In this paper, an improved method of adaptive distance preserving level set evolution is proposed. A weighting coefficient depending on information in the image and a stop function are defined. The evolution curve is no longer sensitive to the position of the initial curve, which can be anywhere in the image. The curve of zero level set can detect object boundaries when it is in a region with pixels having the same gray value. The method enhances capability of detecting boundary concavities. Experiments on images with different object boundaries show that the proposed method can detect the object contour effectively and has strong anti-noise ability.