Abstract: Digital image matting is a classical method in digital image processing and editing. Affinity based methods, which share the same level with the famous sampling based one, currently play a significant role in image matting domain. The major advantages of affinity based method over sampling are processing speed and smoothing effect of alpha results. In this paper, in consideration of two early attempts in closed form solutions of affinity based methods, we briefly review two newly popular categories of image matting in closed form manner, and analyze their advantages and disadvantages, including the classical Matting Laplacian based method and the recently popular Nonlocal methods. Experimental results indicate that Nonlocal methods are superior to Matting Laplacian based method in condition of much more dissimilarity between foreground and background colors, and the simpler structure of foreground edges. Besides, we also give some suggestions and plans to those path-based and additional methods derived from the closed form solutions.