Oriented Animal-mask on the bronze vessels in the Shang and Zhou dynasty is an art of abstraction and stylization from animal face. The non-photo realistic rendering technology tires to establish a system to reproduce the style after analyzing how people create artworks. In this paper, we develop a pattern generation method for the animal-mask decorations. By analyzing the structure and identifying association between animals and animal-mask decorations, we generate facial features from employing a structural matching method to find correspondence from an sample database. For pattern stylization, we propose a modified reaction-diffusion method to stylize the input image patterns and enhance the features in animal-mask decoration. And the proposed reaction-diffusion method can also simulate the diffusion alone user designated flow vectors to generate background decoration patterns. Combining the processes above, we can generate an animal-mask decoration style image. We demonstrate results with a number of animal input images.