Abstract: Times New Roman, 8.5. Level set methods have been extensively used in image segmentation, but their implementation is complex and computationally expensive. By analyzing all segmentation algorithms based on level set, a new one based on template is proposed. This method firstly preprocessed the images by gray processing and de-noising. Then, optimized the image contour by using level set algorithm based on partial sample. Finally, extracted the target object by using the mean value replaced extraction algorithm. The experiment results show that this algorithm can obtain the satisfactory effect in segmentation precision and speed.