NON-PARAMETRIC INFORMATION GEOMETRY AND MULTI-SCALE MODELS OF TEXTURE (WedAmOR11)

Author(s) : Washington Mio (Florida State University, United States) Dennis Badlyans (Florida State University, United States) Xiuwen Liu (Florida State University, United States)

Abstract : We develop a novel algorithmic representation of textures using the statistics of multiple spectral components of images. Histograms of filter responses are viewed as elements of a non-parametric statistical manifold, and dissimilarities of local texture patterns are quantified using a metric derived from information geometry. This representation of textures is applied to the development of a spectral cartoon mode of images, and several image segmentation experiments are discussed.

Menu