An approach to a fuzzy color detection method, robust with regard to variable illuminant level conditions

THREECOND: An Automated and Unsupervised Three Colour Fuzzy-Based Algorithm for Detecting Nuclei in Cervical Pap Smear Images

“…We consider color representation because prevents nonhomogeneity problems due to illumination and shadows, allowing color recognition process to be independent of illumination [25]. Moreover, it has to be taken into consideration that within cervical smear images there is blood, which appears as red stains that in a monochrome image is converted into a dark grey-level very similar to the one of the cells nuclei, so obstructing their location.…”

Obtaining the relevant colors of an image through stability-based fuzzy color histograms

“…A new fuzzy color histogram has been proposed based on the color stability model defined in [5]. By segmenting such histograms through a watershed-like algorithm, we can get a group of fuzzy sets that characterize the main relevant colors of an image properly, both in its number and domain.…”

“…the color components' stability of the pixels have to be considered. So, keeping in mind the color perceptual stability model given in [5], the next properties have to be implemented in thepd definition. cpd-1:If a pixel x corresponds to a stable color, variations in the illumination, or due to other sources of uncertainty, will produce at the most slight modifications of its x H and xs values.…”

“…Taking into account that HSI Smith's model [9] assures a high independence degree among H, S and I color components, and following the framework of [5] and [6], in the present work we will use Smith's Hue (H) and Saturation (S) components to represent color (from now on HS representation). Although Intensity (l) component suppression entails loss of capability for distinguishing between bright and dark shades of thetsame color, using HS representation our algorithm prevents non-homogeneity problems due to illumination and shadows, because these artifacts modify only the Smith's I-component.…”

“…Although Intensity (l) component suppression entails loss of capability for distinguishing between bright and dark shades of thetsame color, using HS representation our algorithm prevents non-homogeneity problems due to illumination and shadows, because these artifacts modify only the Smith's I-component. Considering variability of color components, in [5] we defined the Hue and Saturation deviation predictors and, using them, the H'and S stability degrees -FH, FS were introduced as the inverses of their deviation according to next definition.…”