eduzhai > Socail Sciences > Psychology >

The intensity JND comes from Poisson neural noise: Implications for image coding

  • paiqiu
  • (0) Download
  • 20210831
  • Save

... pages left unread,continue reading

Document pages: 12 pages

Abstract: While the problems of image coding and audio coding have frequentlybeen assumed to have similarities, specific sets of relationshipshave remained vague. One area where there should be a meaningfulcomparison is with central masking noise estimates, whichdefine the codec s quantizer step size.In the past few years, progress has been made on this problem in the auditory domain (Allen and Neely, J. Acoust. Soc. Am.,{ bf 102}, 1997, 3628-46; Allen, 1999, Wiley Encyclopedia ofElectrical and Electronics Engineering, Vol. 17, p. 422-437,Ed. Webster, J.G., John Wiley & Sons, Inc, NY).It is possible that some useful insights might now be obtainedby comparing the auditory and visual cases.In the auditory case it has been shown, directly from psychophysicaldata, that below about 5 sones(a measure of loudness, a unit of psychological intensity),the loudness JND is proportional to the square root of the loudness$ DL( L) propto sqrt{ L(I)}$.This is true for both wideband noise and tones, havinga frequency of 250 Hz or greater.Allen and Neely interpret this to mean that the internal noise isPoisson, as would be expected from neural point process noise.It follows directly that the Ekman fraction (the relative loudness JND),decreases as one over the square root of the loudness, namely$ DL L propto 1 sqrt{ L}$.Above ${ L} = 5$ sones, the relative loudness JND$ DL L approx 0.03$ (i.e., Ekman law).It would be very interesting to know if this samerelationship holds for the visual case between brightness $ B(I)$and the brightness JND $ DB(I)$. This might be tested by measuringboth the brightness JND and the brightness as a function ofintensity, and transforming the intensity JND into a brightness JND, namely [ DB(I) = B(I+ DI) - B(I) approx DI frac{d B}{dI}. ]If the Poisson nature of the loudness relation (below 5 sones)is a general result of central neural noise, as is anticipated,then one would expect that it would also hold in vision,namely that $ DB( B) propto sqrt{ B(I)}$. The history of this problem is fascinating, starting with Weber and Fechner.It is well documented that the exponent in the S.S. Stevens powerlaw is the same for loudness and brightness (Stevens, 1961) nocite{Stevens61a}(i.e., both brightness $ B(I)$ and loudness $ L(I)$ are proportional to$I^{0.3}$). Furthermore, the brightness JND data are more likeRiesz s near miss data than recent 2AFC studies of JND measures cite{Hecht34,Gescheider97}.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...