Pan Tilt Zoom Camera Sensing
Video Analytics provides useful tools to Video Surveillance, in order to improve efficiency and to help human agent in focusing on relevant events.
Many different sensors demonstrated to be useful in such a scenario and Pan-Tilt-Zoom (PTZ) Cameras appear to fit the purpose too, as they can both cover a wide area and provide high resolution images from a suspicious person or a relevant event. Computer Vision research in this field aims to automate the PTZ camera management in order to exploit is capability to track a moving target at a higher resolution than a normal fixed camera could do. The next step is extracting from the collected frames some information useful for the identification of the target or the recognition of the detected event.
Developing a working system that fully exploits the capability of such a sensor is extremely challenging and exciting, since it combines methods and techniques from different areas such as camera geometry and image processing algorithms.
For example the tracking algorithm cannot rely on a fixed background and the camera calibration parameters must be continuosly updated according to the current camera pose. Finally all of this is bounded by the time constraints of the camera controller, to achieve an effective online camera management.
- P. Salvagnini, F. Pernici, M. Cristani, G. Lisanti, A. Del Bimbo, V. Murino
"Non-myopic information theoretic sensor management of a single pan–tilt–zoom camera for multiple object detection and tracking"
Computer Vision and Image Understanding, Volume 134, Pages 74-88, May 2015
- P. Salvagnini, L. Bazzani, M. Cristani, V. Murino
"Person Re-identification with a PTZ Camera: an introductory study"
IEEE International Conference on Image Processing (ICIP), 2013
- P. Salvagnini, M. Cristani, A. Del Bue, V. Murino
"An experimental framework for evaluating PTZ tracking algorithms"
8th International Conference on Computer Vision Systems (ICVS), 2011