A closer look at Texture metrics

This poster presents some insights into perceptual metrics for texture pattern categorization. An increasing number of researchers in the field of visualization are trying to exploit texture patterns to overcome the innate limitations of three dimensional color spaces. However, a comprehensive understanding of the most important features by which people group textures is essential for effective texture utilization in visualization. There have been a number of studies aiming at finding the perceptual dimensions of the texture. Among the pioneering works is Tamura et. al.s research which identified contrast, coarseness and directionality as the most significant texture dimensions. Liu and Picard presented an image model based on 2-D Wold decomposition of homogeneous random fields and defined their texture metrics as: periodicity, directionality and randomness. Ware and Knight argue that the primary texture dimensions in the context of human perception are orientation, size and contrast. In two separate studies, Rao and Lohse identified repetitive vs. non-repetitive, high contrast and non-directional vs. low-contrast and directional; granular, coarse and low-complexity vs. non-granular fine and high-complexity as the three most significant dimensions of texture. In this poster we discuss the results of our three recent studies intended to gain greater insight into perceptual texture metrics. The experiments investigate the role that orientation, scale and contrast play in characterizing a texture pattern. We particularly wanted to know how subjects would react to differences of scale, rotation, and contrast of each texture and how these results can be reconsolidated with Ware and Knights texture features: scale, orientation and contrast. Our hypothesis was that orientation, scale, and contrast are useful manipulable features within a particular single texture pattern but that between different texture patterns it is other visual features that we are at times relying on to group/classify the textures. In particular we hypothesized that human subjects categorized textures based on scale, orientation, and contrast metrics more often when under the time constraint but other metrics were involved when they had enough time to process the similarity between the contours and forms of the textures. In our first study, for each texture that Rao and Lohse used, we added versions that differed from the original in scale, contrast, and rotation. Subjects who participated in the study were asked to cluster textures together, thus replicating Rao and Lohse's study. We also designed a computerized experiment to determine if different texture metrics were used under different time constraints. After showing four images for a period of time, subjects were asked to group the images either horizontally or vertically into two groups of two. The duration in which the subjects viewed the images was one second in one experiment, and without limit in another. Anova analysis was applied to the results of the computerized experiment. The result confirmed our hypothesis.