Computer Vision News - October 2016

Every month, Computer Vision News reviews a successful project. Our main purpose is to show how diverse image processing applications can be and how the different techniques can help to solve technical challenges and physical difficulties. This month we review software for fully automated surveillance with PTZ cameras , developed for a client by RSIP Vision engineers . Do you have a project in computer vision and image processing? Contact our consultants . Computer vision technologies contributed major improvements to the field of surveillance. From fully manual systems, in which videos were monitored by humans, the first step towards modern surveillance methods happened when software was able to monitor the video feeds of cameras : using predefined rules, alarms were automatically activated when intruders were detected entering specific areas. The second step occurred recently when big data and deep learning were introduced into this field. Big data allowed to process the huge amount of information coming from cameras and sensors and integrate it into some usable database. Deep learning enabled the recognition of patterns (behavior and others) which could not be learned before, nor set by simple rules. Thanks to these breakthroughs, we can use computers to track events which previously could not be brought to our attention. RSIP Vision has participated to this progress along both said steps. Regarding the first step, we have developed algorithms and complex mathematic transformations to allow tracking intruders using a PTZ camera : this is a camera providing pan, tilt and zoom functions, so that humans or objects can always be in the center of the frame. From a fully human intervention based process, RSIP Vision algorithms were able to automatically track the target intruder or event (unattended baggage, suspicious behavior), starting from the moment it was detected. On any movement of the target, the algorithm detecting its new location inside the frame adjusts the camera to keep it in the center. The algorithm can keep track of the target also in case of partial occlusion or temporary complete disappearance. Thanks to our work, security personnel can continue to observe the target as long as needed. Regarding the second step, the breakthrough is only beginning: technologies to process big amounts of information already exist and engineers are gaining expertise in the use of deep learning to make further progress in the surveillance area. We shall learn more about these new developments in next month’s Computer Vision News. Project Fully Automated Surveillance with PTZ Cameras Computer Vision News Project 15 Image: DVTEL ioIMAGE on YouTube