<> "The repository administrator has not yet configured an RDF license."^^ . <> . . "The intensity JND comes from Poisson neural noise: Implications for image coding"^^ . "While the problems of image coding and audio coding have frequently\nbeen assumed to have similarities, specific sets of relationships\nhave remained vague. One area where there should be a meaningful\ncomparison is with central masking noise estimates, which\ndefine the codec's quantizer step size.\nIn the past few years, progress has been made on this problem \nin the auditory domain (Allen and Neely, J. Acoust. Soc. Am.,\n{\\bf 102}, 1997, 3628-46; Allen, 1999, Wiley Encyclopedia of\nElectrical and Electronics Engineering, Vol. 17, p. 422-437,\nEd. Webster, J.G., John Wiley \\& Sons, Inc, NY).\nIt is possible that some useful insights might now be obtained\nby comparing the auditory and visual cases.\nIn the auditory case it has been shown, directly from psychophysical\ndata, that below about 5 sones\n(a measure of loudness, a unit of psychological intensity),\nthe loudness JND is proportional to the square root of the loudness\n$\\DL(\\L) \\propto \\sqrt{\\L(I)}$.\nThis is true for both wideband noise and tones, having\na frequency of 250 Hz or greater.\nAllen and Neely interpret this to mean that the internal noise is\nPoisson, as would be expected from neural point process noise.\nIt follows directly that the Ekman fraction (the relative loudness JND),\ndecreases as one over the square root of the loudness, namely\n$\\DL/\\L \\propto 1/\\sqrt{\\L}$.\nAbove ${\\L} = 5$ sones, the relative loudness JND\n$\\DL/\\L \\approx 0.03$ (i.e., Ekman law).\nIt would be very interesting to know if this same\nrelationship holds for the visual case between brightness $\\B(I)$\nand the brightness JND $\\DB(I)$. This might be tested by measuring\nboth the brightness JND and the brightness as a function of\nintensity, and transforming the intensity JND into a brightness JND, namely\n\\[\n\\DB(I) = \\B(I+ \\DI) - \\B(I)\n \\approx \\DI \\frac{d\\B}{dI}.\n\\]\nIf the Poisson nature of the loudness relation (below 5 sones)\nis a general result of central neural noise, as is anticipated,\nthen one would expect that it would also hold in vision,\nnamely that $\\DB(\\B) \\propto \\sqrt{\\B(I)}$.\n%The history of this problem is fascinating, starting with Weber and Fechner.\nIt is well documented that the exponent in the S.S. Stevens' power\nlaw is the same for loudness and brightness (Stevens, 1961)\n \\nocite{Stevens61a}\n(i.e., both brightness $\\B(I)$ and loudness $\\L(I)$ are proportional to\n$I^{0.3}$). Furthermore, the brightness JND data are more like\nRiesz's near miss data than recent 2AFC studies of JND measures\n\\cite{Hecht34,Gescheider97}. "^^ . "2000" . . "3959" . . . . . . . . . . . . . "T.N."^^ . "Pappas"^^ . "T.N. Pappas"^^ . . "B.E."^^ . "Rogowiz"^^ . "B.E. Rogowiz"^^ . . "Jont"^^ . "Allen"^^ . "Jont Allen"^^ . . . . . . "The intensity JND comes from Poisson neural noise: Implications for image coding (PDF)"^^ . . . . . . . . . "Vision.pdf"^^ . . . "The intensity JND comes from Poisson neural noise: Implications for image coding (Indexer Terms)"^^ . . . . . . "indexcodes.txt"^^ . . "HTML Summary of #1513 \n\nThe intensity JND comes from Poisson neural noise: Implications for image coding\n\n" . "text/html" . . . "Psychophysics" . .