Abstract: A method to extract contour feature points of palmprint is proposed. Firstly, a lot of excircles were made along the edge of palmprint, then the approximate corner positions of palmprint can be located by calculating the intersection number of palmprint edges and the excircles, and lastly the accurate corner positions can be achieved by combining the approximate line character of contour. On the basis of corner positions, a consistent ROI (Region of Interest) can be extracted for each palmprint, and feature vectors can be calculated for each ROI to match all the palmprint images in the database. The experimental results show that the method here is more effective and higher in matching precision than traditional methods.