Abstract
Are there differences among human observers in their ability to distinguish between visual textures? We attempt to answer this question by presenting the results of a psychophysics experiment on human texture recognition. We limit our focus to the ability of an observer to distinguish an area containing one ‘figure’ texture from another neighboring or embedding area with a different ‘background’ texture when both textures are presented briefly and simultaneously. A mask containing random grey-scale dots is presented after the stimulus to help prevent afterimaging. A minimum set of 500 different signal-background pairs was used for all the observers. A few observers were tested with a larger set. Only monochromatic elements were used to generate the textures, and care was taken to insure that the texture discrimination could not be interfered by differences in brightness between the figure and background textures. Besides varying the type of texture, we varied the area used for displaying the signal texture, the length of the boundary between the signal and background textures, the steepness of the ‘texture gradient’ at the boundary between signal and background textures, and the nature and distribution of the components of the two textures. Elements of the set of textures used were generated using very different texture generating algorithms, including cellular automata, areal statistical criteria, etc. This allowed usto compare the performance of different observers across a large range of textures. Results are presented to show similarities among the responses of different observers, and where significant inter-observer differences occur.