Apparatus and method for acquiring face image using multiple cameras so as to identify human located at remote site

摘要

An apparatus for acquiring a face image using multiple cameras so as to identify a human being located at a remote site is disclosed. The apparatus for acquiring a face image using multiple cameras allows a PTZ camera to track an interest object from among objects detected/tracked by a fixed camera, and obtains an optimum face image for remote human identification from images generated by the PTZ camera. The apparatus for acquiring the face image using multiple cameras so as to identify a human located at a remote site includes a multi-camera control module for tracking an interest object being detected/tracked by a fixed camera through a Pan-Tilt-Zoom (PTZ) camera, and generating an image of the interest object; and a face-image acquisition module for acquiring a face image appropriate for identifying a face image of the remote human in the interest object image generated by the multi-camera control module.

主权项

1. An apparatus for acquiring a face image using multiple cameras so as to identify a human located at a remote site, comprising:
a multi-camera control module for tracking an interest object being detected and tracked by a fixed camera through a Pan-Tilt-Zoom (PTZ) camera, and generating an image of the interest object; and a face-image acquisition module for acquiring a face image appropriate for identifying a face image of the remote human in the interest object image generated by the multi-camera control module, wherein the multi-camera control module includes:
an object tracking unit for detecting and tracking the interest objects in an image captured by the fixed camera;an online learning unit for learning characteristics of each interest object tracked by the object tracking unit in an online mode; anda PTZ camera controller for controlling the PTZ camera by applying fuzzy inference to the interest object online-learned by the online learning unit, and wherein the PTZ camera controller detects the interest object from an input image using an object characteristic value, performs zoom-in or zoom-out control according to a size of the interest object, adjusts a magnitude of pan and tilt output values in response to the zoom-in or zoom-out control, and performs the fuzzy inference through a fuzzy inference engine using center coordinates of the image and based on the magnitude of the pan and tilt output values, and wherein the PTZ camera controller controls pan, tilt, and zoom operations of the PTZ camera based on an output value of the fuzzy inference engine.