A computer texture analysts method based on statistical model has been used to extract features from the scanning electron microscope images of the catalysts for the oxidation of naphthalene to phthalic anhydride. Extracted texture features are effective measures for the evaluation of catalyst activation. Different kinds of catalysts have been discriminated and classified using the entropy curves with increasing temperature that is relative to the surface properties of the catalysts. In addition, we note that the curves can be used to define the optimum temperature for activation and to predict the thermal deactivation of the catalysts.