Selection and Confirmation of a Visual Transfer Function

Selection and Confirmation of a Visual Transfer Function l Outline u. Motivation and background u. Experimental procedure u. Data analysis u. Raw results from experiment u. Implications and conclusions March 3, 2000 Nancy Goodman 1

Motivation l Produce Banding SIR for Constellation RGT l Develop market acceptability levels for Constellation l Determine “correct” relationship between sensitivity to spatial lightness variation and spatial frequency of the variation March 3, 2000 Nancy Goodman 2

Various Specifications and Visual Transfer Functions March 3, 2000 Nancy Goodman 3

Experimental Procedure l Created sets of patches representing each of 3 VTF / Spec candidates u 3 scaled amplitudes at each of 11 frequencies l 16 observers indicated contours of equiperceptibility for each set of patches u 11 men, 5 women aged 28 - 53 u. Observers were not told which patches represented which VTF / Spec candidate March 3, 2000 Nancy Goodman 4

Visual Transfer Functions represented on print board March 3, 2000 Nancy Goodman 5

Data Analysis l For each VTF / Spec candidate: u. Determined average & standard deviation of “row” for equi-perceptibility contour for each spatial frequency u. Determined correspondence between “row” and DL* u. Converted average and standard deviation of responses to DL* u. This gave a corrected curve of equi-perceptible DL* vs. spatial frequency March 3, 2000 Nancy Goodman 6

Raw Responses (row vs. spatial frequency) Sheet A - IQAF VTF March 3, 2000 Sheet B - NG visible Nancy Goodman Sheet C - F 3 spec 7

Conversion from “row” to DL* Sheet A - IQAF VTF March 3, 2000 Sheet B - NG visible Nancy Goodman Sheet C - F 3 spec 8

Responses in DL* vs. spatial frequency Sheet A - IQAF VTF March 3, 2000 Sheet B - NG visible Nancy Goodman Sheet C - F 3 spec 9

Equi-perceptibility curves derived from each set of patches March 3, 2000 Nancy Goodman 10

Comparison of results to other studies Amp PERC and amp. OBV are on v. MA 3 of market acceptability SIR Note: results of “monitor experiment” are tentative March 3, 2000 Nancy Goodman 11

Implications and Conclusions l Need to abandon old F 3 spec l Need to accept new spec based on confirmed perceptibility vs. spatial frequency relationship u. Actual spec should be determined by program people, with this information as input l Should consider revising IQAF metrics to reflect better representation of human visual response March 3, 2000 Nancy Goodman 12

Threshold vs. Spatial Frequency for Lightness Variation Equi-perceptible amplitudes (2”x 3” patches) Perceptibility threshold (large areas) At lowest frequencies; spec should be tighter due to variations in viewing distance. March 3, 2000 Equi-perceptible (from this study) and perceptibility threshold values are multiples of each other, except at lowest spatial frequency due to limited size of patches in this study. Nancy Goodman 13