A NEW METHOD FOR CHARACTERIZING LACE BASED ON A FRACTAL INDEX

摘要

Pattern recognition requires the extraction of features from the images, and the processing of these features with a pattern recognition algorithm. In this paper, we presented some results which aimed at snowing that fractal feature, based on the estimating fractal dimension, is relevant in pattern recognition tasks. The motivation behind using fractal transformation is to develop a high-speed feature extraction (for complex patterns), and then a high speed pattern index for graphical data bases. The problem considered here was the possibility to dispose a reliable feature to set numerous patterns (some hundred of thousands patterns!). Experiment results show that this approach allows us to obtain new and interesting descriptions of complex patterns. It would be interesting now to use a multi-fractal approach (or an other multiresolution method) to compute information conserving micro-features, and to obtain a finest description of lace patterns. We think that there will be a lot of applications possible for the future in the lace industry.