In recent years, millimeter-wave radars have become increasingly popular as sensors for detecting targets. There are two reasons for the popularity: one is for the detection and tracking of dynamic targets, and the other is for imaging map construction for the sensing environment. At present, the imaging map construction sensors used by most people are optical cameras or laser radars, but their working effects are directly related to the climate and lighting conditions, and the results are not satisfactory when the external environment is poor. However, millimeter-wave radar has good extreme environmental adaptability and cost performance. The significance and prospect of constructing maps using millimeter-wave radar is very important and can be expected. Based on the existing imaging map construction technology of millimeter-wave radar, we propose an improved method for constructing the occupancy grid map based on the radar angle accuracy probability model in this paper. We validate this method using actual data and prove it compensates the angle error of the radar detection targets. The results show that the posterior probability of the imaging map is increased, and the discrete target reflection points become more continuous, which helps the further application of the imaging map, such as target clustering, recognition, and classification.