High-Speed Detection of Intruders from Image Sequences Taken by Rotating Camera

Abstract

This paper proposes a high-speed detection method of walking intruders by a rotating camera. Our detection method is composed of the next two steps. At first, it detects candidate areas including the walking intruders. Then it classifies the candidate area into two groups by pattern recognition technique: one group is the target object and another is background. We made comparison between our detection method and existing methods that are able to detect moving objects in real-time. The comparison results show that our proposed method is the best performance in response times among them. The frame rate of candidate area detection is 67.2 fps and total frame rate that includes pattern recognition is 33.9 fps for images of 360×240 pixels.