Abstract: Two-dimensional images may be analyzed at mutliple resolution with the wavelet decomposition. This paper discusses the wavelet packet analysis and its applications to catalyst surface SEM image recognition. In the study on classification and recognition of catalyst surfaces, features were computed for each wavelet packet and they fully describe the information distribution of surface images at multiple scales space. Experiment results show that the wavelet packet tree is a good description about pattern features, and provides a new approach to image texture classification.