Noise reconstruction for image denoising

Abstract

An apparatus (901) for denoising an image, the apparatus (901) having a processor for receiving an input image, implementing a trained artificial intelligence model to form an estimate of a noise pattern in the input image, and forming an output image by subtracting the estimate of the noise pattern from the input image, the model is used to form the estimate of the noise pattern such that the estimate of the noise pattern represents a noise pattern specific to a particular image sensor type.

Description

technical field [0001] The present invention relates to computer vision, and in particular to image analysis using deep neural networks, such as convolutional neural networks. Background technique [0002] Image denoising aims to estimate potentially clean images from noisy observations. Denoising is an important step in many digital image and computer vision systems. figure 1 (a) shows how the presence of noise affects image quality. figure 1 (b) shows that by figure 1 Noise image of (a) The improvement in image quality obtained after applying image denoising techniques (according to Kai Zhang, Kai Zhang, Yunjin Chen, Deyu Meng, and Lei Zhang) Zhang), "Beyond Gaussian Denoisers: Residual Learning of Deep CNNs for Image Denoising," IEEE Transactions on Image Processing, 2017). [0003] Camera sensors output RAW data in a linear color space, where pixel measurements are proportional to the number of photoelectrons collected. The main sources of noise are shot noise (a Po...

AI Technical Summary

Problems solved by technology

However, even in the presence of skip connections, the AE pathway cannot reconstruct all natural scenes and patterns

In other words, it is unrealistic to use an autoencoder to define a nonlinear manifold that can accurately reconstruct image patterns from various real complex objects / scenes

Therefore, previous methods like RoCGAN often hallucinate complex image structures by introducing severe blurring effects or unnatural image patterns / artifacts

[0036] The heavy computation and memory footprint of these methods also hinders their application on hardware-constrained devices such as smartphones or consumer electronics

Furthermore, these methods attempt to exploit image priors to better model clean images; given the variety of natural image patterns, this is a very complex problem

[0037] While the popularity of smartphones has made them a convenient device for photography, cameras often suffer from higher levels of noise that degrade their image quality due to the smaller sensors and lenses in their cameras

Images