For the inspection of 3D objects, machine vision systems require information on the height and size resolution of the 3D object. Most camera calibration methods use a stereo-camera system or two cameras with laser, which is expensive and involves difficult configuration related to specific variables, such as the power and position of laser and the position of camera. To overcome these problems, a new calibration method is proposed for inspecting 3D objects using a single camera. 3D spatial resolution information is obtained from the images using a new 3D calibration frame. The accuracy of the proposed method is compared with ground truth data for a 3D donut-shape component and a test component manufactured by a CNC machine. The maximum errors for the 3D donut-shape and the test component were less than 0.5 mm and 1 mm, respectively.