Video-based ATM Surveillance pitfalls and their solutions
Source: Nanjing Yuantuo Tech
Date: 2010/07/23
Background analysis
Recent years, ATM crimes are on the increase, especially ATM bank card crimes are increasing rapidly this year. Crimes on ATM happened in Beijing and even in Shanghai where the Expo is holding. From these cases we can see that the existing video surveillance system does not play its proper role. To trace its cause, it is because that the whole industry is over dependent on the video surveillance system. The video equipment suppliers and engineering installation providers all exaggerate the functions of video surveillance. Therefore, all banks and security departments put emphasis on the video surveillance equipment and ignore other prevention and control measures which finally lead to the crucial passive situation.
From technical perspective, video surveillance plays an indispensable role in financial security and is very important on collecting evidence afterwards and detecting the crimes. However, to take a look at its efficacy, it can only have the function of collecting evidence and testification without prevention and early warning functions.
Existing problems
Although the relevant departments publicize the modus operandi through various channels, it is still very hard for the common card holders to judge whether abnormal situations happen on the ATMs of different brands and different models. Currently, to prevent ATM crimes, the police force investments of some places are even more than security equipment investments. What’s worse, it is not effective and the social cost is very high which make the permutation losses of banks even more than investments. In addition, the practicality of current ATM intelligent video surveillance is very low which is mainly because that the video intelligent analysis relies on many factors, for instance, whether the image definition can meet the basic requirement of video analysis； whether the image is influenced (such as holding up an umbrella； wearing a helmet, scarf or masks； or small pieces of reflective sheeting) or broken by human； whether some illegal stickers with polarizing ability are used； or whether the image analysis recognition algorithm is accurate or not, etc. Only if the requirements are satisfied can it do the intelligent analysis. Therefore, the intelligent image recognition system needs very harsh terms. However, in the real ATM withdrawing environment, such terms can hardly exist, and thus the feasibility of video-based intelligent surveillance system is very bad and the actual results are not so good, either.
Solutions
To change the traditional passive video surveillance thoughts and give full play to all kinds of sensors. An omnibearing prevention and control system should be set up with sensor warning as the basis, video surveillance recording as the backup to make up the defects of pure video surveillance. It is a realistic way to solve the current plight.
Seeing from various current ATM crimes, to steal the bank card number and passwords and then clone a bank card is the most difficult to identify and prevent. It will also cause the maximum amount of harm and negative influence. This modus operandi is concealed and has long latent period with large amount of money involved. From these features, it must have two elements: 1 bank car number； 2 passwords. If only one of them can’t be achieved, the crimes will fail. Usually, the passwords can be stolen by installing a camera. However, as the installation position and method is different （some suspects even install a fake anti-peep keyboard covering which has a camera installed in it）, it is very hard to prevent.
The only way to steal card number is to install a card number theft on the card slot of the ATM (card access system is banned). Card number theft is a physical device and once installed the thickness of the card slot changes. By using a specialized equipment to measure the thickness, people can judge whether the card slot is abnormal. Meanwhile, laser has high measurement accuracy, which means that even the criminals make the card number theft thinner and smaller, laser ranging sensor can also identify precisely. The laser ranging sensor can measure the change of the thickness of the card slot, and once anything abnormal, it will raise the alarm by Internet of Things module and thus to prevent and arrest the criminals in time.
(Picture 1: Anti-card theft/ Anti-card theft host)
（Picture 2: human activity detector）
Prevention and control mode of Self-service banking outlets
(Picture 3: Anti-card number theft network system of self-service Banking outlets)
(Picture 4: stand-alone anti-card number theft system of self-service banking)
Network operation mode
Anti-card number theft system deploys network independently which forms the real Internet of Things and avoid the defects caused by being incorporated in the video surveillance host (it is very common that the video surveillance host is crashed and unstable). Although the video surveillance host is crashed, the sensor warning system can still be effective. Thus a new surveillance mode that has diversification of risk and central management can be formed.