Embodiment of the present invention is pre-treatment process for converting the RGB color image the image of the input image frame in the gray scale image; The preprocessed input image through the learning (Learning) process of the pixel units on the in the frame image to generate a background image frame, the difference between the background image frame in the input image frame inputted in turn background to create each of the background differential image the differential process; Noise removal process of removing noise from each of the background differential image; Motion area detection process of detecting a motion area in the background differential image to remove noise; And human body detecting process of detecting a person using the object with the elliptical ratio of the aspect ratio and the movement area of the movement area; may include.