Abstract: Objective To propose an automatic computer method to localize facial features in macaque. Methods A color model is constructed in HSI color space based on the facial points of macaque. The threshold of the color model is applied to segment the macaque image. Then the comparative integrated facial image of the macaque can be obtained from the result by applying with erosion and dilation in mathematical morphology. Results Based on the comparative proportion relationship of the facial component's position, the macaques' facial components can be automatically located, which includes eyes, mouths and so on, it is also caught the detailed contour features of facial component with edge extraction methods. Conclusion The method demonstrated in this work is robust with quite high accuracy which can detect the facial components in macaques' facial image, and its result can support the further facial identification.

Key words: Color Space model, Thresholding segmentation, Edge extraction, Mathematical morphology