Abstract: Active contour model is an efficient method for contour detection of objects on an image, and the contour detection is important for computer vision and pattern recognition. The energy minimized active contour models （Snake）, proposed by Kass, are used successfully for extracting salient object contours in images. In the paper, a detailed analysis of Kass’model is given with emphasis on its limitations and shortcomings , and then the model is improved and a new active contour model is proposed.This model can solve the problems of segmenting all deformable objects and multiobjects. Under all conditions, energy function of the new model is stable and oscillation does not appear.