Analysis of retinal and cortical components of Retinex algorithms

Yeonan-Kim, Jihyun; Bertalmío, Marcelo

doi:10.1117/1.jei.26.3.031208

Cited by 10 publications

(7 citation statements)

References 71 publications

(138 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Retinex models of color vision aim to reproduce the perception of the colors of a scene inspired by psychophysical/physiological knowledge about color vision [5,31]. They can be interpreted as the averaging of perceptual distances between image pixels, as pointed out in [1].…”

Section: A Geometric Model Of Color Perceptionmentioning

confidence: 99%

A Connection Between Image Processing and Artificial Neural Networks Layers Through a Geometric Model of Visual Perception

Batard¹,

Maldonado

Steidl

et al. 2019

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

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 classification.

show abstract

Section: A Geometric Model Of Color Perceptionmentioning

confidence: 99%

A Connection Between Image Processing and Artificial Neural Networks Layers Through a Geometric Model of Visual Perception

Batard¹,

Maldonado

Steidl

et al. 2019

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

show abstract

“…Since then, many Retinex formulations have been proposed in order to improve the results of the original algorithm and extending it to more complex scenes (see e.g. [6], [20], [33], [34] for more details about Retinex theory, Retinex formulations, and their connections with vision).…”

Section: Geometry Of Brightness Perception and Image Processingmentioning

confidence: 99%

A Geometric Model of Brightness Perception and Its Application to Color Images Correction

Batard

Bertalmío

2018

J Math Imaging Vis

Self Cite

View full text Add to dashboard Cite

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 in the corresponding Retinex formulations. To that purpose, we present a generalization of the perceptual distance between image pixels by considering the parallel transport map associated to a covariant derivative on a vector

show abstract

“…For example He algorithm [9], Retinex algorithm based on dark-channel theory [2], Single image de-fog algorithm based on dark color and incident light assumption [13],and other dark channel de-fog algorithms . The sky area will emerge lots of stripes or overenhancement phenomenon or color block phenomenon.…”

Section: Phenomenon Of Over-enhancement and Its Improvementmentioning

confidence: 99%

“…At present, the single image defog algorithm can be broadly two categories: 1, algorithm does not take into account the cause of image degradation, you get sensitive information in the image after processing the data, but it's easy to distort, such as histogram equalization method [1], Retinex [2] etc. The other algorithm consider the degradation of the image, according to the physical model to elicit the original image, the resulting image is more realistic [3][4][5], but it is difficult to estimate the atmospheric light and light transmittance.…”

Section: Introductionmentioning

confidence: 99%

Over Enhancement Inhibition in Sky Region Based on Dark Channel

Ding¹,

Du²,

Song³

2018

Proceedings of the 2018 2nd International Conference on Artificial Intelligence: Technologies and Applications (ICAITA 2018)

View full text Add to dashboard Cite

Abstract-The dark channel de-fog algorithm has a landmark significance in the field of single image fogging removal. In the original algorithm, the sky region is prone to over enhancement. Existing improved algorithms has to take lots of time and space. In order to improve the effect of defogging and make sure the operation efficiency, this paper presents a sky region overgrowth suppression algorithm, which is based on the phenomenon of that these three channel values in sky area are approaching the maximum. First, the RGB channel in the original image is separated, then histogram equalization is got. Second, a proposed limiting algorithm is used to calculate the oversaturation grayscale information in the sky region, and then the transmission rate is obtained based on the correction of the dark channel de-fog algorithm. Finally, the output image is generated. The validity of the algorithm in this paper is verified. The suppression algorithm and the original algorithm is evaluated by using the visible side evaluation indicators. It is shows that the ratio of supersaturated pixels has been significantly reduced with slightly affecting new visible edge ratio and balanced gradient ratio. The over enhancement suppression algorithm can effectively suppress sky over enhancement. It is proved that the suppression algorithm can effectively correct the transmittance of the sky region in fog images, and be suitable for occasions where require high speed.

show abstract

Analysis of retinal and cortical components of Retinex algorithms

Cited by 10 publications

References 71 publications

A Connection Between Image Processing and Artificial Neural Networks Layers Through a Geometric Model of Visual Perception

A Connection Between Image Processing and Artificial Neural Networks Layers Through a Geometric Model of Visual Perception

A Geometric Model of Brightness Perception and Its Application to Color Images Correction

Over Enhancement Inhibition in Sky Region Based on Dark Channel

Contact Info

Product

Resources

About