This paper introduces a method for texture classification using wavelet packet signatures. The authors propose a new approach to characterize textures at multiple scales by analyzing wavelet packet spaces. They measure the performance of these spaces in terms of sensitivity and selectivity for classifying 25 natural textures. Both energy and entropy metrics were computed for each wavelet packet and incorporated into distinct scale space representations. Each wavelet packet (channel) reflects a specific scale and orientation sensitivity. Wavelet packet representations for 25 natural textures were classified without error by a simple two-layer network classifier. An analyzing function with large regularity (D20) was shown to be slightly more efficient in representation and discrimination than a similar function with fewer vanishing moments (D6). Energy representations computed from the standard wavelet decomposition alone (17 features) provided classification without error for the 25 textures. The reliability of texture signatures based on wavelet packet analysis suggests that the multiresolution properties of such transforms are beneficial for segmentation, classification, and subtle discrimination of texture. The paper also discusses the use of wavelet packet decomposition and defines two measures of information used as signatures for texture discrimination. The methodology involves selecting and sampling 25 distinct natural textures, computing wavelet packet representations, and evaluating the performance of different classifiers. The results show that wavelet packet signatures can provide a powerful and efficient means for signal classification. The study compares the performance of energy and entropy measures for texture discrimination and demonstrates that wavelet packet representations can achieve perfect classification for 25 texture classes. The results indicate that longer analyzing functions are more efficient in representation and discrimination. The paper concludes that wavelet packet analysis holds great potential for robust classification and subtle discrimination of textures.This paper introduces a method for texture classification using wavelet packet signatures. The authors propose a new approach to characterize textures at multiple scales by analyzing wavelet packet spaces. They measure the performance of these spaces in terms of sensitivity and selectivity for classifying 25 natural textures. Both energy and entropy metrics were computed for each wavelet packet and incorporated into distinct scale space representations. Each wavelet packet (channel) reflects a specific scale and orientation sensitivity. Wavelet packet representations for 25 natural textures were classified without error by a simple two-layer network classifier. An analyzing function with large regularity (D20) was shown to be slightly more efficient in representation and discrimination than a similar function with fewer vanishing moments (D6). Energy representations computed from the standard wavelet decomposition alone (17 features) provided classification without error for the 25 textures. The reliability of texture signatures based on wavelet packet analysis suggests that the multiresolution properties of such transforms are beneficial for segmentation, classification, and subtle discrimination of texture. The paper also discusses the use of wavelet packet decomposition and defines two measures of information used as signatures for texture discrimination. The methodology involves selecting and sampling 25 distinct natural textures, computing wavelet packet representations, and evaluating the performance of different classifiers. The results show that wavelet packet signatures can provide a powerful and efficient means for signal classification. The study compares the performance of energy and entropy measures for texture discrimination and demonstrates that wavelet packet representations can achieve perfect classification for 25 texture classes. The results indicate that longer analyzing functions are more efficient in representation and discrimination. The paper concludes that wavelet packet analysis holds great potential for robust classification and subtle discrimination of textures.