EP2422325B1 - Automated teller machine comprising at least one camera that produces image data to detect manipulation attempts - Google Patents

Description

The invention relates to a self-service terminal with at least one image data-generating camera according to the preamble of claim 1. In particular, the invention relates to a self-service terminal, which is designed as an ATM.

In the field of self-service machines, in particular ATMs, criminal acts are often carried out in the form of manipulations with the aim of spying on sensitive data, in particular PINs (Personal Identification Numbers) and / or card numbers, from users of the self-service terminal. In particular, manipulation attempts are known in which so-called skimming devices, such as keyboard superstructures and the like, are installed illegally in the operating area or control panel. Such keyboard superstructures often have their own power supply, as well as a processor, a memory and an operating program, so that an unsuspecting user is spied on entering his PIN or inserting his bank card. The spied out data is then transmitted to a remote receiver via a transmitter built into the keyboard superstructure, or is located in a keyboard overlay Data memory saved. Many of today's skimming devices are very difficult to distinguish with the human eye from original controls (keyboard, card reader, etc.).

To thwart such attempts at manipulation, monitoring systems are often used which have one or more cameras which are mounted in the area of the location of the self-service terminal and detect the entire control panel and often also the area of residence of the user. Such a solution is for example in the DE 201 02 477 U1 described. By means of the local camera monitoring, both the control panel itself and the user's area in front of it can be detected. In order to distinguish whether a person is in the occupied area, a sensor is provided.

In the GB 2 351 585 A1 a trained self-service terminal ATM is described with a camera that is mounted above the screen and detects at least a portion of the panel. The camera is aligned to capture the cash dispenser keypad. Accordingly, a self-service terminal with a control panel and controls disposed therein is known in which a camera for detecting tampering attempts in a surrounding the control panel housing portion of the self-service terminal is mounted. Accordingly, is from the GB 2 351 585 A1 a self-service terminal is known, the control panel provided his elements (keyboard, card slot, etc.), which are provided for users of the self-service terminal, wherein at least one image data generating camera is provided for monitoring the self-service terminal. The camera captures at least one of the elements provided in the control panel and generates image data from a plurality of still images. The camera is connected to a data processing unit (micro-processor), which preprocesses the generated image data into a resulting image that serves to detect tampering attempts. In particular, the difference between two individual images is formed there in order to detect changes which may indicate a manipulation attempt.

From the WO 2007/093977 A1 a self-service terminal with a camera arrangement is known which is equipped with several cameras for monitoring the self-service terminal. To detect manipulation attempts on the self-service terminal, which is eg an ATM, several cameras are mounted in the vicinity of the control panel of the ATM. For example, one camera captures the card slot and another camera captures the cash slot. Another camera is aimed at the user of the ATM.

The DE 203 18 489 U1 describes an ATM with a monitoring device, the two cameras (image pickup elements 21 and 22 in Fig. 2 ) having. One camera is aimed at the user; the other camera is aligned with the area of the cash dispensing opening or the cash dispenser.

Object of the present invention is to propose a solution for a camera surveillance, which allows a secure detection of manipulation attempts without the use of additional sensors. It should be a qualitative created a high-quality database and provided for the detection of manipulation attempts.

The problem is solved by a device having the features of claim 1.

Accordingly, a self-service terminal is proposed, in which the data processing unit combines the generated image data of the single image recordings by means of image data processing, which performs a segmentation and / or edge detection, to the result image, wherein the data processing unit segmented the single image recordings into a plurality of the at least one detected element subregions and the Frame data processed differently in segments. In this case, it is provided that the data processing unit composes the result image from the subareas of different individual image recordings, and that the data processing unit processes the image data from the subareas with different image data processing and / or with different variants of image data processing.

The at least one camera preferably generates the image data of the individual image recordings as a function of predefinable criteria, in particular at predeterminable time intervals and / or at different exposure ratios or ambient brightness. Also vorgebare camera settings, in particular exposure times andlor frame rates, are taken into account. The data processing unit combines this image data (individual image data) by means of image data preprocessing, in particular averaging, median formation and / or so-called exposure blending, to the result image or overall image, which is then available for manipulation recognition. It is also possible to continuously calculate result or overall images (result image sequence) at time intervals in order to then be available for a comparison for the detection of manipulation attempts.

Also, at least one further camera may be provided, which is also mounted on or in the self-service terminal in the vicinity of the control panel and at least one of the controls, such. Keyboard, card slot, cash dispenser, recorded. The image data or individual images generated by this additional camera can also be combined together with the image data of the other camera to form a result image or to a result image sequence.

The result images obtained from the single image recordings have a significantly higher image data quality than the respective frames. Thus, a high quality database in the form of preprocessed image data is provided for tamper detection.

In this case, it may be advantageous if the plurality of individual image recordings are created as a function of at least one predefinable function, which specifies different exposure times for the single image recordings. This ensures that no single images are taken with the same exposure time, which in turn is advantageous for exposure blending. In this context, it can be provided that the at least one predefinable function corresponds to at least one ramp function which indicates increasing and / or decreasing exposure times for a series of single-image recordings. Thus, for example, in the first single image recording with the shortest exposure time of, for example, 0.5 ms started and in the subsequent shots, the exposure time successively increased until at the last recording a max. Exposure time of eg 2000 ms is reached. Alternatively, the ramp can be sloping, ie, the exposure times successively decrease. The total duration of all still images can also be specified and, for example, be 10 seconds. It is also advantageous if one of the predeterminable functions predetermines the different exposure times such that they lie within a specific range of values, ie, for example, within a first lower value range, which, for example, ranges from 0.5 ms to 1000 ms. This value range is preferably suitable for a so-called day mode, ie for the case in which a brightness and / or contrast value of at least one of the individual image recordings has a predefinable threshold value exceeds. In the so-called night mode, ie when a brightness and / or contrast value of at least one of the individual image recordings falls below a predefinable threshold value, the different exposure times are settled within a second upper value range, for example ranging from 1000 ms to 2000 ms. The functions can also be put together to form a function.

These and other particularly advantageous embodiments will become apparent from the dependent claims.

Accordingly, it is also advantageous if the at least one camera generates the image data of the single image recordings as a function of events, in particular of events recorded by the latter or by another camera. Such events may e.g. sudden onset of image lightening or darkening. It can e.g. also be control signals (operation of the keyboard or the like). It may be advantageous if single image recordings are not made (only) during the occurrence of the event, but also afterwards.

Preferably, the data processing unit combines the generated image data of the single image recordings by means of one or more suitable image data processing, such as so-called exposure blending. In this case, an image segmentation and / or edge detection can also be used. In this context, too, it is advantageous if the data processing unit segments the individual image recordings into a plurality of partial regions assigned to the at least one detected element and processes the individual image data differently in different segments. It can be provided the data processing unit composes the result image from the subareas of different single-frame images. It can also be provided that the data processing unit processes the image data from the subregions with different image data processing and / or with different variants of image data processing. The partial regions preferably comprise at least one near or inner region and an ambient or outer region of the detected element, such as the slot region and the surrounding region of a card insertion funnel. It can also be provided that one of the partial regions comprises a transition region between the inner region and the outer region of the element.

The data processing unit is preferably designed such that it carries out both the image data preprocessing and the actual image data evaluation, ie that it calculates the preprocessed image data of the resulting image from the individual image data and evaluates these by means of image processing for the purpose of detecting manipulation attempts. For this purpose, the data processing unit has a first stage, which receives the preprocessed image data, for the actual image processing or image data evaluation, wherein, in particular, shadow removal, edge detection, vectorization and / or segmentation can be carried out. The data processing unit also has a second stage downstream of the first stage for feature extraction, in particular by means of blob analysis, edge position and / or color distribution. In addition, the data processing unit has a third stage downstream of the second stage for classification.

Preferably, the data processing unit is integrated in the self-service terminal.

The elements provided in the control panel of the self-service terminal and detected by the at least one camera include e.g. a cashbox, a keyboard, a tray, a card slot, and / or a screen. It is also envisaged that the data processing unit, if it detects a manipulation attempt on the detected elements by means of processing the preprocessed image data of the result image, triggers an alarm, blocks the self-service terminal and / or triggers the additional camera. This additional camera may be a portrait camera, i. a camera which detects the area where a user, in particular his head, is during the operation of the self-service terminal. Thus, if necessary, a portrait of the user can be included. It is also provided that the respective camera and / or the data processing unit is deactivated during the operation and / or maintenance of the self-service terminal.

Fig. 1

zeigt eine perspektivische Ansicht auf das Bedienfeld eines Selbstbedienungsterminals mit mehreren Kameras;

Fig. 2

gibt den Erfassungsbereich derjenigen Kamera aus Fig. 1 wieder, die das Bedienfeld von der Seite erfasst;

Fig. 3a-d

zeigen beispielhaft drei Einzelbildaufnahmen und ein daraus gewonnenes Ergebnisbild;

Fig. 4

veranschaulicht eine Bilddaten-Vorverarbeitung von mehreren Einzelbildern mittels Kantendetektion und Kombination zu einem Ergebnisbild;

Fig. 5

veranschaulicht eine Bilddaten-Vorverarbeitung von mehreren Einzelbildern mittels einer pixelweisen Median-Bildung;

Fig. 6

gibt den Erfassungsbereich derjenigen Kamera aus Fig. 1 wieder, die das Bedienfeld von Oben erfasst;

Fig. 7a

zeigt die Einbaussituation derjenigen Kamera, die in den Karteneingabetrichter integriert ist;

Fig. 7b

gibt den Erfassungsbereich dieser Kamera aus Fig. 7a wieder;

Fig. 8

zeigt ein Blockschaltbild für eine mit mehreren der Kameras verbundener Datenverarbeitungseinheit und damit verbundener Videoüberwachungs-Einheit;

Fig. 9

veranschaulicht die Durchführung von Einzelbildaufnahmen entsprechend einer vorgebbaren Belichtungsreihe; und

Fig. 10a)-c)

zeigen dazu verschiedene Funktionsverläufe in Form von abfallenden und/oder ansteigenden Rampen.

The invention and the advantages arising therefrom are described below by means of exemplary embodiments and with reference to the attached schematic drawings, which show the following:

Fig. 1

shows a perspective view of the control panel of a self-service terminal with multiple cameras;

Fig. 2

outputs the detection range of that camera Fig. 1 again, which captures the control panel from the side;

Fig. 3a-d

show, by way of example, three individual image shots and a result image obtained therefrom;

Fig. 4

Figure 4 illustrates image data preprocessing of multiple frames using edge detection and combination into a result image;

Fig. 5

Figure 4 illustrates image data preprocessing of multiple frames by means of pixel-wise median formation;

Fig. 6

outputs the detection range of that camera Fig. 1 again, which captures the control panel from above;

Fig. 7a

shows the installation situation of the camera integrated in the card slot;

Fig. 7b

outputs the detection range of this camera Fig. 7a again;

Fig. 8

shows a block diagram for a connected to a plurality of cameras data processing unit and associated video surveillance unit;

Fig. 9

illustrates the performance of frame captures according to a predeterminable bracketing series; and

Fig. 10a) -c)

show different function curves in the form of sloping and / or rising ramps.

The Fig. 1 shows in a perspective view of the basic structure of a self-service terminal in the form of an ATM ATM. To the control panel of the ATM ATM include in particular a cash dispenser 1, also called shutter, and a keyboard 2, ie controls on which preferably manipulation attempts, eg in the form of superstructures, may occur for the purpose of skimming. The ATM ATM is equipped with several cameras for detecting such and similar manipulation attempts.

The Fig. 1 shows first those cameras, which are mounted at different locations, preferably in the vicinity of the control panel. These are a side camera CAMS, a top view camera CAMD and an additional portrait camera CAMO.

The cameras CAMS and CAMD are within a demarcation, framing or the like and are mounted there. Each of these cameras CAMS or CAMD detects in each case from the outside at least one of the elements arranged in the control panel of the ATM, eg the cash dispenser 1 (shutter) and / or the keyboard 2. The lateral camera CAMS preferably detects exactly these two elements 1 and 2; the top view camera CAMD also detects other elements (see also Fig. 6 ). On the other hand, a camera CAMK integrated in the card input hopper 4 also detects the interior of this element. This camera CAMK and her Function will be explained later in detail on the basis of Fig. 7a / b described.

In addition to the camera arranged directly on or in the control panel, the additional camera CAMO is located in the upper housing part of the ATM ATM and is directed to the area in which the user is operating the ATM. In particular, this camera captures CAMO the head or the face of the user and is therefore referred to here as a portrait camera.

The FIG. 2 shows the detection range of the camera CAMS, which is located in a side housing part, which framing the control panel of the ATM ATM or surrounds. In particular, this camera CAMS is equipped with a wide-angle lens in order to capture at least these two elements or subregions of the control panel. The ATM ATM is designed so that said elements 1 and 2 preferably have the most homogeneous surfaces with these delimiting edges. This simplifies object recognition. By attaching the camera CAMS at this particularly suitable position, the said subareas or elements 1 and 2 can be measured very reliably optical. It can be provided that the camera is focused in particular on certain areas.

Another perspective, namely that of the top view camera CAMD, is based on the FIG. 6 clarified. Here is the detection field of this camera CAMD illustrated, which is installed in the upper part of the ATM ATM (see also Fig. 1 ) and which captures the control panel from above. Next to the Cash dispenser 1 and the keyboard 2, in the detection range of the camera also other elements may be provided, such as a shelf near the keyboard, a card input funnel 4, ie the feeding part for the card reader, and eg a screen 5 or display. These further mentioned elements 3, 4 and 5 represent potential targets for manipulation attempts.

In particular, based on the Fig. 3 to 5 the image data pre-processing proposed here is illustrated in the data processing unit (see also FIG Fig. 8 ) a result image or a result image sequence of high quality is calculated from a plurality of single image recordings.

The Fig. 3a-c show by way of example three of the side camera CAMS (see FIG. Fig. 2 ) captured at different times frames F1, F2 and F3. From these, a result image R is computed by means of an image data preprocessing which will be described in more detail later Fig. 3d is shown.

As based on the Fig. 3a to 3c As can be seen, each of the still images F1, F2 and F3 contains certain image distortions or aberrations due to, for example, reflection effects, poor ambient light, appearance of foreign objects in the form of persons and / or objects, etc. These are schematic representations which illustrate the illustrate each recording situation. For example, the first still image F1 was taken with incident sunlight, causing disturbing reflections on the surface of the control panel in the area of the cash dispenser. This is illustrated here by a light beam coming from the left. In the frame F2 appears a Person covering the keyboard of the ATM. In frame F3, in turn, a foreign object or a foreign object appears in the background. Thus, each of the individual images for the actual image processing for the detection of manipulation attempts on weak points, but can be largely eliminated by the image data preprocessing described here. The result is a calculated overall image R (s. Fig. 3d ), which reproduces the control panel and the existing controls there as trouble-free and with very high image quality.

The result image R is combined by combining the individual image data, whereby the interfering effects are detected and eliminated by comparing the individual images with one another. For example, many subregions, except for the reflection region, can be utilized from the individual image F1, with the individual image F1 reproducing the surface structure of the housing and the operating elements particularly well. From the single image F2 also many sub-areas, except for the area of the keyboard and the environment in front of the ATM, can be utilized, in which case in particular the edges of the housing and the operating elements are rendered clearly recognizable. The single image F3 also has many useful portions, in which case in particular the keyboard is reproduced without interference.

The result image R can then be calculated from the various subregions and the many image components of the individual images F1 to F3. In contrast to the individual images, the result image does not represent a real image acquisition, but corresponds to an optimally calculated image composition that captures the captured area or the operating elements in a form freed from interference shows. This achieves a very high image quality, which clearly exceeds the quality of the individual images. Thus, an optimal basis for the later actual image data evaluation is created.

For preprocessing the image data, methods known per se from other fields, such as e.g. the so-called exposure blending. In the process, individual images recorded with different exposure times are combined in such a way that over- and under-exposed areas are largely avoided and more details are retained. The individual shots of an exposure bracketing are combined, whereby the brightest parts of an image are replaced by the corresponding points from the next darker one.

As well as on the basis of Figures 9 and 10 is illustrated, the plurality of individual image recordings can be created in dependence on at least one predetermined function, which specifies the different exposure times for the single image recordings. This ensures that no single image recordings are made with the same exposure time, which in turn is advantageous for exposure blending. In the Fig. 9 schematically a series is shown with a plurality of still images F1 to Fn, where it is illustrated that each frame shot has a different exposure time T1, T2, ... Tn. In this case, the row (exposure row) is preferably given in accordance with a monotone decreasing or increasing function, so that the following applies: T1 <T2 <T3 <... <Tn.

• In der Fig. 10a) ist eine erste ansteigende Rampenfunktion MD dargestellt, die in einem unteren Wertebereich Belichtungszeiten vorgibt, so dass beim ersten Einzelbild "1" eine Belichtungszeit T = 0,5 ms eingestellt wird und für die nachfolgenden Einzelbildaufnahmen jeweils längere Belichtungszeiten eingestellt werden. Der untere Wertebereich W1, der für den Tagmodus gilt, geht z.B. bis zu einer max. Belichtungszeit von 1000 ms. In der Fig. 10a) auch als Alternative eine zweite abfallende Rampenfunktion MN dargestellt, die in einem oberen Wertebereich W2 Belichtungszeiten für den sog. Nachtmodus vorgibt, so dass beim ersten Einzelbild "1" eine max. Belichtungszeit von T = 2000 ms eingestellt wird und für die nachfolgenden Einzelbildaufnahmen jeweils kürzere Belichtungszeiten eingestellt werden. Der obere Wertebereich geht z.B. bis zu einer min. Belichtungszeit von T = 1000 ms. Die Entscheidung ob der Tagmodus oder der Nachtmodus gilt, kann auf der Grundlage einer Schwellwertentscheidung erfolgen. Dabei wird der Helligkeitswert und/oder Kontrastwert mindestens einer Einzelbildaufnahme mit dem Schwellwert verglichen. Ist der Helligkeitswert und/oder Kontrastwert größer als der Schwellwert, so gilt der Tagmodus, andernfalls der Nachtmodus.
• Die Fig. 10b) veranschaulicht eine zusammengesetzte ansteigende Rampe, die zunächst entsprechend dem Tagmodus MD Belichtungszeiten im unteren Wertebereich vorgibt. Anschließend werden höhere Belichtungszeiten im oberen Wertebereich entsprechend dem Nachtmodus vorgegeben.
• Die Fig. 10c) zeigt eine ansteigende Rampe, bei der der Übergang von der Tagmodus-Funktion MD zur Nachtmodus-Funktion MN überlappend erfolgt. Viele andere Funktionsverläufe sind denkbar und können an die Gegebenheiten angepasst werden. Im CCTV-Modus werden z.B. 2 bis 4 Bilder pro Sekunde gemacht.

The FIGS. 10a ) -c) illustrate different function curves, each having a specific ramp shape:

• In the Fig. 10a ), a first rising ramp function MD is shown, which predetermines exposure times in a lower value range, so that an exposure time T = 0.5 ms is set for the first single frame "1" and longer exposure times are set for the subsequent single frame exposures. The lower value range W1, which applies to the day mode, goes up to a max. Exposure time of 1000 ms. In the Fig. 10a ) as an alternative, a second sloping ramp function MN shown, which sets in an upper range of values W2 exposure times for the so-called. Night mode, so that the first frame "1" a max. Exposure time of T = 2000 ms is set and for the subsequent single shots each shorter exposure times are set. The upper value range goes up to a min. Exposure time of T = 1000 ms. The decision whether to apply the day mode or the night mode may be based on a threshold decision. In this case, the brightness value and / or contrast value of at least one single image acquisition is compared with the threshold value. If the brightness value and / or contrast value is greater than the threshold value, the day mode applies, otherwise the night mode.
• The Fig. 10b ) illustrates a composite rising ramp which initially sets exposure times in the lower value range according to the day mode MD. Subsequently, higher exposure times are given in the upper value range corresponding to the night mode.
• The Fig. 10c ) shows a rising ramp where the transition from the day mode function MD to the night mode function MN is overlapping. Many other functional sequences are conceivable and can be adapted to the conditions. In CCTV mode, for example, 2 to 4 frames per second are made.

For example, the still images can also be performed depending on lighting conditions. Also, the exposure times may be dependent on various parameters, such as e.g. Location of the ATM (indoor, outdoor), type and / or mounting location of the camera, lighting conditions, etc ..

Also, for example, an edge detection can be used, as this is based on the Fig. 4 to illustrate the schematic representations:

In the Fig. 4 are shown in a first row as sub-figures 4a1) to 4a3) three frame shots F1 'to F3', which at different exposure times of the lateral camera CAMS (s. Fig. 1 and 2 ) have been recorded. This first row reproduces three differently exposed images, namely a1) a very brightly exposed image F1 ', a2) a normally exposed image F2' and a3) an underexposed image F3 '. In a second row are shown as sub-figures 4b1) to 4b3), the respectively obtained by means of edge detection individual images. These edge images shown in b1) to b3) would have to represent white edge progressions on a black background. In order to meet the requirements for patent drawings, these representations are reproduced here in an inverted manner, ie black edge gradients are displayed on a white background.

The same applies to the overall image R 'shown in c). As a comparison of the edge images b1) to b3) with the images a1) to a3) shows, the edge detection of the individual images does not yield an optimal result. According to the invention is now from the data of the individual edge detections, ie from the individual images of the 4b1) to 4b3) , an overall picture R 'calculated that corresponds to a significantly improved edge detection. Overall, all the edges reliably recognized in the overall image or result image R 'are found again, which are not or only partially found in the respective individual images. In addition, artifacts, in particular virtual edges or "dummy edges", could be eliminated.

The Fig. 5a to 5c illustrate a further variant or additional measure for image data preprocessing of single image F1 ", F2", F3 "etc .. Here, the image data is pixelwise subjected to a median formation Fig. 5a ) schematically shows the image data for the first pixel in each frame. For example, the first pixel in the image F1 "has the value" 3 ", in the image F2" the value "7" and in the image F3 "the value" 5. "The next images F4" and F5 "have the first pixel location Value "5" or "4". As the Fig. 4b) 3, 7, 3, 5 and 4, the result for the first pixel is a sequence of the following image data values: a sorting of the values according to their size follows, so that the following sequence results: 3, 3, 4 , 5 and 7. The median of this sequence is thus the value "4". This value is entered in the result image or target image R "at the first pixel location (s. Fig. 4c ). The formation of the median value has the advantage over averaging (the average value would be "4.4") that the moving objects possibly present in individual images are completely eliminated.

The image data processing, which can also take place on the basis of image data of a plurality of cameras, is carried out in a data processing unit which also performs the actual image analysis and which in the Fig. 8 is shown.

The FIG. 8 shows the block diagram of a data processing unit 10 according to the invention, to which the cameras CAMS and CAMK are connected, as well as a CCTV unit 20, which is connected to the data processing unit 10. The data processing unit 10 receives the image data D from the camera CAMS and the image data D 'from the camera CAMK. Both cameras take frames at predeterminable time intervals, the recordings being controlled by a pre-stage or control stage ST. In particular, the respective exposure time is given, so that a series of individual shots (exposure series) is created (see also later description of Fig. 9 and 10 ). Then, in a first stage 11, the preprocessing of the frame data follows. There, among other things result images are created on the basis of the image data processing methods or similar methods already described above. The thus prepared image data D * has a very high quality and is then used as input data for a subsequent second stage 12 which serves for feature extraction. This is followed by a third stage 13 for the classification of the processed input data. In turn, the stage 13 is connected to an interface 14, via which various alarm or monitoring devices can be activated or addressed. These devices include, among other things, image tampering detection (IFD). The first stage 11 in turn, which serves for image preprocessing, is also connected to a second interface 15 via which a connection to the CCTV unit 20 is established. With the help of this CCTV unit, for example, a remote monitoring or remote diagnosis can be performed. The detection of manipulation attempts and the alarming will be described in more detail later.

First, on the Fig. 7a which illustrates a camera installation situation in which the CAMK camera is integrated directly into the card input hopper 4. In order to achieve good image illumination for this camera CAMK, the already used illumination L of the card slot can be used. The camera CAMK is mounted laterally by card slot or insertion slot, which is made of a special, light-conducting plastic K. The lighting L is realized by one or more light sources, such as light-emitting diodes, wherein the generated light is guided via the light-conducting plastic K to the actual insertion slot in order to illuminate it. The light can be guided coming from above and below, so that the card slot is illuminated as evenly as possible. The generated light can be optimally adapted in intensity to the requirements. Also, the light can be colored by the use of colored LEDs and / or color filters to be adapted in particular to the requirements of the camera CAMK.

In order to detect manipulations due to foreign interventions, changes and the like, predefinable subregions are detected and optically measured. This allows deviations from reference values (normal state with regard to image composition, image content, weighting of pixel areas, etc.) quickly and reliably be recognized. For this purpose, different image processing methods (algorithms) or image processing steps (routines) within a later-described data processing unit (s. Fig. 5 ) carried out. The image data processing can take place partially.

The Fig. 7b 1 illustrates the detection area of the camera CAMK segmented into different partial areas, and clearly shows that it is essentially subdivided into three partial areas I, II and III.

The first subarea I primarily detects the inner area of the card input funnel, ie the actual card slot, the subarea III detects the outer area of the card entry funnel and subarea II detects the intermediate transition area. In connection with the Fig. 7a the following advantages of the construction and installation described here are clear:

Due to the internal camera position, in which the camera CAMK is arranged laterally in the card insertion funnel 4 and detects the subregions I to III, various types of skimming modules, superstructures or manipulations can be detected very accurately. Because this method of installation makes it possible to perform an image segmentation according to sub-areas I to III and to measure these individually. The contrast difference between the subregions can be used well for segmenting the image acquisition.

The camera CAMK is here aligned so that the sub-area III also a person in front of the ATM are located (users or attackers) are detected can. This image data can in particular with those of the portrait camera CAMO (s. Fig. 1 ). The camera CAMK is preferably installed on the same side of the terminal as the camera CAMS, so that the image data of these two cameras can also be compared.

In particular, for the interior I, but also for parts of the transition region II, the lighting L (s. Fig. 7a ) are used to achieve the best possible illumination for the image recordings. A colored illumination in the green area is particularly advantageous because the image sensors or CCD sensors of camera are particularly sensitive to green shades and have the greatest resolution. The lighting L improves the object recognition, especially in low light conditions (location, night time, etc.). In addition, the illumination overcomes possibly by external light (eg solar radiation) occurring reflections on a recognizable superstructure. The already provided lighting L of the card insertion funnel is a reliable light source for the camera CAMK. The actual card slot here has a different color than the card input hopper, so that a larger contrast difference is given, which improves the image analysis.

• einer Bildverarbeitungs-Stufe zur Vorverarbeitung der ankommenden Bilder bzw. Daten (z.B. zwecks Schattenentfernung, Kantendetektion, Segmentierung)
• einer Merkmalsextraktions-Stufe (mittels Blobanalyse, Analyse von Kantenposition, Farbverteilung usw.)
• einer Klassifikations-Stufe (zur Bestimmung von Erkennungsmerkmalen für Manipulationen)

In image data processing, various methods are used, in particular a combination of segmentation and edge detection. The data processing unit (s. Fig. 8 ) consists essentially of the following three stages:

• an image processing stage for preprocessing the incoming images or data (eg for the purpose of shadow removal, edge detection, segmentation)
• a feature extraction stage (using blob analysis, edge position analysis, color distribution, etc.)
• a classification level (for determining identification features for manipulations)

The data processing will be later on the basis of Fig. 8 described in more detail and can be realized eg on a PC.

The CAMK camera is designed here as a color camera with a minimum resolution of 400x300 pixels. In the case of saturated illumination, it is thus possible in particular to use a color value distribution-based method for detecting superstructures and the like. The camera CAMK has a wide-angle lens, so that the outdoor area (sub-area III in Fig. 7b ) is detected well.

• Anbringen eines Tastaturüberbaus
• Anbringen eines Komplettüberbaus an der unteren Auflagefläche
• Anbringen eines Überbaus an dem Geldausgabefach (Shutter) und/oder das Anbringen von Gegenständen zur Aufnahme von Sicherheitsinformationen, insbesondere PIN-Nr., wie z.B. Minikameras, Kamerahandys und ähnliche Spähkameras.

In the example described here, at least the cameras mounted in the vicinity of the control panel are CAMS; CAMD and CAMK are connected to the data processing unit 10 (see FIG. 8 ) in order to significantly improve the recognition of manipulations by combining image data. This data processing unit described later makes it possible to optimally evaluate the image data generated by the camera in order to immediately detect a manipulation attempt, such as a superstructure of the keyboard 2 or a manipulation on one of the cameras, and if necessary to trigger alarms and deactivations. By means of the data processing unit described in more detail later, the following manipulations can be reliably detected, among others:

• Attaching a keyboard superstructure
• Attaching a complete superstructure to the lower support surface
• Attaching a superstructure to the cash dispenser (shutter) and / or attaching items for receiving security information, in particular PIN no., Such as miniature cameras, camera phones and similar spy cameras.

For the detection of superstructures within the data processing unit 10 with the aid of the cameras CAMS and CAMD an optical measurement of the detected elements, such as e.g. the keyboard 2, carried out so as to be able to clearly recognize deviations in the case of manipulation. Tests by the applicant have shown that even reference deviations in the millimeter range can be clearly recognized. For example, a combination of edge detection and segmentation can be used for recognition of foreign objects (scouting camera) in order to be able to clearly recognize the contours of foreign objects in the control panel (for example miniature cameras). The required image data processing takes place predominantly in the data processing unit described below.

The FIG. 8 shows the block diagram of a data processing unit 10 according to the invention, to which the cameras CAMS and CAMK are connected, as well as a CCTV unit 20, which is connected to the data processing unit 10. The data processing unit 10 has in particular the following stages or modules:

A pre-stage or control stage ST controls the single-frame images from the cameras to frame data D or D 'to produce, from which then by means of the above-described method preprocessed image data D * can be calculated for the actual data evaluation.

For the actual image data processing and evaluation, a first stage 11 for image processing thereof, a second stage 12 for feature extraction and a third stage 13 for the classification of the processed data are provided. In turn, the stage 13 is connected to an interface 14, via which various alarm or monitoring devices can be activated or addressed. These devices include, among other things, image tampering detection (IFD). The first stage 11, which serves for image processing, is also connected to a second interface 15 via which a connection to the CCTV unit 20 is established. With the help of this CCTV unit, for example, a remote monitoring or remote diagnosis can be performed.

The control stage ST is responsible for the control of the cameras CAMS and CAMK for generating the individual image data D or D '. The subsequent first stage 11 calculates therefrom the prepared image data D * (calculated total image data), wherein in particular measures such as shadow removal, edge detection, vectorization and / or segmentation are performed. The downstream second stage 12 serves the feature extraction, which can be carried out for example by means of a so-called Blobanalyse, an edge positioning and / or a color distribution. For example, blob analysis is used to detect contiguous regions in an image and to make measurements on the blobs. A Blob (Binary Large Object) is an area of adjacent pixels with it logical state. All the pixels in a picture belonging to a blob are in the foreground. All other pixels are in the background. In a binary image, pixels in the background have values that correspond to zero, while every pixel other than zero is part of a binary object.

Subsequently, in the step 13, a classification is made which determines whether or not an enemy manipulation has occurred at the self-service terminal ATM on the basis of the extracted features.

• Abspeichern eines Fotos des Angreifers, wobei sowohl einzelne der Kameras CAMS und/oder CAMK wie auch die zusätzliche Portraitkamera CAMO aktiviert werden können
• Alarmieren der laufenden Geldautomatenapplikation und/oder eines zentralen Managementservers und/oder einer Person, z.B. über E-Mail
• Einleitung von Gegenmaßnahmen, die z.B. das Sperren bzw. Herunterfahren des Geldautomatens
• Übermittlung von Daten, insbesondere von Bildern, der erkannten Manipulation, beispielsweise über das Internet über eine Zentrale

The data processing unit 10 can be realized, for example, by means of a personal computer which is connected to the ATM ATM or which is integrated therein. In addition to the cameras CAMS and CAMK, which detect the mentioned subregions of the control panel, the additional camera CAMO can also be mounted on the ATM ATM (s. Fig. 1 ), which is directed to the user or customer and in particular captures his face. This additional, also known as a portrait camera, camera CAMO can be triggered upon detection of a tampering attack to make a picture of the person located at the ATM. For example, once a skimming attack is detected, the described system can perform the following actions:

• Storing a photo of the attacker, whereby both individual cameras CAMS and / or CAMK as well as the additional portrait camera CAMO can be activated
• Alert the current ATM application and / or a central management server and / or a person, eg via e-mail
• Initiation of countermeasures, such as the locking or shutdown of the ATM
• Transmission of data, in particular images, the detected manipulation, for example via the Internet via a central office

The size and nature of the actions taken or countermeasures can be configured by the operator of the ATM via the system described here.

As described above, several cameras can be provided directly on the control panel, with the cameras CAMS and CAMD detecting the control panel from the outside and the CAMK camera eg detecting the card input funnel from the inside. In addition, an additional portrait camera can be installed (see CAMO in Fig. 1 ). For the actual manipulation detection the cameras CAMS and CAMD are used on the control panel and the camera CAMK in the card input. For the purpose of documenting a manipulation attempt, the portrait camera CAMO is also used.

Preferably, all cameras have a resolution of at least 2 megapixels. The lenses used have a viewing angle of about 140 degrees and more. In addition, the exposure time of the cameras used in a wide range, for example, 0.25 msec. up to 8,000 msec. freely adjustable. This allows adaptation to a wide variety of lighting conditions. Applicant's experiments have shown that a camera resolution of about 10 pixels per degree can be achieved. Based on a distance of one meter, an accuracy of 1.5 mm per pixel can be achieved. This, in turn, means that manipulation starts at one Reference deviation of already 2 to 3 mm can be reliably detected. The closer the camera lens is to the detected element or object, the more accurate the measurement can be. Thus, in closer areas even an accuracy of less than 1 mm can be achieved.

Depending on the area of use of the ATM, e.g. in the outdoor or indoor area, as well as the given lighting conditions, it may be advantageous to mount the camera CAMs rather in the side case of the ATM ATM or in the upper housing area. Also arise depending on the camera position different monitoring options. When monitoring the various elements or subareas, the following is achieved in particular:

The detection of the cash-out tray (shutter) 1 makes it possible to check manipulations in the form of so-called cash-trappers, i. special superstructures. The detection of the keypad makes it possible to determine there manipulation attempts by superstructures or changes to light protection measures and the like. The detection of the support surface makes it possible in particular to detect complete overbuilding. The detection of the card input funnel 4, in particular by a camera integrated therein, makes it possible to detect local manipulations.

It has been shown that even deviations of 2 mm can be clearly recognized, especially on the keypad and on the card input hopper. Deviations at the rear outer edge of the support surface can be detected as early as 4 mm. Deviations at the lower edge of the shutter can already be detected from 8 mm.

To detect manipulations, the data processing unit 10 (see FIG. 8 ) in particular a comparison of the recorded image data D with reference data. In this case, in particular, an image of the outside area can be examined for its homogeneity and compared with the image of the outside area of the control panel camera.

• Unterscheidung von künstlicher und natürlicher Verdunklung: Wird eine Kamera abgedeckt, so steht ihr aufgenommenes Bild im Widerspruch zu den Bildern der anderen Kameras. Fällt das natürliche Licht (Tageslicht) oder das künstliche Licht (Raumbeleuchtung) aus, so ist der Effekt an allen Kameras gleich oder zumindest ähnlich. Andernfalls erkennt das System einen Manipulationsversuch.
• Erkennung von Täuschungsangriffen auf die Kamera-Anordnung, z.B. durch vorgeklebte Fotos: Zeigt eine einzelne Kamera ein anderes Bild (Helligkeit, Bewegung, Farben, insbesondere bezüglich des Außenbereichs), so deutet dies auf einen Täuschungsversuch hin.
• Erhöhung der Robustheit, insbesondere bei Abdeckung des Karteneinschubtrichters: Wird dieser verdeckt, so zeigt die darin integrierte Kamera (s. CAMK in Fig. 4a) ein anderes Bild an (insbesondere bezüglich des Außenbereiches) als die übrigen Kameras (s. CAMS, CAMD in Fig. 1).

The image data of the different cameras CAMS, CAMD and / or CAMK are also compared with one another in order to determine, for example, whether manipulation of individual cameras has occurred. If, for example, the CAMD camera has been covered, the result is a discrepancy with the images of the other cameras. In particular, it can be determined very quickly on the basis of the image brightness whether a darkening occurs only on a single camera, so that a manipulation or cover of this camera is to be assumed. The combination and evaluation of several camera signals or image data increases the robustness of the manipulation monitoring and the avoidance of false alarms. Among other things, the image data or information can also be used as follows:

• Distinction between artificial and natural obscuration: If a camera is covered, its recorded image is in conflict with the images of the other cameras. If the natural light (daylight) or the artificial light (room lighting) fails, the effect is the same or at least similar on all cameras. Otherwise, the system detects a tampering attempt.
• Detection of deception attacks on the camera arrangement, eg by pre-glued photos: shows a single camera another image (brightness, Movement, colors, especially with regard to the outside), this indicates an attempt to deceive.
• Increased robustness, especially when the card slot is covered: If it is hidden, the integrated camera (see CAMK in Fig. 4a ) displays another image (especially with respect to the outside area) than the other cameras (see CAMS, CAMD in Fig. 1 ).

Furthermore, e.g. Also, the environment to be examined for a radiation of lighting for the card entry 4. The connection of the system to the Internet via the interface 23 makes it possible to remotely control the camera or the various cameras. The acquired image data can also be transmitted via the Internet connection to a video server. Thus, the respective camera virtually acts as a virtual IP camera. In particular, the CCTV unit 20 described above is used for such a video surveillance facility, wherein the interface 15 to the CCTV unit is designed for the following functions:

Query an image, adjust frame rate, color model, image resolution, trigger an event in the CCTV service when a new image is provided, and / or possibly visually highlight detected manipulations on a rendered image.

The system is designed so that no false alarms are generated by hands and / or objects in the picture during normal operation (eg withdrawing money, checking account balance, etc.). Therefore, the tamper detection in the period of a normal Machine usage disabled. Also, in time periods in which, for example, a cleaning or a short-term other use (storage of account statements, interactions before and after the start of a transaction) are not used for tamper detection. Essentially, therefore, only rigid and immovable manipulation attempts are preferably analyzed and recognized. The system is designed to work in a wide variety of lighting conditions (day, night, rain, cloudy, etc.). Also, briefly changing light conditions, such as light reflections, shadows and the like, are compensated or ignored during image processing to avoid a false alarm. In addition, technically occurring events, such as the failure of a lighting and the like, are taken into account. These and other special cases are recognized and solved in particular by the third stage for classification.

The method for manipulation detection performed by the described system has in particular the following sequence (see also FIG Fig. 8 ):

First, preprocessed total image data D * are calculated from the original frame data D or D ', which serve as the basis for the actual evaluation of the data.

In this case, a picture is taken in a first step, wherein the camera parameters are adjusted to produce suitable recordings. In particular, a series of images or corresponding image data D or D 'is recorded, which then serve as the basis or reference for the preprocessing.

Subsequently, a further processing of the image data, whereby these are processed so that they are as well suited for the evaluation. For this purpose, for example, several images are combined to form a target image and optimized by means of image enhancement algorithms. In particular, the following steps are carried out:

Shadow removal, removal of moving objects, removal of noise and / or summary of various exposed shots.

The cameras are u.a. set to different exposure times to remove reflections and to collect well-lit areas. Preferably, the images are collected over a predetermined period of time to obtain the best possible output images for manipulation detection. These steps can be performed by means of the recording control ST in the first stage 11. In a further step, a feature extraction (step 12) is carried out for the actual data evaluation, in which image analysis methods are carried out on the preprocessed images or image data in order to check these for certain features, such as e.g. on edge positions or color distributions. For each feature, a number or a value can be specified, which indicates how well the corresponding feature was found again in the viewed image. The values are summarized in a so-called feature vector.

In a further step, a classification is performed (step 13), ie the feature vector is passed to a classification procedure to make a decision to determine whether there is a manipulation or not. It also uses those types of classifiers that can indicate the confidence, ie probability or certainty, of the decision. Classification mechanisms used can be, for example:

Learning Classify Systems, Bayes Classifiers, Support Vector Machines (SVM) or Decision Trees (CART or C 4.5).

The system described herein is preferably modular in design to allow for different configurations. The actual image processing as well as the CCTV connection are realized in different modules (see FIG. 4 ).

The system presented here is also suitable for documenting the detected manipulations or digitally archiving them. In the case of a detected manipulation, the captured images are provided with corresponding meta information, such as. Timestamp, type of manipulation, etc., stored on a hard disk in the system or in a connected PC. Also, for reporting purposes, messages may be forwarded to a platform, e.g. Error messages, status messages (deactivation, mode change), statistics, suspected manipulation and / or alarm messages. In the event of an alarm, a corresponding message with the respective alarm level can be forwarded to the administration interface or the interface. The following options are also implemented at this interface:

Query of camera data, such as number of cameras, construction status, serial number, etc., camera master data or Set camera parameters and / or register for alarms (notifications).

The invention presented here is particularly suitable for carrying out hostile manipulations on a self-service terminal, such as e.g. at an ATM, reliable to recognize. For this purpose, the control panel is continuously and automatically monitored by at least one camera. By means of image data processing, the elements detected by the camera are optically measured in order to detect deviations from reference data. It has been shown that even deviations in the millimeter range can be reliably detected. For the detection of foreign objects, a combination of edge detection and segmentation is preferably used, so that contours of left objects can be clearly recognized and marked. In the case of a manipulation attempt countermeasures or actions can be triggered.

By the combination of several cameras proposed here and the intelligent image data processing, the invention significantly increases the reliability with which manipulations can be detected.

In a preferred embodiment, the invention has the following camera arrangement:

A camera on the card slot, a camera on the control panel, and a camera on the top of the ATM for taking portraits or videos. In addition, the cameras are connected to the described data processing unit. Within the data processing unit are those of the cameras obtained image data or information among other things used as follows:

Recognition or distinction of artificial and natural obscuration: If a camera is covered, its captured image contradicts the images from the cameras. If natural or artificial light fails, this effect is equally noticeable on all cameras. Detection of deception attacks on the camera system, e.g. through pre-clicked photos: If a camera displays a different picture (different brightness, movement, colors, etc.), this indicates an attempt to deceive. Increasing the robustness of the cover detection on the card input hopper: If the card slot is hidden, the integrated camera CAMK displays a different image of the outside area than the other cameras do.

The preprocessing of the camera image data described here, with low-interference or even interference-free overall images calculated from individual images, leads to an increase in the reliability of the subsequent data evaluation for the detection of manipulation attempts and accordingly also serves to avoid false alarms.

In summary, a self-service terminal is proposed, which has at least one camera for detecting tampering attempts, which detects one or more elements provided in the control panel, such as a keyboard, cash dispenser, card slot, and generates image data from a plurality of single image recordings. The at least one camera is connected to a data processing unit, which preprocesses the generated image data (single image data) into a result image. The preprocessed image data of the result image can, for example are calculated by exposure blending from the frame data and provide a very good database for a data analysis for tamper detection.

The present invention has been described using the example of an ATM, but is not limited thereto, but can be applied to any type of self-service terminals.

LIST OF REFERENCE NUMBERS

ATM

Self-service terminal, designed as an ATM, with a control panel CP, which includes:
1 cash dispenser, 2 keypads, 3 shelves,
4 card entry funnels, 5 screens
CAMS camera, mounted on the side of the control panel
CAMK camera, mounted in card slot
CAMD camera, mounted above the control panel
CAMO additional portrait camera

F1, F2, F3

Frames generated by the cameras

D, D '

Frame data

R, R '

Result images (calculated total images)

D *

preprocessed (calculated) image data

10

Data processing unit with:
ST control level (for single-frame shooting)
11 first stage for image preprocessing
12 second stage feature extraction
13 third level for classification
14, 15 interfaces to video surveillance (CCTV)
or image counterfeit detection

20

Video surveillance or CCTV unit with:
21 level for image acquisition, 22 level for compression,
23 interface to an IP network,
24 level for parameterization

Claims (18)

1. Automated teller machine (ATM) having elements (1, 2, 3, 4, 5) that are made available for users of the automated teller machine (ATM) and located in a control panel of the automated teller machine (ATM), wherein at least one camera generating image data is provided for monitoring the automated teller machine (ATM), wherein the at least one camera (CAMS) captures one or more (1, 2) of the elements located in the control panel to detect attempts of manipulating the automated teller machine (ATM) and generates image data (D) of a plurality of individual image recordings (F1, F2, F3), and wherein the camera (CAMS) is connected to a data processing unit (10) which pre-processes the generated image data (D) into a resulting image (R) which serves for detecting manipulation attempts,
   characterized in that
   the data processing unit (10) combines the generated image data (D) of the individual image recordings (F1', F2', F3') into the resulting image (R) using an image data processing operation which effects segmentation and/or edge detection,
   in that the data processing unit (10) segments the individual image recordings (F1, F2, F3) into a plurality of partial regions (I, II, III) that are assigned to the at least one captured element (4) and processes the image data (D') in various ways in segment-wise fashion by way of the data processing unit (10) assembling the resulting image (R) from the partial regions (I, II, III) of different individual image recordings (F1, F2, F3), and
   in that the data processing unit (10) processes the image data (D) from the partial regions (I, II, III) with different image data processing operations and/or with different variants of an image data processing operation.
2. Automated teller machine (ATM) according to Claim 1, characterized in that the at least one camera (CAMS) generates the image data (D) of the individual image recordings (F1, F2, F3) in dependence on prespecifiable criteria, in particular in prespecifiable time intervals and/or at different exposure conditions or ambient brightnesses.
3. Automated teller machine (ATM) according to Claim 1 or 2, characterized in that the at least one camera (CAMS) generates the image data (D) of the individual image recordings (F1, F2, F3) in dependence on events, in particular events captured by said camera or by another camera.
4. Automated teller machine (ATM) according to one of Claims 1 to 3, characterized in that the at least one camera (CAMS) generates the image data (D) of the individual image recordings (F1, F2, F3) in dependence on prespecifiable camera settings, in particular prespecifiable light exposure times and/or image rates.
5. Automated teller machine (ATM) according to one of the preceding claims, characterized in that the data processing unit (10) combines the generated image data (D) of the individual image recordings (F1, F2, F3) into the resulting image (R) by way of an image data pre-processing operation, in particular forming an average, forming a median and/or exposure blending.
6. Automated teller machine (ATM) according to Claim 1, characterized in that the partial regions comprise at least one near or inner region (I) and a surrounding or outer region (III) of the captured element (4), in particular in that one of the partial regions comprises a transitional region (II) between the inner region (I) and the outer region (III) of the element (4).
7. Automated teller machine (ATM) according to one of the preceding claims, characterized in that the data processing unit (10) evaluates the pre-processed image data of the resulting image (R) to detect manipulation attempts using an image processing operation, in particular in that the data processing unit (10) has a first stage (11), receiving the pre-processed image data (D*) of the resulting image (R), for image processing, in particular for shadow removal, edge detection, vectorization and/or segmentation.
8. Automated teller machine (ATM) according to Claim 7, characterized in that the data processing unit (10) has a second stage (12), connected downstream of the first stage (11), for feature extraction, in particular using blob analysis, edge position and/or colour distribution, in particular in that the data processing unit (10) has a third stage (13), connected downstream of the second stage (12), for classification.
9. Automated teller machine (ATM) according to one of the preceding claims, characterized in that the automated teller machine (ATM) has at least one other camera (CAMK), which is mounted at or in the automated teller machine (ATM) within the near region of the control panel and captures at least one of the elements (1, 2, 3, 4, 5) .
10. Automated teller machine (ATM) according to Claim 9, characterized in that the data processing unit (10) pre-processes the image data (D, D') generated by the two or more cameras (CAMS, CAMK) into a resulting image.
11. Automated teller machine (ATM) according to one of the preceding claims, characterized in that the elements, which are located in the control panel and are captured by the at least one camera (CAMS, CAMK) comprise a money dispensing compartment (1), a keypad (2), a placement area (3), a card insertion hopper (4) and/or a screen (5).
12. Automated teller machine (ATM) according to one of Claims 1 to 9, characterized in that, upon detection of an attempt to manipulate the captured elements (1, 2) by way of evaluating the image data (D*) of the resulting image (R), the data processing unit (10) triggers an alarm, locks the automated teller machine (ATM) and/or triggers the additional camera (CAMO).
13. Automated teller machine (ATM) according to one of Claims 1 to 9, characterized in that the respective camera (CAMS) and/or the data processing unit (10) is deactivated during operation and/or maintenance of the automated teller machine (ATM).
14. Automated teller machine (ATM) according to one of the preceding claims, characterized in that the data processing unit (10) produces the plurality of individual image recordings (F1, F2,... Fn) in dependence on at least one prespecifiable function (MD, MN) which indicates different exposure times (T1, T2,... Tn) for the individual image recordings, in particular in that the at least one prespecifiable function (MD, MN) corresponds to at least one ramp function which indicates increasing and/or decreasing exposure times (T1, T2,... Tn) for a series of individual image recordings (F1, F2,... Fn) .
15. Automated teller machine (ATM) according to Claim 14, characterized in that one of the prespecifiable functions (MD) indicates, for a series of individual image recordings (F1, F2,... Fn), different exposure times (T1, T2,... Tn) within a first, lower value range (W1) if a brightness value and/or contrast value of at least one of the individual image recordings exceeds a prespecifiable threshold value.
16. Automated teller machine (ATM) according to Claim 14, characterized in that one of the prespecifiable functions (MN) indicates, for a series of individual image recordings (F1, F2,... Fn), different exposure times (T1, T2,... Tn) within a second, upper value range (W2) if a brightness value and/or contrast value of at least one of the individual image recordings falls short of a prespecifiable threshold value.
17. Automated teller machine (ATM) according to one of Claims 14 to 16, characterized in that the exposure times (T1, T2,... Tn) are dependent on various parameters, in particular the setup location of the automated teller machine (ATM) and/or the type and/or the mounting location of the camera.
18. Automated teller machine (ATM) according to one of Claims 14 to 17, characterized in that the data processing unit (10) produces the individual image recordings in dependence on illumination conditions.

Images