SOFTWARE FOR MULTISCALE IMAGE-ANALYSIS - THE NORMALIZED OPTIMIZED ANISOTROPIC WAVELET COEFFICIENT METHOD

The two-dimensional Anisotropic Wavelet Transform aims to decipher ima ges in which distributions are combined at different scales. Based on the Optimized Anisotropic Wavelet Coefficient method (OAWC), we presen t two C++ programs which enable multi-scale analysis and discriminatio n of objects, or groups of objects, depending on their area, shape rat io, orientation, and location. These programs allow the identification of the different levels of organization that are present in an image (objects, clusters, alignments of clusters), and the quantification of their sizes, shape ratios, and orientations. The programs run on a st andard IBM-PC. A complete analysis, performed on a constructed image, is presented as an example for its graphical results, and illustrates the potential of quantifying the anisotropies of orientation, shape, a nd spatial distribution of organized structures at different scales. ( C) 1997 Elsevier Science Ltd.