Image segmentation using the Wigner-Ville distribution

This article demonstrates how time-frequency methods can be used in applications of image processing, such as monitoring, detecting inhomogeneities in textures and texture segmentation. A Pseudo Wigner Distribution (PWD) for 2D images free of aliasing and simultaneously increasing the spatial frequency support is defined through the use of a new analytic signal. In addition to that, cross-terms can be diminished by spatial and spatial-frequency smoothing. A subsequent Principal Component Analysis stage provides a feature extraction and reduction of the dimensionality. Performance of the combined PWD and PCA method has been evaluated for a classification task by using VisTex texture composites and some examples of fabric samples with defects and medical texture images. These examples demonstrate the highly discriminative power of the present method for detecting even very subtle changes in the homogeneity of textures.