In order to improve the adjustable focus range of the image, the previous adaptive sharpness autofocus algorithm adopts different evaluation methods for deep and light defocused images, but does not give a clear method for judging the depth of defocusing of the image, which affects the feasibility of the algorithm. In order to solve this problem, the image edge is extracted from Sobel operator, and the step edge is selected through the threshold in this paper, so as to use the step edge width to judge the defocusing degree of the image. Experimental results show that this method can judge the sharpness of single image without reference. Compared with SMD and Laplace methods, it has the advantage of being scene-independent. The standard deviation of the edge width of the same definition image in different scenes is only 0.0676. In addition, the edge width is positively correlated with the defocus degree of the image, which proves the effectiveness of the proposed method.The accuracy of the algorithm in this paper has reached 86.8%, which has certain advantages compared with the high-frequency sum algorithm and the algorithm based on the clarity of the no-reference structure.