An infrared filter is often mounted in front of a camera either to capture objects at night or to reduce the light interference. In recent years, such a combined camera with an infrared filter has been extensively used in many spatial vision applications. However, a rigorous imaging model of such a combined camera is necessary for spatial vision, and this imaging model does not raise the due awareness for ether researchers or practitioners in the field. Addressing this problem, we firstly analyze whether the imaging model of this combined camera is of the pinhole one in theory, and then we analyze the influences of four factors on the imaging model in experiments, including the distance from the camera optical center to an arbitrary spatial line, the thickness of the infrared filter, the intersection angle between the arbitrary line and its corresponding imaging line, the refractivity of the infrared filter. Both our theoretical analysis and experimental results demonstrate that, the imaging model of such a com- bined camera is no longer of the pinhole one even if the imaging model of the original camera is of the ideal pin- hole. We note also that for the spatial vision applications, special attention should be paid to this ＂non-pinhole- hess＂ in order to avoid unnecessary systematical position errors, because for such applications, selection of cam-era＇s appropriate imaging model is crucial to determine the relationship between object and camera. Our result could be of reference to those people working on such combined cameras.