Counting People by Infrared Depth Sensors

People counting is not only an important key in video surveillance, but also it is the base for a range of analysis, which can help to manage buildings or shops. The solutions based on traditional visible cameras have problems like lighting conditions. Starting from the basic ideas of an existing algorithm, which uses depth cameras located above the scene, we formulate a new algorithm based on mathematical analysis. It also raises the study of several classifiers to discern whether the objects detected in the depth maps are people or not.