The surfaces of many ancient relics as murals and oil paintings we can see today appear three-dimensional features to some extent, which poses a challenging problem for line scan imaging to capture the features on these objects. With this goal in mind, a line scan camera-based shape from focus (SFF) method is proposed for obtaining three-dimensional information in planar cultural heritages with three-dimensional features. First, line scan imaging is employed to obtain high-resolution image sequence of objects. Next, spatial frequency (SF) is used to evaluate the focus quality of a specific local area of each image in the image sequence to find out the position of the best focused image. Through analyzing the relationship between this position and the relative movement distance of the camera-object, the three-dimensional shape of the object can be roughly estimated. Then, a fitting method was utilized to refine the local details of the three-dimensional shape of the object. An experimental setup is established to restore the three-dimensional shape of the test sample using the proposed SFF algorithm. Root mean square error (RMSE) was used as the evaluation criteria, and the result demonstrated the effectiveness of the proposed method, which shows the high precision three-dimensional shape recovery and also gains the benefits of high resolution and uniform point cloud distribution. Finally, the factors affecting the results of 3D reconstruction was discussed in detail, which may guide the selection of parameters in practice.