scale mixture priors with local adaptivity in the dual tree complex wavelet packet transform domain

Properties of Color Images Pictured on Sunlight and Different Wave Lengths

Contact Info

Product

Resources

About