MULTI-VIEWPOINT LINKED TRACKING SYSTEM AND METHOD USING PARTICLE FILTER IN THE OBJECT AREA DETECTED BY BACKGROUND MODELING

摘要

Disclosed are a multi-viewpoint linked tracking system using a particle filter in an object area detected through background modeling, and a method for the same. The multi-viewpoint linked tracking system includes: a camera image input unit for receiving an image from at least one CCTV camera; an image pre-processor for removing noise from the image received from the camera input unit by using an average filter; an object detector for performing the background modeling through a spatial-temporal Gaussian mixture model, if the image without noise is a static image; an object labeling unit for performing parallel-labeling using a plurality of core processors with respect to a plurality of pixels of the object detected from the object detector; and a within-viewpoint-tracking unit for assigning an identifier (ID) to each object and matching an ID with each object by determining the merging or demerging, and disappearing or reappearing between the objects according to the information of the color, size, and type of object labeled by the object labeling unit using only an extended Kalman filter.