Doug Kerr
Well-known member
In various areas (such as metrics for various aspects of image quality) we run into "JND" notation.
As with many such matters, this flows from a very clear concept, but the "editorial" way in which it is presented has been screwed up in a way that can lead to confusion (as when we call the circle of confusion diameter limit the "circle of confusion".)
In the subjective assessment of various things, it can be useful to adopt a score in which the unit is the smallest amount of change in the actual technical property that can be noticed by a "standard observer". The unit is called the "just noticeable difference (JND). (Because of the logarithmic nature of many human perceptual properties, this is often a ratio.)
Now if we compare "case A" with "case B", and the typical human observer will find the perceived property for "case B" less (maybe worse) than "case A" by twice the amount one could just notice, we can say that the property for "case B", compared to "case A", scores as -2 JND.
Again, "JND" designates the unit.
But of course technical authors could not hold to that notion. So it gets a bit twisted, linguistically.
Today, we might read of our "case A, case B" example that:
In other words, "JND", which originally meant a unit of perceived difference, has come to mean "a perceived difference, expressed in the unit JND."
It is as if we called the length of a board, to be expressed expressed in the unit foot, its "foot". "What foot do we need for the shelves at the back of the shed?" "Oh, about 6."
"The sharpness JND for that system is -3". "Oh, what units is that in?"
Best regards,
Doug
As with many such matters, this flows from a very clear concept, but the "editorial" way in which it is presented has been screwed up in a way that can lead to confusion (as when we call the circle of confusion diameter limit the "circle of confusion".)
In the subjective assessment of various things, it can be useful to adopt a score in which the unit is the smallest amount of change in the actual technical property that can be noticed by a "standard observer". The unit is called the "just noticeable difference (JND). (Because of the logarithmic nature of many human perceptual properties, this is often a ratio.)
For example, for human perception of the "loudness" of an acoustic signal, at 1000 Hz and in a certain general neighborhood of acoustic pressure, the JND is a ratio of acoustic pressure of about 1.28 (very nearly one decibel, interestingly enough).
Now if we compare "case A" with "case B", and the typical human observer will find the perceived property for "case B" less (maybe worse) than "case A" by twice the amount one could just notice, we can say that the property for "case B", compared to "case A", scores as -2 JND.
Again, "JND" designates the unit.
But of course technical authors could not hold to that notion. So it gets a bit twisted, linguistically.
Today, we might read of our "case A, case B" example that:
The JND for that certain property of case B (compared to case A) is -2.
In other words, "JND", which originally meant a unit of perceived difference, has come to mean "a perceived difference, expressed in the unit JND."
It is as if we called the length of a board, to be expressed expressed in the unit foot, its "foot". "What foot do we need for the shelves at the back of the shed?" "Oh, about 6."
"The sharpness JND for that system is -3". "Oh, what units is that in?"
Best regards,
Doug