Semantic color constancy

Download
2021-2-12
Üstün, İhsan Emre
Color constancy aims to perceive the actual color of an object, disregarding the effectof the light source. Recent works showed that utilizing the semantic information inan image enhances the performance of the computational color constancy methods.Considering the recent success of the segmentation methods and the increased numberof labeled images, we propose a color constancy method that combines individualilluminant estimations of detected objects which are computed using the classes of theobjects and their associated colors. Then we introduce a weighting system that valuesthe applicability of the object classes to the color constancy problem. Lastly, weintroduce another metric expressing the detected object and how well it fits the learnedmodel of its class. Finally, we evaluate our proposed method on a popular colorconstancy dataset, confirming that each weight addition enhances the performanceof the global illuminant estimation. Experimental results show promising results,outperforming the conventional methods while competing with the state of the artmethods.

Suggestions

Multi-attitudinal approaches of colour perception: construing eleven basic colours by repertory grid technique
Akbay, Saadet; Börekçi, Naz Ayşe Güzide Z.; Department of Industrial Design (2013)
Colour is a basic aspect of perception and the perception of colour varies from individual to individual. This indicates that the perception of colours mean different semantics in various contexts to different individuals. Therefore, these differences in perception forms to behave in different attitudes towards colours among individuals and it is likely to achieve different attitudinal responses to colours from individuals. Relying on the effects of colours on individuals, the initial interest of this thesi...
Automatic target recognition and detection in infrared imagery under cluttered background
GÜNDOĞDU, ERHAN; KOÇ, AYKUT; Alatan, Abdullah Aydın (2017-09-14)
Visual object classification has long been studied in visible spectrum by utilizing conventional cameras. Since the labeled images has recently increased in number, it is possible to train deep Convolutional Neural Networks (CNN) with significant amount of parameters. As the infrared (IR) sensor technology has been improved during the last two decades, labeled images extracted from IR sensors have been started to be used for object detection and recognition tasks. We address the problem of infrared object ...
Localized surface plasmons in metal nanoparticles engineered by electron beam lithography
Güler, Urcan; Turan, Raşit; Department of Physics (2009)
In this study, optical behavior of metal nanoparticles having dimensions smaller than the wavelength of visible light is studied experimentally and numerically. Gold (Au) and silver (Ag) nanoparticles are studied due to their superior optical properties when compared to other metals. A compact code based on Discrete Dipole Approximation (DDA) is developed to compute extinction efficiencies of nanoparticles with various different properties such as material, dimension and geometry. To obtain self consistent ...
Utilization of False Color Images in Shadow Detection
Aksoy, Yagiz; Alatan, Abdullah Aydın (2012-10-13)
Shadows are illuminated as a result of Rayleigh scattering phenomenon, which happens to be more effective for small wavelengths of light. We propose utilization of false color images for shadow detection, since the transformation eliminates high frequency blue component and introduces low frequency near-infrared channel. Effectiveness of the approach is tested by using several shadow-variant texture and colorrelated cues proposed in the literature. Performances of these cues in regular and false color image...
Effects of hue, saturation, and brightness on preference: A study on Goethe's color circle with RGB color space
Camgoz, N; Yener, C (2001-06-29)
In order to investigate preference responses for foreground-background color relationships, 85 university undergraduates in Ankara, Turkey, viewed 6 background colors (red, yellow, green, cyan, blue, and magenta) on which color squares of differing hues, saturations, and brightnesses were presented. All the background colors had maximum brightness (100%) and maximum saturation (100%). Subjects were asked to show the color square they preferred on the presented background color viewed through a computer moni...
Citation Formats
İ. E. Üstün, “Semantic color constancy,” M.S. - Master of Science, Middle East Technical University, 2021.