Apparent intensity is best judged by a wide-band photo-diode, aimed directly on axis with the LED center of axis. The maker of LED's like OPTEK supply various parameters to judge color purity and brightness in mcd.

Brightness as observed from looking down the center with a photo-diode (not your eyes, as some LED's are intense enough to damage the eyes retina) must include overall brightness, stated as mcd, which accounts for lens type and viewing angle.

As an example OPTEK makes a high efficiency white LED (OVLEW1CB9) with a brightness level of an intense 24000 mcd at just 20 mA of current, 90 mA pulsed. But that is with a water-clear lens and a tight 15\$^{o}\$ viewing angle. This LED is good for flashlights and spotlights only, as it is much too harsh for a reading lamp. Expand the viewing angle and add a smoky or tinted lens and now you have an LED that is much easier on the eyes. But that is all relative.

Pure color LED's are constantly being improved. Samsung now has 'Quantum LED's which are supposed to be more close to being true red, green and blue, but the goal is to have laser LED's become the norm where color purity is mandatory, which means TV's and monitors.

A calibrated colorometer would not give accurate readings unless it was focused on the center of the LED viewing angle. How the eye sees color is non-linear and also relative, so these LED manufactures use expensive laboratory grade colorometers from many angles to assemble their datasheets.

How the eyes see color and brightness does not count in the lab, it is a marketing issue. One can only say that a high mcd value means a bright LED, but is the brightness from the drive current, the lens type or the viewing angle? It is all 3 of these parameters, so picking just one is not accurate at all for judging an LED's brightness.