Alpha matting is an important topic in areas of computer vision. It has various applications, such as virtual reality, digital image and video editing, and image synthesis. Conventional approaches for alpha matting do not perform well when they encounter complicated background or when foreground and background color distributions overlap. It is also difficult to extract alpha matte accurately when the foreground objects are semi-transparent or hairy. In this paper, we propose a residual convolutional grid network for alpha matting, which deals with those matting problems well and has a performance comparable to the best image matting method in the literature. Meanwhile, the number of parameters in our method is less than one-third of the number of parameters in the best method.