Abstract: Current methods for finger tip and valley points extraction require the hand is strictly placed.In order to avoiding this drawback, an improved disk method is proposed to extract hand feature points under natural placement.The hand image is captured by a digital camera and is subsequently converted to gray image and binary image.Subseguently the single pixel contour of the hand image is extracted.Thereafter the finger tip and valley points are located by using the proposed disk method.The experiment results applied on 310 images belonging to 155 users show that the method can extract the single pixel contour of the hand image and the correct extraction ration of finger tip and valley points are 95.6% and 94.8% respectively.The method avoids the low efficiency of traditional disk method applied on high resolution hand image and the reliability of the method is not impacted by fingers' opening degree, proving its advantage.