Image-Based Security Technologies

A camera image is not enough

Enlarge image CCTV surveillance (© picture-alliance/dpa) Does CCTV surveillance of sensitive areas make us safer? What do technological innovations have to do with human safety? Scientists have developed some impressive projects in the field of security research as a component of the German government’s High-Tech Strategy.

The elderly gentleman has to push hard to open the door. The toilet must be here somewhere. Behind the door he finds a wide corridor – and all of a sudden a siren wails. Security forces race up. It was an emergency exit, monitored by one of the 1,700 security cameras in operation at Munich Airport.
 
When do people react to a camera image?
Do security staff always react immediately if they see something out of the ordinary on a monitor? Unfortunately not, because it would take an awful lot of people to monitor 1,700 images around the clock. Are the many cameras only there to give us the illusion of security then? Or are they designed to solve crimes after the event?
 
Spotting anything out of the ordinary
Enlarge image CCTV surveillance cameras (© picture-alliance/dpa) Scientists are called on to enable the system itself to spot anything at all that is out of the ordinary. Germany's Research Ministry is promoting research work on image recognition within the framework of the project "Improving the Security of Transport Infrastructure" (SiVe). The aerospace concern EADS is managing the joint project, in which Munich Airport, the Fraunhofer Centre for Logistics System Planning and Information Systems (ALI) and Munich Technical University are involved.
 
The project is working on automatically evaluating video images. As soon as a critical event is noted an alarm is triggered. If, for instance, anyone intended to blow up one of the airport bridges across a road a bomb would probably be thrown out of a passing car or placed by a passing car. The new system records any moving object depositing an object and carrying on, while the object remains stationary. The video image appears immediately on the monitor.
 
Recognising individuals
Another application on which scientists are working helps identify an individual or object.

Let us assume for instance that a warning is received that a terrorist wearing a green jacket with a blue suitcase is at the airport. The image recognition software must now search all possible camera images immediately and retrospectively to try to find the suspect.
 
First of all the entrance area. Did any person answering to the description enter the airport? What does the person really look like? And where is he now? Nobody could comb through all possible images at the speed required. And nobody could sift through all the possible camera images looking for the suspect.
 
Scientists have developed a system that always looks for people in conjunction with items of luggage. The computer takes the moving camera images and uses them to calculate a three-dimensional model of this combination of individual and luggage. If we are looking for a man with a green jacket and a blue suitcase, the computer can give security staff a list of images of individuals wearing green and carrying a suitcase.
 
Computers do not make decisions
Despite his enthusiasm Dirk Dickmanns, EADS project manager is quite certain that computers will not be able to make decisions independently even in the long term. They offer people an image, because they can search more swiftly. They cannot decide. Human security staff still have to make the decisions.
 

Enlarge image CCTV surveillance at Frankurt Airport (© picture-alliance/dpa) The sub-project "Image-Based Security Technologies", which is part of the SiVe project, is one of several projects sponsored by the German Research Ministry in the research area "Protection of Transport Infrastructures".
 
The research projects are seeking new ways of better protecting various transport infrastructures – from roads and rails to air and water transport. They are closely networked and designed on a supra-subject level. The projects also look at ethical, psychological and organisational issues.
 
Several young men attacked and seriously injured a man in a Berlin subway station. The press demanded to know why nobody had intervened in spite of the security cameras at the subway station. The cameras recorded the incident, but nobody noticed. Did the employees fail to do their duty? Certainly not, because a vast number of people would be needed to monitor every camera in Berlin’s subway system. So why are the cameras there?
 
First of all they are there to identify the perpetrators and help capture them. That, of course, is not a great deal of comfort to the victim. Here too image recognition would be needed. And work is being done on just this. The programme is to evaluate the movements in the video images and filter out all normal-speed movements of travellers. What movements are typical for an act of violence? They must be identified. There is still a lot of work for scientists.
 
The Security Research Programme has been running since 2007. The ministry has already provided 235 million Euro for research work designed to make life safer for citizens. It is not only a technology programme that focuses on technical innovations. Several projects are also looking at new organisational concepts and strategies for action. Research scientists and practitioners from a wide spectrum of different fields are cooperating closely.