A brief analysis of the basic foreign body detection and early warning functions of ATM intelligent monitoring

At present, most ATMs and self-service banks are equipped with video surveillance management systems. The existing system can basically prevent financial crimes, ensure that depositors can conduct normal and safe transactions at ATMs, and bank managers can timely and accurately grasp the on-site working conditions of ATMs. However, crimes against ATMs are becoming more and more frequent, and the methods are becoming more and more diverse. The existing system cannot provide early warning and can only review video recordings for evidence after the crime occurs.

Intelligent video analysis: the protective umbrella of ATM machines

The emergence of intelligent video analysis technology solves this problem. According to the actual application environment of the ATM machine, targeted rules are preset. When the monitoring system is running, if there is a violation of the preset rules in the scene, the system will actively warn in real time to nip possible criminal behavior in the bud. The intelligent transformation of the existing monitoring system can be completed through a network video server with intelligent video analysis function. The intelligent video server is customized for the needs of the financial industry. It can encode the video analysis results and transmit the video stream through the network. It can also display the target recognition results through the video analog output port, realizing seamless connection with the existing ATM video monitoring management system.

The functions that the ATM intelligent video surveillance management system can achieve are:

· Sticker detection: detects the presence of stickers on or next to the ATM machine or on the wall of the self-service bank, and generates an alarm;

· Camera and card reader installation detection: detects illegal card readers, cameras and other foreign objects on the panel and side walls of the ATM machine and generates an alarm;

Abnormal face recognition: When the ATM cash withdrawal person's face is detected to be abnormal (such as masked or wearing a large mask), an alarm is generated;

1-meter line warning: When two or more people appear within the 1-meter line of the ATM, an alarm is generated;

· Homeless person monitoring: alarm is generated when a homeless person is detected sleeping in a self-service bank;

· People counting: Count the number of customers entering and leaving self-service banks or outlets within a unit time;

Fight detection: generates an alarm when detecting violent behaviors such as robbery or fighting;

· Detection of suspicious items left behind in business outlets;

· Detection of prolonged lingering in front of business outlets;

· Cash box monitoring in the cash dispensing area behind the ATM;

· Perimeter protection of business premises;

· Video analysis and identification of vibration and prying damage.

Adding foreign objects: the most common crime at ATMs

In practical applications, how to prevent ATM withdrawal fraud is the primary problem that intelligent monitoring systems need to solve. This article will not elaborate on the traditional perimeter defense intelligent video monitoring function, but will focus on how to prevent monitoring when various foreign objects are installed on ATMs - the ATM foreign object detection and warning function (sticker detection and installation of cameras and card readers).

At present, the main criminal methods of adding foreign objects to ATM machines are as follows:

· Installing foreign objects to steal bank card information: In addition to installing cameras, fake keyboards, recorders and other special devices on self-service bank access control systems and self-service equipment to steal cardholders' bank card numbers, passwords and other security information, criminals also use devices such as recorders to steal magnetic track information on customers' magnetic cards, and then use the stolen information to make counterfeit cards for large-scale consumption and cash withdrawals, causing significant economic losses to cardholders and issuing banks;

· Creating false impressions of card swallowing and cash failure: Criminals first place a homemade device into the ATM card reader to create the illusion of "card swallowing", or set up obstacles at the ATM cash outlet to prevent the ATM from dispensing cash. At the same time, they post fake "customer service complaint hotline" signs next to the ATM to lure cardholders to reveal security information such as card numbers and passwords, and transfer funds to their designated accounts. They even pretend to be bank customer service hotlines, which is more covert and deceptive than earlier methods.

Posting false notices: Criminals pretend to be ATM management units and post false notices or announcements on ATMs, asking cardholders to transfer funds from their bank cards to designated accounts through ATMs in order to steal cardholders’ deposits.

Foreign object detection and warning: the basis of ATM intelligent monitoring

The original monitoring system uses the camera facing the cash outlet to monitor the entire panel of the ATM or a specific area on the panel. The intelligent video server will memorize the normal situation of the ATM panel. When a certain area of ​​the panel changes, the intelligent video server will intelligently analyze the image. If a foreign object is indeed found, the ATM intelligent video monitoring management system will link the bank's superior monitoring center to the alarm, automatically upload alarm information and images, etc., and can trigger various local alarm devices. The monitoring center personnel will give voice warnings to criminal behavior or link 110 to the alarm.

Time settings

[page]

The time deployment can be set according to actual needs. Users can check the date according to actual needs. After confirmation, all deployment time configurations can be queried in the date column.

Foreign object detection area - warning area intrusion detection

The user can set the detection area according to actual needs and change the position and size of the set area. The area range after setting is the foreign body detection area. The foreign body detection area can be selected multiple times in a single screen, and the selection results can be superimposed or subtracted. The final detection area is based on the area range set in the last operation.

When a foreign object appears or disappears in the detection area, the system will identify its behavior; the system will also identify whether the foreign object exists in the detection area; once the behavior in the scene violates the preset rules, an early warning will be issued.

Foreign Object Size Settings - Event Size Filter

The foreign object size setting provides the system with a detection range for the largest and smallest foreign objects. When the trigger condition is reached within the detection range, the system will alarm the foreign object and mark the location of the foreign object with a red frame.

The method for setting the foreign object size is the same as the method for setting the foreign object detection area. For the method for setting the foreign object size, please refer to the method for setting the foreign object detection area.

The set foreign object size is the maximum and minimum foreign object size to be detected. If the foreign object is smaller than the minimum size or larger than the maximum size, the system will not detect the foreign object.

Setting the minimum residence time of foreign objects - disposal identification

The minimum residence time of foreign objects is set in the preset detection area, that is, the foreign object must stay in the detection area for a certain time at least before the system triggers an alarm. The minimum residence time of foreign objects is calculated from the time when the foreign object enters the detection area and its position does not change.

The minimum residence time of foreign objects is an important basis for the system to determine whether the target object is a foreign object. Once the target object meets the minimum time setting, the system will use a red frame to mark the location of the foreign object as an alarm.

Users can set it according to the actual daily use needs of ATM. This rule can be mainly used to detect posted notes, advertisements, etc.

Foreign body detection time setting - wandering identification

Set the foreign object detection time in the preset detection area. When the target object enters the detection area and its position keeps changing, that is, the target object keeps wandering in the area, the system will calculate the wandering time of the object, and the wandering time is calculated from the time when the position of the object changes after entering the detection area. The foreign object detection time is a time period preset by the system, that is, the foreign object wandering within this time period will not trigger an alarm.

Linkage alarm

The foreign object detection and warning function has two output modes: linked video recording and linked external alarm.

Linked video recording: Once the foreign object detection function triggers an early warning, the system will capture a snapshot and place the relevant video in the local storage area. It can also be transmitted to the bank's superior monitoring center through the Internet, which is convenient for the monitoring center to dispatch and manage personnel information.

Linked external alarm: The system sends out early warning signals and uses sound, light, voice and other methods to link alarms in real time. Other parts of the monitoring system (access control system, voice intercom, intelligent fire protection, etc.) can be networked and monitored according to needs.

Conclusion

Compared with the existing video surveillance system, intelligent video analysis technology can better solve the problem of ATM crime warning. The actual application of intelligent video analysis on ATMs is no longer new, but more manufacturers focus on the traditional perimeter defense function of intelligent video analysis, and have not invested in the necessary research and analysis of the real and primary demand for ATM monitoring - various crimes against the ATM body. Intelligent video analysis technology should innovate technology based on the actual application scenarios and business needs of ATMs and self-service banks, and cannot stop at the existing rule algorithms and expect to meet all the needs of users.