Pierced carving is an art form that perforates the given shape by texture patterns or images, widely used in furniture design and decorations. Usually this is manually done by expert designers and carvers, such that the produced carved model is of fine details and structural soundness. The emergence of 3D printing offers substantial opportunities for generalizing manual efforts to an automatic fabricating manner. However, geometric modeling of the image pierced carving is non-trivial due to two challenging problems: how to adapt the given image to variant and limited 3D printing resolutions with high fidelity; how to ensure the printability and physical soundness of the carved model. In this paper, we take binary images as the input and attempt to tackle the pierced image carving problem from the above two folds. Given an arbitrary binary image and a 3D surface model, we formulate the image carving into image details adaption and structure enhancement problems, iteratively optimize the image and carved shape, and obtain a 3D printable and physically sound model. Experimental results show that our approach produces fabricable and visually satisfactory results.