Candela is, in my opinion, an unnecessary base physical unit. i don't think that it (or the mole) should exist in SI.
essentially it is a unit of luminosity which is a perceptual quantity and only indirectly related to a physical quantity of radiant intensity (1/683 watt per steradian that can be translated to a quantity of power because there is [itex]4 \pi[/itex] steradians comprizing all directions in 3 dimensional space) but defined at a particular frequency of visible light 540 x 10
12 Hz (which is, i think yellowish green). but to talk about equivalent luminousity for other frequencies or colors, you might need to refer to data that looks like this:
http://en.wikipedia.org/wiki/Candela
http://www.fourmilab.ch/documents/specrend/
all's i can say is that if you have an omni-directional radiating element with total radiant power of 0.018398785 watts that radiates at frequency 540 x 10
12 Hz (a wavelength of 555 nm), you have something that has, by definition, a perceived brightness of
1 candela. if it's another frequency but the same power, the luminosity will be less than 1 candela (i think you use that solid black curve in the Wikipedia article).
it's not a physical unit, it has all sorts of heuristic perceptual dependency in it, and it doesn't belong as a base unit in a system of units like SI. it's kinda a dumb inherited thing.