State-of-the-Art video surveillance systems are based on analogue technologies. Recently, a trend towards IP based monitoring and recording systems can be identified. These systems allow the archiving and the analysis of the video material where the problem of huge amounts of data arises. Therefore, big command and control centers only need the visual information in case of a security event in order to derive further action by the human operators.
An additional trend is to add computational power and image processing functions on-the-spot of smart cameras. In parallel, the image resolution is increased which mainly results in higher prices for the consumer. The project miniSPOT.net is firstly focusing on adding the previously addressed computational power on low cost cameras to perform on-the-spot image analysis. Low cost cameras have a fixed plastic lens, very little dimensions, low power consumption and a rugged housing which make them a perfect fit for such purposes. Nowadays such sensors are produced in very large quantities for the automotive industry and are thus readily available at the desired prices.
The goal of the project is to implement an automated event recognition system which will solely provide visual data to the operators of command and control centers in the case of certain security relevant events. Based on this principle the entire hardware and network chain will be streamlined and re-designed in order to gain the maximum benefit in term of costs, robustness and reliability. Enablers of the project are recent developments in the field of computer vision, the use of mini camera modules (<1 EUR/unit) and digital signal processors (<10 EUR/unit) as well as adaption of the control networks for building automation systems.
Due to the involvement of the partner Hel-Wacht and the fitting of their main command and control building with networked miniSPOTs a very important and central intersection regarding critical infrastructure security is included. This building has links to safety and security installations of many important end-users from public administration, health and other sectors and by improving the security of this „system of systems“ the overall security in Austria will be improved decisively.
This project is performed in cooperation with:
CogVis Software and Consulting GmbH
Hel-Wacht Bewachungsdienst GmbH
Institut f. höhere Studien, IHS
TU Wien, Automation Systems Group
Publications
M. Hödlmoser, M. Kampel, “Multiple Camera Self-Calibration and 3D Reconstruction Using Pedestrians” – In Proc. of the 6th International Symposium on Visual Computing (ISVC’10), Part 2, pp 1 – 10, Las Vegas, USA, November 2010.
A. Zweng, T. Rittler, M. Kampel – “Evaluation of Histogram-Based Similarity Functions for Different Color Spaces”, 14th International Conference on Computer Analysis of Images and Patterns (CAIP 2011), pp.455-462, Sevilla, Spain, Aug. 2011.
M. Hödlmoser, B. Micusik, M. Kampel – “Exploiting Spatial Consistency for Object Classification and Pose Estimation” – In Proc. of the 18th IEEE International Conference on Image Processing (ICIP), pp 1009-1012, Brussels, Belgium, September 2011.
A. Zweng, M. Kampel – “Introducing Confidence Maps to Increase the Performance of Person Detectors”, 7th International Symposium on Visual Computing (ISVC’11), pp.446-455, Las Vegas, Nevada, Sep. 2011.
M. Hödlmoser, B. Micusik, M. Kampel – “Camera Auto-Calibration Using Pedestrians and Zebra-Crossings” – In Proc. of the 11th IEEE International Conference on Computer Vision Workshop on Visual Surveillance (ICCV-VS), pp 1697-1704, Barcelona, Spain, November 2011.