Exploring the spatial information contained in a single photo is an important research topic in computer vision and photogrammetry. Theoretically, more than two photos covering the same area are needed to restore the 3D scene of the imaged areas, while those photos were taken. This research focuses on collecting the 3D spatial information from a single photo by using a developed 3D perspective grid-plate. In the proposed approach, Grassmann-Cayley geometric algebra is employed to establish the geometric relationship between the real world and the single photo such that 3D spatial information can be precisely collected without the calibrated parameters of cameras. In general, extracting the 3D spatial information from a single photo is difficult, unless additional geometric conditions are provided. A new perspective grid-plate is developed to provide extra geometric conditions during the collection of spatial information. With the perspective grid plate and the proposed approach, the spatial information can be precisely extracted to within centimeter accuracy.