Espace de couleur | EPFL Graph Search

vignette|Vue de côté de l'espace de couleurs CIE Lab*. Un espace de couleur ou espace chromatique est un concept de présentation des couleurs. On dispose les couleurs dans un espace à trois dimensions . Les trois nombres qui, ensemble, décrivent la couleur dans un système colorimétrique, s'assimilent à des coordonnées dans cet espace. La représentation des couleurs dans un espace chromatique ne tient pas compte de l'influence que des plages de couleur ont l'une sur l'autre . Les couleurs que l'espace représente sont des stimulus vus sur un fond neutre. Comme chaque appareil produisant des images en couleurs a des caractéristiques légèrement différentes, il produit un résultat dans un espace de couleurs particulier. Pour permettre le passage de l'un à l'autre, on communique le plus souvent par l'intermédiaire d'un profil colorimétrique. Usages du volume vignette|redresse=1.5|Sphères de couleur de Runge et de Munsell. La métaphore du volume des couleurs permet de dispose

Incrustation de microstructures par écarts chromatiques

Electronic imaging devices produce the illusion of synthesizing continuous colors by applying halftones, which are repetitive structures of discrete micro-elements. By themselves, these microstructures do not carry any meaning: their only purpose is to fool the eye. We explore a new approach for creating color images with two layers of information, one layer of information being the global image and the second layer of information being represented by a meaningful microstructure embedded within the global image. When seen from far away, only the global image should be visible. From a close viewing distance, the microstructure should provide a clear visual impact. A microstructure is incrusted into a color image by synthesizing chromatic differences between the foreground and background of the microstructure shapes. When viewed from a large distance, microstructure shapes embedded in the global image should not be distinguishable by the human eye. The resulting perceived color should be a combination of foreground and background colors, weighted by their respective surfaces. We have much freedom in creating chromatic differences expressed in terms of lightness, saturation or hue difference components. The first problem consists in creating continuous chromatic differences within a 3D color space. A second problem is the creation of perceptually similar chromatic differences across the color space. A further problem resides in determining the local coverage percentages of the microstructure's foreground and background, in order to appropriately choose their respective colors so as to ensure that when viewed from a certain distance, the combination of foreground and background appears as the desired target color. The tools we created allow without much technical efforts to create global images incorporating freely designed or scanned bi-level microstructures. They may be used for advertisement as well as for creating huge posters incorporating artistically designed microstructures such as text, logos, and ornaments. In addition, the incrustation of characters and numbers into valuable document images (e.g. photographs in ID cards) provides a protection against counterfeiting attempts. The incrustation of such information can be specifically adapted to each instance of a security document, increasing its resistance against falsification by substitution.

EPFL2003

Configural and perceptual factors influencing the perception of color transparency

Peggy Gerardin

The mechanisms by which the brain represents colors are largely unknown. In addition, the large number of color phenomena in the natural world has made understanding color rather difficult. Color transparency perception, which is studied in this thesis, is precisely one of these interesting phenomena: when a surface is seen both in plain view and through a transparent overlay, the visual system still identifies it as a single surface. Processes of the visual system have widely inspired researchers in many domains such as neurosciences, psychology, as well as computer vision. The progress of digital imaging technologies requires research engineers to deal with issues that demand knowledge of human visual processing. To humans, an image is not a random collection of pixels, but a meaningful arrangement of regions and objects. One thus can be inspired by the human visual system to investigate color representation and its applicability to digital image processing. Finding a model of perception is still a challenging matter for researchers among multidisciplinary fields. This thesis discusses the problem of defining an accurate model of transparency perception. Despite the large number of studies on this topic, the underlying mechanisms are still not well understood. Investigating perceptual transparency is challenging due to its interactions with different visual phenomena, but the most intensively studied conditions for perceptual transparency are those involving achromatic luminance and chromatic constraints. Although these models differ in many aspects, a broad distinction can be drawn between models of additive and subtractive transparency. The General Convergence Model (GCM) combines both additive and subtractive color mixtures in showing that systematic chromatic changes in a linear color space, such as translation and convergence (or a combination of both), lead to perceptual transparency. However, while this model seems to be a necessary condition, it is not a sufficient one for transparency perception. A first motivation of this thesis was to evaluate and define situations more general than the GCM. Several chromatic changes consistent or not with the GCM were generated. Additional parameters, such as configural complexity, luminance level, magnitude of the chromatic change and shift direction were tested. The main results showed that observers' responses are influenced by each of the above cited parameters. Convergences appear significantly more transparent when motion is added for bipartite configurations, or when they are generated in a checkerboard configuration. Translations are influenced by both configuration and motion. Shears are described as opaque, except when short vector lengths are combined with motion: the overlay tends to be transparent. Divergences are strongly affected by motion and vector lengths, and rotations by a combination of checkerboard configuration with luminance level and vector length. These results question the generality of the GCM. We also investigated the effects of shadows on the perception of a transparent filter. An attempt to extend these models to handle transparency perception in complex scenes involving surfaces varying in shape and depth, change in conditions of illumination and shadow, is described. A lightness-matching task was performed to evaluate how much constancy is shown by the subject among six experimental conditions, in which shadow position, shadow blur, shadow and filter blending values were varied. The results showed that lightness constancy is very high even if surfaces were seen under both filter and shadow. A systematic deviation from perfect constancy in a manner consistent with a perceived additive shift was also observed. Because the GCM includes additive mixture and is related to color and lightness constancy, these results are promising and may be explained ultimately by this model.

EPFL2005

A Stepwise Domain Adaptive Segmentation Network With Covariate Shift Alleviation for Remote Sensing Imagery

Yinlin Hu, Yunpeng Li, Rui Song

Semantic segmentation for remote sensing images (RSI) is critical for the Earth monitoring system. However, the covariate shift between RSI datasets under different capture conditions cannot be alleviated by directly using the unsupervised domain adaptation (UDA) method, which negatively affects the segmentation accuracy in RSI. We propose a stepwise domain adaptive segmentation network with covariate shift alleviation (Cov-DA) for RSI parsing to solve this issue. Specifically, to alleviate domain shift generated by different sensors, both the source and target domains are projected into a colorspace with normalized distribution through an elaborate colorspace mapping unified module (CMUM). The color distributions of these two domains tend to be more uniform. Furthermore, in the target domain, the multistatistics joint evaluation module (MJEM) is proposed to capture different statistical characteristics of subscenarios for selecting plain scenarios regarded as high-confidence segmentation results to assist the further improvement of segmentation performance. In addition, a pyramid perceptual attention module (PPAM) containing omnidirectional features without computational burdens is added to our network for effectively enhancing the multiscale feature capture ability. In the cross-city DA experiments based on the International Society for Photogrammetry and Remote Sensing (ISPRS) and aerial benchmarks, the superiority of our algorithm is significantly demonstrated. Furthermore, we release a large-scale Martian terrain dataset noted as "Mars-Seg" containing 5 K images with pixel-level accurate annotations regarding issues, such as the lack of semantic segmentation datasets for unknown scenes.

IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC2022

Incrustation de microstructures par écarts chromatiques

EPFL2003

Configural and perceptual factors influencing the perception of color transparency

Peggy Gerardin

EPFL2005