Image Contour Clustering by Vector Quantization on Multiscale Gradient Planes and Its Application to Image Coding

摘要

We introduce an image contour clustering method based on a multiscale image representation and its appli- cation to image compression. Multiscale gradient planes are ob- tained from the mean squared sum of 2D wavelet transform of an image. The decay on the multiscale gradient planes across scales depends on the Lipshitz exponent. Since the Lipshitz exponent indicates the spatial differentiability of an image, the multiscale gradient planes represent smoothness or sharpness around edges on image contours. We apply vector quatization to the multiscale gradient planes at contours, and cluster the contours in terms of represntative vectors in VQ.