computer vision - Why is color segmentation easier on HSV? -
i've heard if need color segmentation on software (create binary image colored image setting pixels 1 if meet threshold rules r<100, g>100, 10< b < 123) better first convert image hsv. true? , why?
the big reason separates color information (chroma) intensity or lighting (luma). because value separated, can construct histogram or thresholding rules using saturation , hue. in theory work regardless of lighting changes in value channel. in practice nice improvement. singling out hue still have meaningful representation of base color work better rgb. end result more robust color thresholding on simpler parameters.
hue continuous representation of color 0 , 360 same hue gives more flexibility buckets use in histogram. geometrically can picture hsv color space cone or cylinder h being degree, saturation being radius, , value being height. see hsv wikipedia page.
Comments
Post a Comment