Open AccessShadow detection on color images
Author(s) -
E. E. Kurbatova,
V. A. Lyalina
Publication year - 2019
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1368/3/032018
Subject(s) - artificial intelligence , thresholding , shadow (psychology) , hsl and hsv , segmentation , otsu's method , computer vision , computer science , color space , pattern recognition (psychology) , image segmentation , component (thermodynamics) , connected component labeling , image (mathematics) , scale space segmentation , psychology , virus , physics , virology , psychotherapist , biology , thermodynamics
The shadow areas cause significant problems in objects recognition and classification applications. The shadows have uniform properties in some colour spaces. Thus, the shadow detection can be effectively done by applying the threshold processing. This paper proposes a simple method to detect shadows in images using the combination of two components from different colour spaces. The B component of LAB colour space is used to detect homogenous areas using contour segmentation algorithm. The method uses the V component of HSV colour space to decide whether the obtained areas (the result from the segmentation stage) are shadowed or not. This stage is done using the mean value of the V component for these areas and subsequent thresholding based on Otsu’s method. The experimental results show that the proposed approach can get accurate detection results like the state-of-art-methods, while it is more stable for the difference of characteristics of initial images.