K-means RGB or HSV grouping?

Question

I want to segment the image, but someone told me that the Euclidean distance for RGB is not as good as HSV, but for HSV, since not all H, S, V have the same range, so I need to normalize it. Good to normalize HSV and then do clustering? If so, how do I normalize the HSV scale?

thanks

Answer 1

First, you need to know why HSV is preferred over RGB in image segmentation. HSV separates color information ( Chroma ) and image intensity or brightness level ( Luma ), which is very useful if you want to segment the image. For example, if you are trying to use the RGB approach for photography with the sea as the background, there is a high probability that the dominant RGB component in the sea is not blue (usually due to shadow or lighting). But if you use HSV, the value is split, and you can create histogram or threshold rules using only saturation and hue.

, RGB HSV, , → http://www.cse.msu.edu/~pramanik/research/papers/2002Papers/icip.hsv.pdf

Answer 2

.

(, , k-) .

Answer 3

HSV , , , , , Hue ( ) , , - , ( Intensity), HSV , ( ) , (V) , , , Hue .

, ( ), , (147,175,208) (208,175,147), , , Co- , .

, RGB, mahalanobis ( , ), HSV, Hue V .

Hope this helps. Thank you.