This project aims to analyze texture cues to three-dimensional shape in terms of the modulation of apatial frequencies contained in the texture, and to separate the information conveyed by color and brightness variations. There are three sections to the project: 1. Analysis of natural textures. We will acquire a set of natural textures, transform each into three images containing variations along each of the cardinal directions of color space, and decompose each of these images into three individual structural components: indeterministic, harmonic, and evanescent. II. Frequency modulation analysis of texture cues. We will express the various texture cues to shape in terms of spatial frequency modulation, and measure sensitivity to spatial frequency modulation in color and brightness when the carrier is purely harmonic, harmonic plus indeterministic, and harmonic plus indeterministic plus evanescent. A model of detecting spatial frequency modulation will be developed. III. Surface from texture. We will measure perceived surface curvature and slant and how these influence shape perception using curved and slanted surfaces as defined by color and achromatic textures. Results will provide insight to mechanisms that use frequency modulation to infer surface shape and curvature. These results will be used to extend models of texture segmentation to surface reconstruction.
