See my ColorSpace Explorer to have a general idea about the RGB, YCrCb and HSL color spaces. This applet is quite similar but it computes a 3D Color histogram of the image. Basically it divides the color space in bins or equally sized boxes and counts how many pixels of the image are in each box (the size of the box corresponds to the number of pixels).
You can see how the distribution of the pixels in each color space and this can help us to group the pixels of the image by similar colors:
The histograms are easy to compute (just counting pixels) and it shows which are the dominant colors of the image. The main drawback is that it does not say anything about the distribution of the colors in the image, so we cannot infer information about textures or shapes using only the histogram. For shape descriptors you can take a look to my Master's thesis.