To achieve a face recognition system robust to the presence of spectacles, we have developed a spectacle detection and extraction algorithm. The detection is realized using edge information within a small area defined between the eyes. The extraction is achieved with a deformable contour, combining edge features such as strength and orientation and geometrical features such as convexity, symmetry, smoothness and continuity. The final position of the deformable contour is obtained using dynamic programming. Experimental results demonstrate the effectiveness of the method, and the robustness to the variation of spectacle shapes and colours.