Traffic monitoring device, traffic monitoring system and method

Abstract

The invention relates to a traffic monitoring device (10a; 10b) for detecting vehicles (12a, 14a, 16a, 18a, 20a; 12b, 14b) on a roadway (22a; 22b), with a detection unit (24a; 24b) comprising at least one stereo camera (26a; 26b) with two image sensors (32a, 34a; 32b, 34b) arranged along a baseline (28a; 28b) at a predefined distance (30a; 30b) from each other, which are directed at at least partially overlapping sections of the roadway (22a; 22b), and with an evaluation unit (36a) for evaluating data acquired by the image sensors (32a, 34a; 32b, 34b).To increase efficiency, it is proposed that the image sensors (32a, 34a; 32b, 34b) of the stereo camera (26a; 26b) are configured as event-based image sensors (32a, 34a; 32b, 34b) and each comprise a pixel matrix (38a) with a plurality of pixels (40a, 42a), wherein the pixels (40a, 42a) are each configured to detect relative changes in light intensity independently and asynchronously as events (44a, 46a), wherein the detection unit (24a; 24b) is configured to detect events (44a, 46a, 44a', 46a') occurring simultaneously at identical object points in an overlap area (48a; 48b) of the subsections using both event-based image sensors (32a, 34a; 32b, 34b) to capture the stereo camera (26a; 26b) and to provide it as corresponding events (44a, 46a, 44a', 46a') of an event data set (50a, 52a).

Description

Technical field

[0001] The invention relates to a traffic monitoring device according to the preamble of claim 1, a traffic monitoring system according to claim 10, and a method for monitoring vehicles on a roadway according to the preamble of claim 11. State of the art

[0002] Driving at excessive speed is one of the most frequent causes of accidents. Furthermore, road safety is significantly compromised by other traffic violations, such as following too closely or running red lights. Traffic monitoring devices can therefore contribute to improving road safety. They can also help prevent increased emissions caused by excessive speed. A variety of such devices are already known from the prior art, based on different technologies, such as radar, lidar, induction loops, optical distance-time measuring devices in the form of single-sided sensors, or light barriers. The use of such devices is generally limited to speed detection. Therefore, camera-based devices are increasingly being used to monitor compliance with other traffic regulations.For example, EP 3 053 155 B1 discloses a method for monitoring traffic behavior based on a stereoscopic camera device. However, disadvantages of using stereo cameras with conventional image sensors, such as CMOS or CCD sensors, include the very large amounts of data generated when recording traffic events and a dependence on lighting conditions, which necessitates adjusting the exposure time and makes reliable traffic monitoring, especially at night, difficult.

[0003] The object of the invention is in particular to provide a generic device and a generic method with improved efficiency characteristics. Description of the invention

[0004] The problem is solved according to the invention by the features of the independent claims, while advantageous embodiments and further developments of the invention can be found in the dependent claims.

[0005] The invention relates to a traffic monitoring device for detecting vehicles on a roadway, comprising a detection unit comprising at least one stereo camera with two image sensors arranged along a baseline at a predefined distance from each other, which are directed at at least partially overlapping sections of the roadway, and with an evaluation unit for evaluating data detected by the image sensors.

[0006] It is proposed that the image sensors of the stereo camera are designed as event-based image sensors and each comprise a pixel matrix with a plurality of pixels, wherein the pixels are each designed to independently and asynchronously detect relative changes in light intensity as events, wherein the detection unit is configured to detect events occurring simultaneously at identical object points in an overlap area of ​​the subsections using both event-based image sensors of the stereo camera and to provide them as corresponding events of an event data set.

[0007] This design allows for the advantageous provision of a particularly efficient traffic monitoring device. Since the pixels of the event-based image sensors, which are not triggered by changes in light intensity, are not activated, no events are recorded in static scenes, for example, when no cars are currently driving on a section of road being monitored. Furthermore, only dynamic elements of a scene are captured, while static elements in the background are not recorded. This enables particularly energy-efficient operation of the traffic monitoring device with comparatively low data volumes and simultaneously high data rates. In addition, event-based image sensors are characterized by a very high dynamic range.Therefore, reliable traffic monitoring can be achieved largely independent of lighting conditions and without adjusting exposure times, since event-based image sensors react to relative changes in light intensity regardless of their initial value. Furthermore, the traffic monitoring device according to the invention advantageously enables the simultaneous detection of multiple vehicles.

[0008] The traffic monitoring device according to the invention can be designed as a part, in particular as a sub-assembly, of a traffic monitoring system, which can alternatively also be referred to as a speed measuring system or colloquially as a "speed camera". However, it would also be conceivable for the traffic monitoring device to constitute the entire traffic monitoring system.

[0009] The detection unit of the traffic monitoring device includes at least one stereo camera with two event-based image sensors, but can also include multiple stereo cameras, for example, two, three, four, or more, each with two event-based image sensors. Event-based image sensors differ fundamentally from conventional image sensors, such as CMOS or CCD sensors, which encode image brightness and generate a high volume of data at a fixed frame rate, regardless of scene activity. In contrast, the pixels of the event-based image sensors in the stereo camera are each designed to detect relative changes in light intensity independently and asynchronously as events.Each pixel of an event-based image sensor is self-signaling and designed to respond individually, i.e., independently of other pixels of the event-based image sensor, and preferably in real time, to relative changes in light intensity. The pixels of the event-based image sensors are thus each configured as an independent photodetector. The design of the event-based image sensors is not limited to a specific type of photodetector. Pixels of the pixel matrices of the event-based image sensors can, for example, be configured as photodiodes, photocells, phototransistors, photoresistors, or the like, without being limited to this.

[0010] The pixel matrices of event-based image sensors each have a specific number of rows and columns, so that each pixel within a pixel matrix can be assigned unique coordinates. The pixel matrices are not limited to a specific number of rows and / or columns. For example, currently available event-based image sensors have pixel matrices with 1280 columns and 720 rows. However, within the scope of the invention, the pixel matrices of the event-based image sensors can also have a lower or higher number of rows and / or columns, and thus a lower or higher total number of pixels. This allows, in particular, flexible adaptation of the size and resolution of the stereo camera to different operating conditions, such as the width of the roadway to be monitored and / or the number of lanes to be monitored, and / or the like.Preferably, the pixel matrices of both event-based image sensors of the stereo camera each have the same number of rows and columns, particularly to facilitate easy calibration of the stereo camera. The event-based image sensors are arranged at a predefined distance from each other along the baseline, which is an imaginary straight line extending through the geometric centers of the pixel matrices of both event-based image sensors. Preferably, the predefined distance between the image sensors is variably adjustable, particularly to adapt the depth resolution of the stereo camera to different operating conditions.

[0011] The event-based image sensors each have a dynamic sampling rate, which is preferably at least 0.5 million events per second, advantageously at least 0.6 million events per second, particularly advantageously at least 0.7 million events per second, preferably at least 0.8 million events per second, preferably at least 0.9 million events per second and particularly preferably up to 1.0 million events per second.

[0012] Further information on the functionality of event-based image sensors can be found, for example, in WO 2019 / 129790 A1.

[0013] Preferably, sections of the pixel matrices, and more preferably each individual pixel, are designed to be individually activatable and deactivatable. This advantageously allows for flexible adaptation to individual requirements. For example, individual sections of the image area of ​​the event-based image sensors, which are not relevant for traffic monitoring, can be deactivated to increase the transmission rate. At the same time, this can also further reduce energy consumption and / or data volume. It is also conceivable to selectively deactivate sources of interference in sections of the image area of ​​the event-based image sensors, such as sources of regularly recurring changes in light intensity, like warning lights, traffic lights, construction site lighting, and the like.Furthermore, pixels not aligned with the overlap area of ​​the roadway segments can be deactivated to further increase efficiency. The overlap area describes a subset of the real object points that, in a mounted state of the capture unit, can be simultaneously detected by both event-based image sensors of the stereo camera. The size of the overlap area can be varied depending on the orientation of the event-based image sensors relative to each other.

[0014] Since the event-based image sensors of the stereo camera do not capture events in a clocked manner, no time quantization takes place during the event capture itself. Therefore, the event-based image sensors preferably each have at least one timing unit, in particular at least one clock, for the precise capture of the times of event occurrence. Preferably, the timing units each have a temporal resolution of at least milliseconds, more preferably at least microseconds. Preferably, the event-based image sensors each have a quantization unit, which in turn comprises a timing unit.The quantization units are connected to the pixel matrix and are designed to convert events detected by the pixels and output as analog signals into digital signals, assigning each event at least the coordinates of the pixel that detected it and a unique time. The quantization units can also assign further parameters to the detected events, such as an absolute value of the detected relative change in light intensity and / or a light polarity, indicating whether the detected relative change in light intensity represents an increase or decrease relative to an initial value, and / or similar parameters.Preferably, the quantization units of the event-based image sensors of the stereo camera are synchronized with each other, particularly to reliably provide events simultaneously captured by both image sensors at identical object points in the overlap area as corresponding events. The quantization units can be part of a bus controller for the interfaces of the event-based image sensors. The provision of event data sets by the acquisition unit can include intermediate storage of events, for example in the memory of the quantization units, and / or direct data transmission to the evaluation unit.

[0015] The evaluation unit preferably comprises at least one digital signal processor and at least one buffer memory, preferably a first-in, first-out (FIFO) buffer memory. The buffer memory is configured to temporarily store events detected by the event-based image sensors and provided via their interface in the chronological order of their occurrence and to transmit them to the digital signal processor for further processing. The interfaces of the event-based image sensors are preferably connected to and / or integrated into their respective quantization units. Various bus systems known to those skilled in the art are conceivable as interfaces for the event-based image sensors.

[0016] Theoretically, the event-based image sensors could be configured for direct alignment with the roadway without any additional optical instruments installed in front of them. Preferably, however, the detection unit has a lens for each event-based image sensor, which, when mounted, is positioned in front of the respective sensor in such a way that light can pass through the lens and strike the pixel matrix of the sensor. The lenses are configured to determine the magnification of an image area from the event-based image sensors and the size of a detectable image section. Depending on the requirements, lenses with different focal lengths and / or angles of view can therefore be used.Preferably, the lenses for the two event-based image sensors of a stereo camera in the detection unit are essentially identical to each other, particularly to facilitate easy calibration of the stereo camera. If the detection unit has lenses, it is advantageous to calibrate a distortion correction of the lenses when commissioning the traffic monitoring device, especially to avoid errors in traffic monitoring due to imaging errors in the image areas of the event-based image sensors, which may be caused by lens distortions.

[0017] Furthermore, it is proposed that the evaluation unit be designed to determine the spatial position of at least two corresponding events within an event dataset, based at least on the predefined distance between the event-based image sensors and their mutual orientation. This advantageously enables the determination of vehicle positions on the roadway directly from event data without the need for additional measuring devices. Steps for determining the spatial positions of corresponding events within event datasets, as well as further steps for determining other parameters from event datasets, can be implemented in at least one algorithm executable by the evaluation unit.Possibilities for determining spatial positions of at least two corresponding events of an event data set based at least on the predefined distance of the event-based image sensors and their mutual orientation are described in more detail in the embodiments of the invention.

[0018] Furthermore, it is proposed that the evaluation unit be designed to determine the spatial positions of at least two corresponding events in sequentially recorded event datasets and to calculate the speed of at least one vehicle on the roadway from this. This would advantageously enable particularly efficient speed measurement.The determination of vehicle speeds based on previously determined spatial positions of corresponding events can also be implemented in an algorithm executable by the evaluation unit and, for example, be carried out in the unit pixels per second, which can then be converted into a vehicle speed in a desired unit, such as kilometers per hour or miles per hour, whereby conversion factors can be stored, for example, in a lookup table, in particular in a memory of the evaluation unit.

[0019] Furthermore, it is proposed that the evaluation unit be configured to assign at least two corresponding events from at least one event data set to each of at least two vehicles traveling on the roadway and to determine the distance between the vehicles. Such a configuration advantageously enables the detection of traffic offenses related to failure to maintain minimum distances. It also advantageously allows for the detection of traffic jams or slow-moving traffic based on small distances between vehicles combined with low recorded vehicle speeds. Preferably, the evaluation unit is configured to assign at least two corresponding events from at least one event data set to each of at least two vehicles traveling in the same direction on the roadway and to determine the distance between the vehicles in that direction of travel.The evaluation unit is not limited to determining the distance between vehicles in the direction of travel, but can alternatively or additionally be designed to determine distances between vehicles at angles, particularly perpendicular to their directions of travel. Monitoring distances between vehicles that do not extend along their direction of travel can be used, for example, at bottlenecks, such as in construction zones or tunnels, which require staggered driving on narrowed, adjacent lanes. In principle, the detection unit is also not limited to detecting vehicles traveling in the same direction, but can be designed to simultaneously detect vehicles on opposite carriageways using at least one stereo camera.Similarly, the evaluation unit can be designed to determine parameters such as the positions, speeds, distances, and / or the like of several vehicles, particularly those traveling partially in opposite directions, from one or more sequentially recorded event data sets. The evaluation unit can include an outline recognition algorithm, particularly for determining distances between vehicles and / or for determining other parameters. Using this algorithm, outlines of vehicles or parts of vehicles, such as the front and / or rear of a vehicle, and / or the like, can be identified and assigned from event data, for example, based on comparative data stored in a database of the evaluation unit.Alternatively or additionally, it is conceivable that outline recognition could be based on a spatial arrangement and / or sequence of detected events. For example, several events that were simultaneously detected by adjacent pixels within a row, while no events were detected by the pixels in the row below or above at the same time, could be used to detect the front or rear of a vehicle, depending on the direction of travel in the monitored section of the roadway.

[0020] Furthermore, it is proposed that the evaluation unit be designed to verify compliance with at least one traffic regulation valid at a given time of recording, based on at least one characteristic value derived from at least one event data set for at least one vehicle. This advantageously provides a particularly flexible traffic monitoring device. In addition to monitoring speed limits and minimum distances, it can also enable, but is not limited to, monitoring of overtaking bans and / or red light phases, and the like.

[0021] Furthermore, it is proposed that the evaluation unit be designed to determine the vehicle class of at least one vehicle on the roadway from at least one event data set comprising a plurality of corresponding events. This advantageously enables a more detailed recording and analysis of various traffic situations. A classification of vehicle classes that can be determined by the evaluation unit can, but is not limited to, include, for example, two-wheelers, passenger cars, passenger cars with trailers, buses, trucks, semi-trailer trucks, and the like. The determination of vehicle classes from event data by the evaluation unit can, for example, be carried out using the outline recognition algorithm described above.

[0022] Furthermore, it is proposed that the evaluation unit be trained to take the previously determined vehicle class into account when verifying compliance with traffic regulations. This can advantageously increase the accuracy of traffic monitoring. For example, compliance with traffic regulations that only apply to certain vehicle classes, such as speed limits and / or overtaking bans and / or driving bans on Sundays and public holidays for trucks or similar vehicles, can be verified.

[0023] In an advantageous embodiment of the invention, it is proposed that the detection unit be arranged in a mounted state above the roadway, with its baseline oriented at least substantially perpendicular to a longitudinal extent of a section of the roadway. This advantageously ensures particularly reliable vehicle detection. "At least substantially perpendicular" here refers to an angle between 85° and 95°, preferably between 88° and 92°, and particularly advantageously 90°. The section of the roadway is preferably substantially straight, such that its longitudinal extent corresponds to the longest edge of an imaginary geometric cuboid that just completely encloses the section of the roadway.The detection unit, in its assembled state, could be attached, for example, to a support structure extending over the roadway, or to a bridge, tunnel, or the like.

[0024] In an alternative advantageous embodiment of the invention, it is proposed that the detection unit, in its mounted state, be arranged elevated next to the roadway, with its baseline oriented at an angle to a longitudinal extension of the roadway. An arrangement of the detection unit in its mounted state could, for example, be a temporary arrangement of the detection unit on a tripod or a similar frame. It is also conceivable that an arrangement of the detection unit in its mounted state could be a fixed arrangement of the detection unit within or on a permanently installed housing. Such an embodiment advantageously increases flexibility. In particular, it allows for flexible arrangement of the detection unit at various traffic hotspots. The detection unit could, in its mounted state, be arranged in a fixed, elevated position next to the roadway.Alternatively, the detection unit could be designed as a mobile unit for temporary placement at various locations alongside the roadway. For this purpose, the traffic monitoring unit could include a support and / or stabilizing element, such as a tripod or similar device, for the temporary installation and adjustment of the detection unit. The traffic monitoring unit could also include an energy storage device, in particular a battery, such as a lithium-ion battery, or similar device, for an independent power supply to the detection unit.

[0025] The invention further relates to a traffic monitoring system with at least one traffic monitoring device according to one of the previously described embodiments and with at least one camera for capturing license plates and / or vehicle interiors of vehicles on the roadway. Such a traffic monitoring system is characterized in particular by the advantageous properties of the traffic monitoring device described above. The traffic monitoring system could comprise several cameras, each of which could be directed towards a lane of the roadway. The camera(s) can additionally include a flash. Capturing the interior of a vehicle can be used in particular for identifying the driver. The traffic monitoring system can include a communication unit for wireless communication between the evaluation unit and the camera(s).The communication unit can be configured for wireless, preferably bidirectional, communication using a suitable wireless communication standard, such as Bluetooth, BLE, WLAN, WPAN, infrared, ISM 433 MHz, ISM 868 MHz, or similar. The evaluation unit can be configured to send a trigger signal to the camera to capture the license plate and / or the vehicle interior when a traffic regulation is violated. It is also conceivable that the images captured by the camera can be wirelessly transmitted via the communication unit to an external storage unit, for example, to allow for redundant storage and / or to relieve the burden on local memory in the camera. The external storage unit can, for example, be connected to or integrated into the evaluation unit, or be part of an external data server connected to the communication unit via the internet.

[0026] The invention further relates to a method for monitoring vehicles on a roadway, wherein two image sensors of at least one stereo camera are arranged along a baseline at a predefined distance from each other, are directed towards overlapping sections of the roadway, data are captured by means of the image sensors and subsequently evaluated.

[0027] It is proposed that event-based image sensors be used as image sensors, each comprising a pixel matrix with a plurality of pixels. The pixels are configured to independently and asynchronously detect relative changes in light intensity as events. Prior to data acquisition, the event-based image sensors are configured such that simultaneously occurring events at identical object points within an overlapping area of ​​the sub-sections are detected by both event-based image sensors and provided as corresponding events in an event data set. This advantageously provides a particularly efficient method for traffic monitoring. Preferably, the method is implemented using the traffic monitoring device described above.The method includes, in particular, a configuration step in which the event-based image sensors of at least one stereo camera are arranged along the baseline at a predefined distance from one another and aligned with overlapping sections of the roadway, enabling the simultaneous detection of events occurring at identical object points within the area of ​​intersection of the sections. The method further includes, in particular, a subsequent detection step in which relative changes in light intensity are detected independently and asynchronously by the pixels of the event-based image sensors as events, and events occurring simultaneously and detected by both event-based image sensors are provided as corresponding events in an event data set.The method preferably also includes an evaluation step, occurring after the data acquisition step, for evaluating event data records. The evaluation step may comprise several sub-steps. Preferably, the evaluation step of the method is implemented in at least one algorithm, which is stored in the evaluation unit of the traffic monitoring device.

[0028] In an advantageous embodiment of the method, it is proposed that, particularly in a sub-step of the evaluation step of the method, a spatial position of at least two corresponding events of an event data set is determined based at least on the predefined distance between the event-based image sensors and their mutual orientation. This advantageously provides an efficient method for determining the position of vehicles on the roadway based on event data, which does not require any additional distance measurements.

[0029] Furthermore, it is proposed that, particularly in a sub-step of the evaluation phase of the procedure, the spatial positions of at least two corresponding events in temporally sequentially recorded event datasets be determined, and the speed of at least one vehicle on the road be calculated from this. This would advantageously provide a particularly efficient method for speed measurement.

[0030] Furthermore, it is proposed that, particularly in a sub-step of the evaluation phase of the procedure, at least two vehicles traveling on the road should each be assigned at least two corresponding events from at least one previously recorded event data set, which comprises a plurality of corresponding events, and that a distance between the vehicles should be determined from this. Such a design of the procedure advantageously enables the recording of traffic offenses related to failure to maintain minimum distances, as well as the detection of congestion or slow-moving traffic based on small distances between vehicles combined with low recorded vehicle speeds.

[0031] Furthermore, it is proposed that, particularly in a sub-step of the evaluation phase of the procedure, the vehicle class of at least one vehicle on the roadway be determined from at least one previously recorded event dataset, which comprises a plurality of corresponding events. This advantageously provides a procedure for the particularly detailed evaluation of various traffic situations. In addition to verifying compliance with traffic regulations that apply only to certain vehicle classes, the procedure can also be used, for example, to record statistical parameters, such as the number of vehicles of a specific vehicle class using the roadway monitored by the procedure within a specific period, and similar data.

[0032] Individual steps of the procedure can be implemented in a computer program. The computer program can include instructions which, when executed by a computer, cause it to perform at least one, several, or all of the sub-steps of the evaluation step of the procedure.

[0033] The traffic monitoring device, the traffic monitoring system, and the method are not to be limited to the applications and embodiments described above. In particular, the traffic monitoring device and / or the traffic monitoring system may, to achieve a functionality described herein, comprise a different number of individual elements, components, and / or units than specified herein. The method may comprise a different number of process steps than specified herein. Furthermore, for the value ranges specified in this document, values ​​within the stated limits are also to be considered disclosed and freely usable. Brief description of the drawings

[0034] The invention is explained in more detail below by way of example with reference to the drawings. The drawings show two embodiments of the invention. The drawings, the description, and the claims contain numerous features in combination. A person skilled in the art will expediently consider the features individually and combine them into meaningful further combinations.

[0035] They show: Fig. 1 a schematic representation of a traffic monitoring system with a traffic monitoring device for detecting vehicles on a roadway and with a camera for detecting license plates and / or vehicle interiors of vehicles on the roadway, Fig. 2 a schematic view of the traffic monitoring device with a detection unit comprising two event-based image sensors, and with an evaluation unit, Fig. 3a simplified graphical representation of an event data set with events that were captured by the event-based image sensors of the acquisition unit in a first time interval, Fig. 4 a simplified graphical representation of another event data set with events that were captured by the event-based image sensors of the acquisition unit in a second time interval, Fig. 5 a schematic diagram illustrating a procedure for monitoring vehicles on a roadway, Fig. 6 a schematic representation of a traffic monitoring system with a traffic monitoring device for detecting vehicles on a roadway and with a camera for detecting license plates and / or vehicle interiors of vehicles on the roadway in an alternative embodiment of the invention, and Fig. 7A schematic diagram illustrating the determination of a spatial position of a real object point based on two corresponding events of an event data set when event-based image sensors are angularly aligned with each other. Ways to implement the invention

[0036] Figure 1 Figure 60a shows a traffic monitoring system 60a in a highly simplified and schematic top view. The traffic monitoring system 60a comprises a traffic monitoring device 10a for detecting vehicles 12a, 14a, 16a, 18a, 20a on a roadway 22a and a camera 58a for detecting license plates and / or vehicle interiors, in particular for identifying the drivers, of the vehicles 12a, 14a, 16a, 18a, 20a on the roadway 22a.

[0037] The traffic monitoring device 10a has a detection unit 24a. The detection unit 24a comprises a stereo camera 26a with two image sensors 32a, 34a arranged along a baseline 28a at a predefined distance 30a from each other. The baseline 28a extends between the geometric centers of the image sensors 32a, 34a. The image sensors 32a, 34a are directed at at least partially overlapping sections of the roadway 22a. The traffic monitoring device 10a also includes an evaluation unit 36a (see...). Fig. 2 ) for evaluating data acquired by the image sensors 32a, 34a. In this case, the image sensors 32a, 34a of the stereo camera 26a are designed as event-based image sensors 32a, 34a. The event-based image sensors 32a, 34a each comprise a pixel matrix 38a with a plurality of pixels 40a, 42a (see Fig. 2). Pixels 40a and 42a are each configured to register relative changes in light intensity independently and asynchronously as events 44a and 46a (see below). Fig. 3 ) to capture. In the present case, the capture unit 24a has a lens 76a, which is arranged in front of the event-based image sensor 32a such that light passes through the lens 76a onto the pixel matrix 38a (cf. Fig. 2) of the event-based image sensor 32a, such that the focal point lies at the geometric center of the event-based image sensor 32a. The detection unit 24a also has a lens 92a, which is arranged in the same manner in front of the event-based image sensor 34a. In this case, the lenses 76a and 92a of the detection unit 24a are essentially identical to each other. Alternatively, however, it would also be conceivable that the lenses 76a and 92a have different focal lengths and / or angles of view. Before the traffic monitoring device 10a is put into operation, a distortion correction of the lenses 76a and 92a is calibrated. The sections of the roadway 22a, towards which the event-based image sensors 32a and 34a are aligned, are defined by the angles of view of the lenses 76a and 92a and are shown in the Figure 1schematically symbolized by dashed lines. The detection unit 24a is configured to detect simultaneously occurring events 44a, 46a, 44a', 46a' at identical object points in an overlap area 48a of the roadway sections 22a (see...). Fig. 3 ) to be captured by means of both event-based image sensors 32a, 34a of the stereo camera 26a and as corresponding events 44a, 46a, 44a', 46a' of an event data set 50a, 52a (cf. Figs. 3 and 4 to provide.

[0038] In this case, the detection unit 24a is arranged in a mounted state above the roadway 22a. The baseline 28a is aligned at least substantially perpendicular to a longitudinal extension 56a of a section of the roadway 22a. The detection unit 24a is arranged above the roadway 22a by means of a support structure. Alternatively, it would also be conceivable, for example, to attach the detection unit 24a to a bridge or tunnel for arrangement above the roadway 22a (not shown).

[0039] Figure 2 The figure shows the recording unit 24a and the evaluation unit 36a of the traffic monitoring device 10a in a highly simplified schematic view.

[0040] In the Figure 2The event-based image sensors 32a and 34a of the acquisition unit 24a are each represented by a pixel matrix 38a with a plurality of pixels 40a and 42a. The event-based image sensors 32a and 34a are essentially identical in design. Therefore, for further description, only the event-based image sensor 32a will be referred to below, although the described features can be applied analogously to the event-based image sensor 34a. For illustrative purposes and to simplify representation, the pixel matrix 38a of the event-based image sensor 32a is shown as an 8x8 pixel matrix with a total of 64 pixels. For clarity, not every pixel is labeled. The pixel matrix 38a has, for example, eight columns 62a and 64a and eight rows 66a and 68a, but for clarity, only two columns and two rows are labeled.Based on columns 62a and 64a and rows 66a and 68a of pixel matrix 38a, unique coordinates can be assigned to each of pixels 40a and 42a. For example, pixel 40a is located in column 64a and row 66a, and pixel 42a is located in column 62a and row 68a. However, pixel matrix 38a could also have a significantly higher or lower number of columns 62a and 64a and / or rows 66a and 68a, and thus a different total number of pixels 40a and 42a, than shown in the example. Figure 2 shown, exhibit.

[0041] The event-based image sensors 32a, 34a of the acquisition unit 24a are each connected to the evaluation unit 36a via interfaces 72a. In this case, the event-based image sensors 32a, 34a each have a quantization unit 70a, which is configured to process the events 44a, 46a (see figure) captured as analog signals by the pixels 40a, 42a. Figs. 3 and 4) to convert into digital signals and transmit them to the evaluation unit 36a via the respective interfaces 72a. The quantization units 70a each comprise a time measurement unit with a temporal resolution at least in the range of milliseconds. If a relative change in light intensity, which is detected by at least one of the pixels 40a, 42a of the event-based image sensors 32a, 34a, exceeds a previously defined threshold value, this is recorded as an event 44a, 46a, whereby the respective quantization unit 70a determines the coordinates of the respective pixel 40a, 42a that detected the event 44a, 46a, the exact time of the event 44a, 46a, as well as the polarity of the event 44a, 46a, and transmits them to the evaluation unit 36a.

[0042] The evaluation unit 36a comprises a digital signal processor and a buffer memory (not shown) which is connected upstream of the digital signal processor. The buffer memory is designed as a first-in, first-out (FIFO) buffer memory, so that the events 44a and 46a can be processed by the digital signal processor in the order in which they occur.

[0043] Figure 3 Figure 1 shows a simplified graphical representation of an event data set 50a, which was recorded by the event-based image sensors 32a, 34a within a first time interval. In the Figure 3 The image area 74a of the event-based image sensor 32a with events 44a, 46a and the image area 74a' of the event-based image sensor 34a with corresponding events 44a', 46a' are shown. For clarity, the Figure 3In image areas 74a, 74a', only two events 44a, 46a, 44a', 46a' are provided with reference signs.

[0044] The evaluation unit 36a is designed to determine a spatial position of at least two corresponding events 44a, 44a' of the event data set 50a based at least on the predefined distance 30a (cf. Fig. 1) of the event-based image sensors 32a, 34a and their mutual orientation. In the present embodiment, the optical axes of the event-based image sensors 32a, 34a are aligned parallel to each other and parallel to the longitudinal extent 56a of the roadway section 22a. The spatial position of the at least two corresponding events 44a, 44a' can be determined by the evaluation unit 36a using an algorithm executable by the digital signal processor. A horizontal pixel distance 78a of event 44a and a horizontal pixel distance 78a' of the corresponding event 44a' can be used for this purpose. The horizontal pixel distances 78a, 78a' are derived from the respective coordinates of pixels 40a, 42a (see Figure 1). Fig. 2), which captured the events 44a, 44a'. From a difference in the horizontal pixel distances 78a, a disparity D of the events 44a, 44a' in the image areas 74a, 74a' can be determined. A distance Z between a real object point that triggered the events 44a, 44a', and the baseline 28a (cf. Figure 1 ) can be determined using the following formula (1): Z = f × T d × D

[0045] In formula (1) f represents the focal length of the lenses 76a arranged in front of the event-based image sensors 32a, 34a (see Figure 1 ), d for the pixel width of pixels 40a, 44a (cf. Figure 2 ) and T for the predefined distance 30a, in which the event-based image sensors 32a, 34a are arranged at intervals along the baseline 28a (see Fig. 1 ).

[0046] Figure 4Figure 1 shows a simplified graphical representation of another event data set 52a, which was recorded by the event-based image sensors 32a, 34a within a second time interval following the first time interval. In the Figure 4 The image area 80a of the event-based image sensor 32a with events 44a, 46a and the image area 80a' of the event-based image sensor 34a with corresponding events 44a', 46a' are shown. For clarity, the following are also shown in the Figure 4 In image areas 80a, 80a', only two events 44a, 46a, 44a', 46a' are provided with reference signs.

[0047] The evaluation unit 36a is designed to determine spatial positions of at least two corresponding events 44a, 44a' in temporally successive event data sets 50a, 52a and to derive from this a speed of at least one vehicle 12a, 14a, 16a, 18a, 20a on the roadway 22a (cf. Fig. 1to determine.

[0048] A summary of Figures 3 and 4 shows that events 44a, 46a, 44a', 46a' in image areas 74a, 74a' of the further event data set 52a in Figure 4Compared to events 44a, 46a, 44a', 46a' in image areas 74a, 74a' of the previously recorded event data set 50a, they are shifted downwards in the vertical direction. It should be noted that events 44a, 46a, 44a', 46a' in event data sets 50a, 52a are each independent events, each captured by different pixels 40a, 42a of the event-based image sensors 32a, 34a. The evaluation unit interprets these events as related, for example, based on their underlying relative changes in light intensity and their relative positions to each other. The triggers for each of these events are attributable to the same moving real object points.The evaluation unit 36a is designed to determine the speed of at least one vehicle 12a, 14a, 16a, 18a, 20a on the roadway 22a by means of a distance-time calculation of the spatial positions of related events 44a, 46a, 44a', 46a' previously determined from both temporally successive event data sets 50a, 52a.

[0049] The evaluation unit 36a is further designed to assign at least two corresponding events 44a, 46a, 44a', 46a' from at least one event data set 50a, 52a to each of at least two vehicles 12a, 14a traveling on the roadway 22a and to determine a distance 54a between the vehicles 12a, 14a from this (cf. Figure 1The evaluation unit 36a can, for example, include an algorithm for outline recognition. Using this algorithm, the outlines 82a, 82a', 84a, 84a', which are drawn as examples in image areas 80a, 80a', can be recognized and assigned to the event data record 52a, for example, based on comparison data stored in a database of the evaluation unit 36a. Thus, the outlines 82a, 82a' could be assigned to the front of a vehicle and the outlines 84a, 84a' to the rear of a vehicle, for example, vehicle 14a (see...). Fig. 1 ) be assignable.

[0050] The evaluation unit 36a is also designed to verify compliance with at least one traffic regulation valid at a given time of recording, based on at least one characteristic value determined from at least one event data set 50a, 52a of at least one vehicle 12a, 14a, 16a, 18a, 20a. This characteristic value may, but is not limited to, be, for example, the current position and / or speed of one or more of the vehicles 12a, 14a, 16a, 18a, 20a and / or the distance 54a between the vehicles 12a, 14a and / or another traffic-relevant characteristic value. Based on at least this one parameter, the evaluation unit 36a could, for example, verify compliance with a maximum speed limit valid in the overlap area at the time of recording and / or compliance with minimum distances between vehicles 12a, 14a, 16a, 18a, 20a and / or the like. In the Figure 1In the traffic scenario shown, the evaluation unit 36a could, for example, determine from the event data sets 50a, 52a, that the determined distance 54a between the vehicles 12a, 14a falls below a minimum distance to be maintained and / or that the speeds of the vehicles 12a, 14a exceed a permissible maximum speed.

[0051] Furthermore, the evaluation unit 36a is designed to determine the vehicle class of at least one vehicle 12a, 14a, 16a, 18a, 20a on the roadway 22a from at least one event data set 50a, 52a, which comprises a plurality of corresponding events 44a, 46a, 44a', 46a'. The determination of vehicle classes can, in turn, be carried out using the outline recognition algorithm and / or based on spatial positions of corresponding events 44a, 46a, 44a', 46a' previously determined from an event data set 50a, 52a, the distances between these spatial positions determined therefrom, and the vehicle lengths and / or widths and / or heights derived therefrom. In the traffic scenario shown in Figure 1, the evaluation unit 36a could, for example, determine the vehicle class "passenger car" of vehicles 12a, 14a and 16a and the vehicle class "truck" of vehicles 18a and 20a based on the event data sets 50a, 52a.

[0052] The evaluation unit 36a is also trained to take the previously determined vehicle class into account when checking compliance with traffic regulations. For example, in the traffic scenario in Figure 1 On the section of road 22a shown, at the time of recording the event data sets 50a, 52a, an overtaking ban for trucks applies, and the evaluation unit 36a could determine, based on the specific vehicle class of vehicle 20a and its position on road 22a, which can be determined from the specific spatial positions underlying corresponding events 44a, 46a, 44a', 46a', that vehicle 20a is violating this overtaking ban.

[0053] Figure 5 shows a schematic process flow diagram to illustrate a procedure for monitoring vehicles 12a, 14a, 16a, 18a, 20a on the roadway 22a (see. Fig. 1), wherein event-based image sensors 32a, 34a are used as image sensors 32a, 34a. The method can be carried out using the traffic monitoring device 10a. The method comprises at least three process steps 86a, 88a, 90a. In a configuration step 86a of the method, the two event-based image sensors 32a, 34a of the at least one stereo camera 26a are arranged along the baseline 28a at the predefined distance 30a from each other and aligned with overlapping sections of the roadway 22a (see ). Fig. 1 ). Before data acquisition, the event-based image sensors 32a, 34a are configured in configuration step 86a such that simultaneously occurring events 44a, 46a, 44a', 46a' at identical object points in the overlap area 48a of the subsections (see Fig. 3) can be detected by means of both event-based image sensors 32a, 34a. In a subsequent detection step 88a of the method, relative changes in light intensity are detected independently and asynchronously by the pixels 40a, 42a of the event-based image sensors 32a, 34a as events 44a, 46a, whereby simultaneously occurring events 44a, 46a, 44a', 46a' detected by both event-based image sensors 32a, 34a are provided as corresponding events 44a, 46a, 44a', 46a' of an event data set 50a, 52a.

[0054] The procedure comprises an evaluation step 90a following the acquisition step 88a, which includes several sub-steps and can be carried out, in particular, by means of algorithms executable by the evaluation unit 36a. In evaluation step 90a, the spatial position of at least two corresponding events 44a; 44a' of an event data set 50a, 52a is determined based on at least the predefined distance 30a between the event-based image sensors 32a, 34a and their mutual orientation, as described above. Figure 3 explained in more detail.

[0055] In evaluation step 90a, spatial positions of at least two corresponding events 44a, 44a' in temporally successive event data sets 50a, 52a are further determined, and from this a speed of at least one vehicle 12a, 14a, 16a, 18a, 20a on the roadway 22a is determined, as previously explained in more detail with reference to Figures 3 and 4.

[0056] In evaluation step 90a, at least two corresponding events 44a, 46a, 44a', 46a' are assigned to at least two vehicles 12a, 14a traveling on the roadway 22a from at least one previously recorded event data set 50a, 52a, and the distance 54a between the vehicles 12a, 14a is determined from this, whereby the assignment of events 44a, 46a, 44a', 46a' to the vehicles 12a, 14a can be carried out, for example, using the outline recognition algorithm described above.

[0057] In evaluation step 90a, the vehicle class of at least one vehicle 12a, 14a, 16a, 18a, 20a on the roadway 22a is determined from at least one previously recorded event data set 50a, 52a, whereby the determination of the vehicle class can again be carried out, as described above, using the outline recognition algorithm and / or based on vehicle lengths and / or vehicle widths and / or vehicle heights that can be determined from the event data sets 50a, 52a.

[0058] In the Figure 6 Another embodiment of the invention is shown. The following descriptions are essentially limited to the differences between the embodiments, whereby with regard to identical components, features and functions, reference is made to the description of the embodiment of the Figures 1 to 5 Reference can be made to. To distinguish the embodiments, the letter a in the reference numerals of the embodiment is used in the Figures 1 to 5by the letter b in the reference numerals of the embodiment of the Figure 6 replaced.

[0059] Figure 6 Figure 60b shows an alternative design of a traffic monitoring system 60b in a highly simplified and schematic top view. The traffic monitoring system 60b comprises a traffic monitoring device 10b for detecting vehicles 12b, 14b on a roadway 22b and a camera 58b for detecting license plates and / or vehicle interiors of the vehicles 12b, 14b on the roadway 22b.

[0060] Analogous to the traffic monitoring device 10a of the first embodiment, the traffic monitoring device 10b comprises a detection unit 24b, which includes a stereo camera 26b with two image sensors 32b, 34b arranged along a baseline 28b at a predefined distance 30b from each other, which are directed at at least partially overlapping sections of the roadway 22b, and an evaluation unit (not shown) for evaluating data acquired by the image sensors 32b, 34b. The image sensors 32b, 34b of the stereo camera 26b are designed, analogous to the first embodiment, as event-based image sensors 32b, 34b, each of which has a pixel matrix with a plurality of pixels (not shown here, cf. Fig. 2 ) include, where the pixels are each designed to register relative changes in light intensity independently and asynchronously as events (not shown here, see below). Figs. 3 and 4) to capture. The capture unit 24b is in turn configured to capture simultaneously occurring events at identical object points in an overlap area 48b of the subsections using both event-based image sensors 32b, 34b of the stereo camera 26b and to provide them as corresponding events of an event data set.

[0061] In contrast to the first embodiment, the detection unit 24b is arranged in a mounted state elevated next to the roadway 22b, with the baseline 28b being oriented at an angle to a longitudinal extension 56b of a section of the roadway 22b.

[0062] Another difference from the first embodiment is that the event-based image sensors 32b, 34b are not aligned parallel to each other, but at an angle to each other. Therefore, when determining the spatial position of at least two corresponding events of an event data set based on the predefined distance 30b between the event-based image sensors 32b, 34b, the evaluation unit of the traffic monitoring device 10b must also take into account the mutual alignment of the event-based image sensors 32b, 34b.

[0063] One in the Figure 7The schematic diagram shown serves to illustrate the determination of the spatial position of a real object point 100b by the evaluation unit based on two corresponding events of an event data set when the event-based image sensors 32b and 34b are angularly aligned with each other. Using an algorithm executable by the evaluation unit, an intersection point 94b between an optical axis 96b of the event-based image sensor 32b and an optical axis 98b of the event-based image sensor 34b can first be determined as a temporary measure. The optical axis 96b passes through an optical center 104b of a lens 76b, which is arranged in front of the event-based image sensor 32b (see figure). Figure 6 ) . The optical axis 98b passes through an optical center 106b of a lens 92b, which is arranged in front of the event-based image sensor 34b (see Figure 6). A distance Z 0 between the intersection point 94b and the baseline 28b can be determined using the following formula (2): Z 0 = T tan Φ

[0064] In formula (2) T represents the predefined distance 30b at which the event-based image sensors 32b, 34b are spaced apart from each other along the baseline 28b and Φ represents an inclination angle 102b between the optical axes 96b, 98b of the event-based image sensors 32b, 34b.

[0065] A distance Z between the real object point 100b, which triggered corresponding events in the event-based image sensors 32b, 34b, and the baseline 28b can then be determined using the following formula (3): Z = f × T d × D + f × T Z 0

[0066] As in the first embodiment based on the Figure 3In further detail, formula (3) explains that the expression D represents the disparity between two corresponding events that were simultaneously detected by one of the event-based image sensors 32b and 34b. Figure 7 A horizontal pixel distance of 108b between the geometric center of the event-based image sensor 32b and a pixel 40b of the event-based image sensor 32b is shown. Figure 7The figure also shows a horizontal pixel distance of 110b between the geometric center of the event-based image sensor 32b and a pixel 42b of the event-based image sensor 34b. Pixels 40b and 42b of the event-based image sensors 32b and 34b captured two corresponding events at the real object point 100b. The difference between the horizontal pixel distances 108b and 110b yields the disparity D. The expression f in formula (3) again represents the focal length of the lenses 76b and 92b arranged in front of the event-based image sensors 32b and 34b. The focal length f is given in the Figure 7 Designated with reference numeral 112b. In the present embodiment, the lenses 76b and 92b have the same focal length 112b. The expression d in formula (3) represents the pixel width of pixels 40b and 42b of the event-based image sensors 32b and 34b.

[0067] By substituting formula (2) into formula (3), it can be simplified as follows: Z = f × T d × D + f × tan Φ

[0068] Therefore, in order for the evaluation unit to correctly determine the distance Z, the inclination angle Φ must be calibrated before the traffic monitoring device 10b is put into operation.

[0069] Analogous to the first embodiment, the evaluation unit of the traffic monitoring device 10b is also configured to determine the spatial positions of at least two corresponding events in temporally successive event data sets and to determine the speed of at least one vehicle 12b, 14b on the roadway 22b from this. Likewise, the evaluation unit of the traffic monitoring device 10b is configured to assign at least two corresponding events from at least one event data set to each of at least two vehicles 12b, 14b traveling on the roadway 22b and to determine the distance between the vehicles 12b, 14b from this.Since the optical axes 96b, 98b of the event-based image sensors 32b, 34b are not aligned parallel to the longitudinal extent 56b of the roadway section 22b in the present embodiment, the angle of inclination between the baseline 28b and the longitudinal extent 56b of the roadway section 22b is taken into account by the evaluation unit when calculating a distance between two spatial positions determined from corresponding events in an event data set or in temporally successive event data sets, and when determining the speeds of vehicles 12b, 14b and / or distances between vehicles 12b, 14b on the roadway 22b. Accordingly, the angle of inclination between the baseline 28b and the longitudinal extent 56b of the roadway section 22b must also be calibrated before the traffic monitoring device is put into operation. Reference symbol list

[0070] 10 Traffic monitoring device 12 Vehicle 14 Vehicle 16 Vehicle 18 Vehicle 20 Vehicle 22 Roadway 24 Detection unit 26 Stereo camera 28 Baseline 30 Predefined distance 32 Event-based image sensor 34 Event-based image sensor 36 Evaluation unit 38 Pixel matrix 40 Pixel 42 Pixel 44 Event 46 Event 48 Overlap area 50 Event record 52 Event record 54 Distance 56 Longitudinal extent 58 Camera 60 Traffic monitoring system 62 Column 64 Column 66 Row 68 Row 70 Quantization unit 72 Interface 74 Image area 76 Lens 78 Horizontal pixel pitch 80 Image area 82 Outline 84 Outline 86 Configuration step 88 Detection step 90 Evaluation step 92 Lens 94 Intersection point 96 Optical axis 98 Optical axis 100 Real object point 102 Tilt angle 104 Optical center 106 Optical center 108 Horizontal pixel spacing 110 Horizontal pixel spacing

Claims

1. Traffic monitoring device (10a; 10b) for detecting vehicles (12a, 14a, 16a, 18a, 20a; 12b, 14b) on a roadway (22a; 22b), comprising a detection unit (24a; 24b) comprising at least one stereo camera (26a; 26b) with two image sensors (32a, 34a; 32b, 34b) arranged along a baseline (28a; 28b) at a predefined distance (30a; 30b) from each other, which are directed at at least partially overlapping sections of the roadway (22a; 22b), and comprising an evaluation unit (36a) for evaluating data acquired by the image sensors (32a, 34a; 32b, 34b), characterized by the fact thatThe image sensors (32a, 34a; 32b, 34b) of the stereo camera (26a; 26b) are configured as event-based image sensors (32a, 34a; 32b, 34b) and each comprise a pixel matrix (38a) with a plurality of pixels (40a, 42a), wherein the pixels (40a, 42a) are each configured to detect relative changes in light intensity independently and asynchronously as events (44a, 46a, 44a', 46a'), wherein the detection unit (24a; 24b) is configured to detect events (44a, 46a, 44a', 46a') occurring simultaneously at identical object points in an overlap area (48a; 48b) of the subsections using both event-based image sensors (32a, 34a; 32b, 34b) to capture the stereo camera (26a; 26b) and to provide it as corresponding events (44a, 46a, 44a', 46a') of an event data set (50a, 52a).

2. Traffic monitoring device (10a; 10b) according to claim 1, characterized by the fact thatthe evaluation unit (36a) is designed to determine a spatial position of at least two corresponding events (44a, 44a') of an event data set (50a, 52a) based at least on the predefined distance (30a; 30b) of the event-based image sensors (32a, 34a; 32b, 34b) and their mutual orientation to each other.

3. Traffic monitoring device (10a; 10b) according to claim 2, characterized by the fact that the evaluation unit (36a) is designed to determine spatial positions of at least two corresponding events (44a, 44a') in temporally successive event data sets (50a, 52a) and to determine from this a speed of at least one vehicle (12a, 14a, 16a, 18a, 20a; 12b, 14b) on the roadway (22a; 22b).

4. Traffic monitoring device (10a; 10b) according to one of the preceding claims, characterized by the fact thatthe evaluation unit (36a) is designed to assign at least two corresponding events (44a, 46a, 44a', 46a') from at least one event data set (50a, 52a) to each of at least two vehicles (12a, 14a; 12b, 14b) traveling on the roadway (22a, 22b) and to determine a distance (54a) between the vehicles (12a, 14a; 12b, 14b).

5. Traffic monitoring device (10a; 10b) according to one of claims 2 to 4, characterized by the fact that the evaluation unit (36a) is designed to verify compliance with at least one traffic regulation valid at a time of recording, based on at least one characteristic value determined from at least one event data set (50a, 52a) of at least one vehicle (12a, 14a, 16a, 18a, 20a; 12b, 14b).

6. Traffic monitoring device (10a; 10b) according to one of the preceding claims, characterized by the fact thatthe evaluation unit (36a) is designed to determine a vehicle class of at least one vehicle (12a, 14a, 16a, 18a, 20a; 12b, 14b) on the roadway (22a, 22b) from at least one event data set (50a, 52a) which comprises a plurality of corresponding events (44a, 46a, 44a', 46a').

7. Traffic monitoring device (10a; 10b) at least according to claims 5 and 6, characterized by the fact that the evaluation unit (36a) is trained to take into account the previously determined vehicle class when checking compliance with traffic regulations.

8. Traffic monitoring device (10a) according to one of the preceding claims, characterized by the fact that the detection unit (24a) is arranged in a mounted state above the roadway (22a), wherein the baseline (28a) is aligned at least substantially perpendicular to a longitudinal extent (56a) of a section of the roadway (22a).

9. Traffic monitoring device (10b) according to any one of claims 1 to 7, characterized by the fact that the detection unit (24b) is arranged in a mounted state elevated next to the roadway (22b), the baseline (28b) being oriented at an angle to a longitudinal extension (56b) of a section of the roadway (22b).

10. Traffic monitoring system (60a; 60b) comprising at least one traffic monitoring device (10a; 10b) according to one of the preceding claims and comprising at least one camera (58a; 58b) for capturing license plates and / or vehicle interiors of vehicles (12a, 14a, 16a, 18a, 20a; 12b, 14b) on the roadway (22a; 22b).

11. Method for monitoring vehicles (12a, 14a, 16a, 18a, 20a; 12b, 14b) on a roadway (22a; 22b), wherein two image sensors (32a, 34a; 32b, 34b) of at least one stereo camera (26a; 26b) are arranged along a baseline (28a; 28b) at a predefined distance (30a; 30b) from each other, directed at overlapping sections of the roadway (22a; 22b), data are acquired by means of the image sensors (32a, 34a; 32b, 34b) and subsequently evaluated, characterized by the fact thatEvent-based image sensors (32a, 34a; 32b, 34b) are used as image sensors (32a, 34a; 32b, 34b), each comprising a pixel matrix (38a) with a plurality of pixels (40a, 42a), wherein the pixels (40a, 42a) are each configured to detect relative changes in light intensity independently and asynchronously as events (44a, 46a), wherein the event-based image sensors (32a, 34a; 32b, 34b) are configured prior to data acquisition such that simultaneously occurring events (44a, 46a, 44a', 46a') at identical object points in an overlap area (48a; 48b) of the subsections are detected by both event-based image sensors (32a, 34a; 32b, 34b) and recorded as corresponding events (44a, 46a, 44a', 46a') of an event record (50a, 52a) are provided.

12. Method according to claim 11, characterized by the fact thata spatial position of at least two corresponding events (44a, 44a') of an event data set (50a, 52a) is determined based at least on the predefined distance (30a; 30b) of the event-based image sensors (32a, 34a; 32b, 34b) and their mutual orientation to each other.

13. Method according to claim 12, characterized by the fact that spatial positions of at least two corresponding events (44a, 44a') in temporally successive event data sets (50a, 52a) are determined and from this a speed of at least one vehicle (12a, 14a, 16a, 18a, 20a; 12b, 14b) on the roadway (22a; 22b) is determined.

14. Method according to any one of claims 11 to 13, characterized by the fact thatfrom at least one previously recorded event data set (50a, 52a), which includes a plurality of corresponding events (44a, 46a, 44a', 46a'), at least two vehicles (12a, 14a; 12b, 14b) traveling on the roadway (22a; 22b) are each assigned at least two corresponding events (44a, 46a, 44a', 46a') and a distance (54a) between the vehicles (12a, 14a; 12b, 14b) is determined from this.

15. Method according to any one of claims 11 to 14, characterized by the fact that from at least one previously recorded event data set (50a, 52a) which includes a plurality of corresponding events (44a, 46a, 44a', 46a'), a vehicle class of at least one vehicle (12a, 14a, 16a, 18a, 20a; 12b, 14b) on the roadway (22a; 22b) is determined.

Images