Search

Comunicació presentada a: the 7th International Conference on Scale Space and Variational Methods in Computer Vision, celebrada del 30 de juny al 4 de juliol de 2019 a Hofgeismar, Alemanya. In this paper, we establish a connection between image processing, visual perception, and deep learning by introducing a mathematical model inspired by visual perception from which neural network layers and image processing models for color correction can be derived. Our model is inspired by the geometry of visual perception and couples a geometric model for the organization of some neurons in the visual cortex with a geometric model of color perception. More precisely, the model is a combination of a Wilson-Cowan equation describing the activity of neurons responding to edges and textures in the area V1 of the visual cortex and a Retinex model of color vision. For some particular activation functions, this yields a color correction model which processes simultaneously edges/textures, encoded into a Riemannian metric, and the color contrast, encoded into a nonlocal covariant derivative. Then, we show that the proposed model can be assimilated to a residual layer provided that the activation function is nonlinear and to a convolutional layer for a linear activation function. Finally, we show the accuracy of the model for deep learning by testing it on the MNIST dataset for digit classiffication.

This work proposes novel hyperparameter-free losses for single view 3D reconstruction with morphable models (3DMM). We dispense with the hyperparameters used in other works by exploiting geometry, so that the shape of the object and the camera pose are jointly optimized in a sole term expression. This simplification reduces the optimization time and its complexity. Moreover, we propose a novel implicit regularization technique based on random virtual projections that does not require additional 2D or 3D annotations. Our experiments suggest that minimizing a shape reprojection error together with the proposed implicit regularization is especially suitable for applications that require precise alignment between geometry and image spaces, such as augmented reality. We evaluate our losses on a large scale dataset with 3D ground truth and publish our implementations to facilitate reproducibility and public benchmarking in this field.

Image denoising has been a topic extensively investigated over the last three decades and, as repeatedly shown in this book, denoising algorithms have become incredibly good, so much so that many researchers have started questioning the need to further pursue this line of research. In this chapter, we argue that there is indeed room for improvement of denoising results, and we propose three different avenues to explore, none of which requires the development of new denoising methods. First, we describe how it can be better to denoise a transform of the noisy image rather than denoise the noisy image directly. We mention several possible transforms, and an open problem is to find a transform that is optimal for denoising, according to a proper image quality metric. Next, we point out the importance of having a proper noise model for JPEG pictures, so that a variance stabilization transform can be developed that transforms noise in JPEG images into additive white Gaussian noise, enabling existing denoising methods to be properly applied to the JPEG case. Finally, we highlight the fact that while virtually all denoising methods are optimized and validated in terms of the PSNR or SSIM measures, these metrics are not well correlated with perceived image quality, and therefore, it could be best to optimize the parameter values of denoising methods according to subjective testing. A remaining challenge is to develop perceptually based image quality metrics that match observer preference.

_ Human perception involves many features like contours, shapes, textures, and colors to name a few. Whereas several geometric models for contours, shapes and textures perception have been proposed, the geometry of color perception has received very little attention, possibly due to the fact that our perception of colors is still not fully understood. Nonetheless, there exists a class of mathematical models, gathered under the name Retinex, that aim at modeling the color perception of an image, that are inspired by psychophysical/physiological knowledge about color perception, and that can geometrically be viewed as the averaging of perceptual distances between image pixels. Some of the Retinex models turn out to be associated to an efficient image processing technique for the correction of camera output images. The aim of this paper is to show that this image processing technique can be improved by including more properties of the human visual system. To that purpose, we first present a generalization of the perceptual distance between image pixels by considering the parallel transport map associated to a covariant derivative on a vector bundle, and from which can be derived a new image processing model for color images correction. Then, we show that the family of covariant derivatives constructed in [T. Batard and N. Sochen, J. Math. Imaging Vision, 48(3) (2014), pp. 517-543] can model some color appearance phenomena related to brightness perception. Finally, we conduct experiments in which we show that the image processing techniques induced by these covariant derivatives outperform the original approach. This work was supported by the European Research Council, Starting Grant ref. 306337, by the Spanish government and FEDER Fund, grant ref. TIN2015-71537-P (MINECO/FEDER,UE), and by the Icrea Academia Award.

Comunicació presentada a: IEEE International Conference on Image Processing 2016, celebrada del 25 l 28 de setembre de 2016 a Phoenix, Estats Units d'Amèrica. We propose a fast, local denoising method where the Euclidean curvature of the noisy image is approximated in a regularizing manner and a clean image is reconstructed from this smoothed curvature. User preference tests show that when denoising real photographs with actual noise our method produces results with the same visual quality as the more sophisticated, nonlocal algorithms Non-local Means and BM3D, but at a fraction of their computational cost. These tests also highlight the limitations of objective image quality metrics like PSNR and SSIM, which correlate poorly with user preference. This work was supported by the European Research Council, Starting Grant ref. 306337, by the Spanish government, grant ref. TIN2015-71537-P, and by the Icrea Academia Award.

Comunicació presentada al IS&T International Symposium on Electronic Imaging, celebrat del 14 al 18 de febrer de 2016 a San Francisco (CA, USA) i organitzat per la Society for Imaging Science and Technology. State-of-the-art denoising methods achieve impressive results, even for large noise levels. However, they can not be implemented in camera hardware, mainly due to the fact that they are computationally too intensive. The aim of this paper is then to show that we can obtain comparable denoising results to the ones obtained with state-of-art methods by inserting a well-chosen fast denoising method at the right location in the camera processing pipeline. We evaluate our results visually and with respect to objective measures. This work was supported by the European Research Council, Starting Grant ref. 306337, by the Spanish government, grant ref. TIN2012-38112, and by the Icrea Academia Award.

Due to technical limitations, common display devices can only reproduce images having a low range of intensity values (dynamic range). As a consequence, the dynamic range of images encoding real world scenes, which is large, has to be compressed in order for them to be reproduced on a common display, and this technique is called tone mapping. Because there is no ground truth to compare with, evaluation of a tone mapped image has to be done by comparing with the original high dynamic range image. As standard metrics based on pixel-wise comparisons are not suitable for comparing images of different dynamic range, non local perceptual based metrics are commonly used. We propose a general method for optimizing tone mapped images with respect to a given non local metric. In particular, if the metric is perceptual, i.e. it involves perceptual concepts, we provide an adequate minimization strategy. Experiments on a particular perceptual metric tested with different tone mapped images provided by several tone mapping operators validate our approach. This work was supported by European Research Council, Starting Grant ref. 306337, and by Spanish grants ref. TIN2011-15954-E and ref. TIN2012-38112.

Comunicació presentada a: the 6th International Conference, (ICISP 2014), celebrada del 30. de juny al 2 de juliol de 2014 a Cherbourg, França. In this paper, we provide a new non-local method for image denoising. The key idea we develop is to denoise the components of the image in a well-chosen moving frame instead of the image itself.We prove the relevance of our approach by showing that the PSNR of a grayscale noisy image is lower than the PSNR of its components. Experiments show that applying the Non Local Means algorithm of Buades et al. [5] on the components provides better results than applying it directly on the image. This work was supported by European Research Council, Starting Grant ref. 306337, and by Spanish grants AACC, ref. TIN2011-15954-E, and Plan Na- cional, ref. TIN2012-38112. S. Levine acknowledges partial support by NSF-DMS #0915219.

17 pages; International audience; The aim of this paper is to perform edge detection in color-infrared images from the point of view of Clifford algebras. The main idea is that such an image can be seen as a section of a Clifford bundle associated to the RGBT-space (Red, Green, Blue, Temperature) of acquisition. Dealing with geometric calculus and covariant derivatives of appropriate sections with respect to well-chosen connections allows to get various color and temperature information needed for the segmentation. We show in particular how to recover the first fundamental form of the image embedded in a LSHT-space (Luminance, Saturation , Hue, Temperature) equipped with a metric tensor. We propose applications to color edge detection with respect to a given hue interval and to edge detection in color-infrared images with constraints on temperature. Others applications related to different choices of connections, sections and embedding spaces for nD images may be considered from this general theorical framework.

In this paper, we consider an image decomposition model that provides a novel framework for image denoising. The model computes the components of the image to be processed in a moving frame that encodes its local geometry (directions of gradients and level lines). Then, the strategy we develop is to denoise the components of the image in the moving frame in order to preserve its local geometry, which would have been more affected if processing the image directly. Experiments on a whole image database tested with several denoising methods show that this framework can provide better results than denoising the image directly, both in terms of Peak signal-to-noise ratio and Structural similarity index metrics. The work of G. Ghimpeteanu, T. Batard, and/nM. Bertalmío was supported in part by the Spanish Government under/nGrant TIN2012-38112, in part by the Icrea Academia Award, and in/npart by the European Research Council under Grant 306337. The work/nof S. Levine was supported by the Na/ntional Science Foundation under/nGrant NSF-DMS 1320829.