Method and system for determining camera position

Abstract

The invention relates to a method and system for determining camera position. A set of image event data is determined using a sequence of images recorded by a camera depicting a traffic condition involving a plurality of moving vehicles, where each image event data indicates a respective number of events associated with the moving vehicles occurring during a respective imaging time interval. A set of accident event data is obtained from a database, where each accident event indicates a respective number of events occurring during a respective accident time interval detected by a traffic event detector located at a known detector geographic location. Based on a matching process between the image event dataset and the accident event dataset, it is determined that an event associated with the image event dataset is an event associated with the accident event dataset so that the camera geo-location is associated with the detector geo-location.

Description

Technical Field

[0001] The present invention relates to the geolocation of a camera that records an image sequence depicting a traffic situation involving multiple moving vehicles. Background Art

[0002] Surveillance cameras are typically installed in fixed positions and are used to observe a specific scene, usually as part of a network of multiple cameras. If exact information about the location and orientation of the camera was not provided during installation, or if this information is lost or the camera has been moved, it can be challenging to find out where the camera is located, especially what specific scene it is observing.

[0003] While external signal sources such as Global Navigation Satellite System (GNSS) signals can be used for positioning, these signal sources are not always available or reliable, and also do not provide any details about the scene observed by the camera. For example, if the camera observes a section of road or an intersection in an area with dense intersections such as a city, it would be beneficial if it were possible to determine the exact section of road or intersection it is observing. It would also be beneficial to know from what angle the camera is observing the scene. Therefore, there is a great need for a system that can determine what scene a camera is observing without having to visit the location of the camera.

[0004] U.S. Patent Application Publication 2019 / 0057600 describes an investigation assistance system in which vehicles at a road intersection are monitored. Summary of the Invention

[0005] In view of the above, it is an object of the present invention to provide a system that seeks to alleviate, mitigate or eliminate one or more of the above deficiencies and drawbacks in the art, either alone or in any combination.

[0006] Accordingly, in a first aspect of the present invention, there is provided a method for determining the camera geographical location of a camera. The method includes obtaining an image sequence recorded by the camera. The image sequence depicts a traffic situation involving multiple moving vehicles. Based on the image sequence, determining a plurality of events associated with the moving vehicles. Based on the plurality of events, determining an image event data set. The image event data set may contain one or more image event data. Each image event data indicates the corresponding number of events that occurred during the corresponding imaging time interval T IMG During. Obtaining an accident event data set from a database. The accident event data set may contain one or more accident event data. Each accident event data indicates the corresponding number of events that occurred during the corresponding accident time interval T TE During. The events of the accident event data are events detected by a traffic event detector located at the detector geographical location.

[0007] In other words, the traffic situation can be any scenario of a road including where vehicles travel, such as intersections, roundabouts, road segments, etc. Vehicles can include, but are not limited to, cars, buses, trucks, motorcycles, bicycles, etc. A traffic event can be understood as an event where a vehicle enters or leaves a traffic situation, or changes direction. The accident event data has been collected by traffic event detectors. The traffic event detectors have well-known geographical locations. A traffic detector can be any object or entity that collects information about traffic events at a known geographical location and stores it in a database, such as a person, a camera, a detector built into a traffic light, a detector built into a road, etc. The database can be, for example, a public repository of traffic data managed by a traffic authority, a commercially available database, etc.

[0008] A matching process is performed on the image event data set and the accident event data set. The matching process generates matching parameters. The matching parameters represent a measure of the degree to which the image event data and the accident event data correspond to each other. Based on the matching parameters, it is determined that the event associated with the image event data set is the event associated with the accident event data set. Based on the determination that the event associated with the image event data set is the event associated with the accident event data set, it is determined that the camera geographical location is associated with the detector geographical location.

[0009] The image event data collected by the camera during T IMG is matched with the accident event data from the database, where T TE is similar to T IMG The matching process can be completed in various ways. For example, the matching parameters can be given as a percentage of the overlap between the image event data set and the accident event data set. Then, the accident event data set with the highest percentage of matching parameters can be selected as the match for the image event data set.

[0010] The advantages of this method are at least that it can determine the camera geographical location based on what the camera observes. Using this method means determining the camera geographical location independently of external signal sources such as the Global Navigation Satellite System (GNSS). A major advantage of this method compared to other location determination methods is that, in addition to the location, it can also give information about the scene that the camera is observing. This can be particularly useful in a network including several cameras, where the observations of several cameras can be correlated with each other.

[0011] In various embodiments, the method may include performing image segmentation on at least one image in an image sequence recorded by a camera. A first road segment may be identified from the at least one segmented image. Using the image sequence, a crossing line may be defined to laterally span the first road segment. The crossing line is located at a first road segment position where the first road segment is visible to the camera. Then, it may be determined, based on the image sequence, that a moving vehicle crosses the crossing line. Then, each image event data may be the corresponding number of moving vehicles passing the crossing line at the first road segment position.

[0012] In other words, the crossing line is a virtual line that serves as a tool for determining when a vehicle passes a road segment.

[0013] Each accident event data may be the corresponding number of vehicles passing a traffic event detector located at the geographical location of the detector along the first road segment. A distance offset between the geographical location of the detector and the first road segment position may be determined. Based on the distance offset, a corresponding time offset may be estimated. The time offset may be applied to the image event data set or the accident event data set.

[0014] In various embodiments, the distance offset may be a value from 0 to the distance from the traffic event detector to the next nearest traffic event detector. The time offset may be calculated, for example, based on a known speed limit of the first road segment or, if such information is available, based on the average speed of vehicles on the first road segment. Depending on the known information about the moving direction of the moving vehicle (120), the time offset may be added or subtracted from a plurality of events associated with the moving vehicle (120).

[0015] In various embodiments, the method may include identifying a second road segment from the at least one segmented image. The second road segment may intersect the first road segment at an intersection. In this case, the first road segment position may be defined as the position where the first road segment enters the intersection. The second road segment position may be defined as the position where the second road segment enters the intersection.

[0016] In the case where the traffic event detector and the first road segment are located between two intersections and the camera is located at one of the two intersections, the distance offset may be determined as the distance from the known position of the traffic event detector to the nearest intersection.

[0017] The matching process may include converting at least one segmented image into a two-dimensional top view of the traffic situation. The top view may be rotated so that the matching process is performed for a plurality of rotation positions. Each rotation position yields resulting matching parameters. The matching parameters are a measure of the degree to which the image event data set and the accident event data set correspond to each other.

[0018] In other words, by identifying at least two road segments that form an intersection, the matching process can be improved such that it is more likely to correctly match the image event data with the accident event data. Transforming at least one segmented image into a top view enables the matching process to take into account the rotational position. This transformation enables the determination of the rotation of the camera. This can be particularly useful when the method is used in a camera network, where each camera observes the intersection from its own perspective. In such a camera network, the rotation of the cameras relative to each other can be determined and used as another parameter in the matching process.

[0019] Determining that the camera geographical location is associated with the detector geographical location can include determining the geographical location of the corresponding road segment position. Then, the camera geographical location can be determined based on the geographical location of the corresponding road segment position.

[0020] Determining that the camera geographical location is associated with the detector geographical location can include determining the orientation of the camera based on the geographical location of the corresponding road segment position.

[0021] In this way, the camera geographical location can be determined more precisely based on the road segment position.

[0022] In various embodiments, the method can include determining an additional image event data set based on an image sequence. The additional image event data set can contain multiple additional image event data. Each additional image event data indicates the corresponding number of moving vehicles that cross an additional cross line during an imaging time interval T IMG The additional cross line is defined using the image sequence to laterally span a second road segment and is located at the second road segment position. The additional accident event data set can be obtained from a database. The additional accident event data set can contain multiple additional accident event data. Each additional accident event data indicates the corresponding number of moving vehicles that pass an additional vehicle detector during a traffic event time interval T TE The additional vehicle detector is located at an additional detector geographical location.

[0023] A matching process can be performed on the additional image event data set and the additional accident event data set. Then, the matching process can produce resulting additional matching parameters. The additional matching parameters represent a measure of the degree to which the additional image event data set and the additional accident event data set correspond to each other. Based on the additional matching parameters, it can be determined that an event associated with the additional image event data set is an event associated with the additional accident event data set. Based on determining that an event associated with the additional image event data set is an event associated with the additional accident event data set, it can be determined that the camera geographical location is associated with the additional detector geographical location.

[0024] In other words, image event data and accident event data can be obtained from multiple sections of an intersection. Additional image event data and additional accident event data can be used during the matching process. In this way, the matching process can be further improved to make it more likely to correctly match the image event data with the accident event data. By using the additional image event data and additional accident event data, the camera geographical location can be determined more precisely. Imaging time interval T IMG and traffic event time interval T TE can correspond to or be different from the imaging time interval T as described above IMG and traffic event time interval T TE .

[0025] Another advantage is that the matching process can be completed in a shorter time because the additional image event data makes it more likely to identify good matches.

[0026] In various embodiments, the method can include identifying corresponding vehicle attributes associated with a moving vehicle. Determining that the moving vehicle crosses an intersection line can include determining that a moving vehicle with the identified vehicle attributes crosses the intersection line. Obtaining an accident event data set from a database can include obtaining accident event data associated with a vehicle having the identified vehicle attributes. Thus, the matching process can include matching the vehicle attributes of the image event data with the vehicle attributes of the accident event data.

[0027] The identified vehicle attributes can be any one or more of vehicle color, type of vehicle such as sedan, truck, bus, motorcycle, bicycle, speed of the vehicle, direction of vehicle movement.

[0028] By identifying vehicle attributes and using them during the matching process, the matching process can be further improved to make it more likely to correctly match the image event data with the accident event data, and can be completed in a shorter time.

[0029] In another aspect, a system for determining the camera geographical location of a camera is provided, including a processing circuit configured to obtain an image sequence recorded by the camera. The image sequence depicts a traffic condition involving multiple moving vehicles. The system is further configured to determine a plurality of events related to the moving vehicles based on the image sequence and determine an image event data set. The image event data set can contain a plurality of image event data. Each image event data indicates the corresponding number of events that occurred during the corresponding imaging time interval T IMG period. In addition, the system is configured to obtain an accident event data set from a database. The accident event data set can contain a plurality of accident event data. Each accident event data indicates the corresponding number of events that occurred during the corresponding accident time interval T TEThe corresponding quantity of events occurring during the period. The events of the accident event data are events detected by a traffic event detector located at the geographical location of the detector.

[0030] The system is also configured to perform a matching process on the image event data set and the accident event data set. The matching process generates matching parameters. The matching parameters represent a measure of the degree to which the image event data and the accident event data correspond to each other. The system is also configured to determine, based on the matching parameters, that the event associated with the image event data set is an event associated with the accident event data set. The system is also configured to determine, based on determining that the event associated with the image event data set is an event associated with the accident event data set, that the camera geographical location is associated with the detector geographical location.

[0031] In another aspect, a non-transitory computer-readable storage medium is provided, having instructions stored thereon to cause the system summarized above to perform the steps of the method summarized above.

[0032] These further aspects provide effects and advantages corresponding to those summarized above in connection with the method according to the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The above and other aspects of the present invention will now be described in more detail with reference to the drawings. These drawings should not be considered limiting; rather, they are used for explanation and understanding. The same reference numerals always refer to the same elements.

[0034] Figure 1 The system is schematically illustrated,

[0035] Figure 2a is a flowchart of the method steps performed in the system, and

[0036] Figure 2b is a flowchart of the method steps performed in the system. DETAILED DESCRIPTION

[0037] The present invention will be described more fully hereinafter with reference to the drawings, in which exemplary embodiments of the present invention are shown. However, the present invention may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the present invention to those skilled in the art.

[0038] Now reference will be made to Figure 1 , which schematically shows a system 10 including a processing circuit configured to determine a camera geographical location 102 of a camera 100. As Figure 1 illustrated in, the system 10 may include a processing circuit suitably configured in the form of a processor 12, a memory 14, and an input / output (I / O) unit 16.

[0039] Processor 12 is configured to execute program code stored in memory 14 to perform the functions and operations of system 10.

[0040] Memory 14 can be one or more of a buffer, flash memory, hard disk drive, removable medium, volatile memory, non-volatile storage, random access memory (RAM), or other suitable devices. In a typical arrangement, memory 14 can include non-volatile memory for long-term data storage and volatile memory that serves as system memory. Memory 14 can exchange data with processor 12 via a data bus. There can also be accompanying control lines and an address bus between memory 14 and processor 12.

[0041] The functions and operations of system 10, including embodiments of methods executed in the context of system 10, as will be illustrated below, can be embodied in the form of instructions or executable logic routines (e.g., lines of code, software programs, etc.) that are stored on a non-transitory computer-readable medium (e.g., memory 14) of system 10 and executed by processor 12. Additionally, the functions and operations of system 10 can be a stand-alone software application or can form part of a software application that performs additional tasks related to system 10. The described functions and operations can be considered methods that the corresponding parts of the device are configured to perform.

[0042] In addition, while the described functions and operations can be implemented in software, these functions can also be implemented by dedicated hardware or firmware, or some combination of hardware, firmware, and / or software.

[0043] Camera 100 can form part of system 10 or be an external device and is configured to obtain a sequence of images depicting a traffic situation and provide the sequence of images to processing circuitry, for example, via input / output unit 16.

[0044] System 10 is configured to obtain the sequence of images recorded by camera 100. The sequence of images depicts a traffic situation involving multiple moving vehicles 120. The traffic situation can be any scene that includes a road on which vehicles are traveling (e.g., intersections, roundabouts, road segments, etc.). As Figure 1 illustrated, moving vehicles 120 move along road segments 131, 132, 133, 134. Vehicles can be, but are not limited to, cars, buses, or trucks.

[0045] System 10 is also configured to determine a plurality of events related to moving vehicles 120 based on the sequence of images. Events related to moving vehicles 120 can be, for example, a vehicle entering, leaving the traffic situation, or changing direction within the traffic situation.

[0046] System 10 is also configured to determine an image event data set based on multiple events. The image event data set may include one or more image event data. Each image event data indicates the corresponding number of events that occurred during the corresponding imaging time interval T IMG period.

[0047] System 10 is also configured to obtain an accident event data set from database 160. The database may be, for example, a public repository of traffic data managed by a traffic authority. The accident event data set may include one or more accident event data. Each accident event data indicates the corresponding number of events that occurred during the corresponding accident time interval T TE period. The events of the accident event data are events detected by traffic event detector 111 located at detector geographical location 112. A traffic detector may be any object or entity that collects information on traffic events at a known geographical location, such as but not limited to a person, a camera, a detector built into a traffic light, or a detector built into a road. In some embodiments, each accident event data may be the corresponding number of vehicles 120 passing through traffic event detector 111.

[0048] System 10 is also configured to perform a matching process on the image event data set and the accident event data set. The matching process matches the image event data set and the accident event data set, where the absolute time period of T TE at least partially overlaps with T IMG such that the two sets of data at least partially describe events that occurred during the same time interval.

[0049] The matching process may generate a matching parameter. The matching parameter represents a measure of the degree to which the image event data and the accident event data correspond to each other. System 10 is also configured to determine, based on the matching parameter, that the events associated with the image event data set are events associated with the accident event data set. System 10 is also configured to determine, based on determining that the events associated with the image event data set are events associated with the accident event data set, that camera geographical location 102 is associated with detector geographical location 112.

[0050] Figure 1 Further shown is a non - transitory computer - readable storage medium 15 having instructions stored thereon that cause system 10 to perform the steps of the method shown in FIG. 2.

[0051] In some embodiments, system 10 may be configured to perform image segmentation on at least one image in an image sequence recorded by camera 100. From at least one segmented image, a first road segment 131 may be identified. Using the image sequence, a cross line 135 may be defined to laterally span the first road segment 131. The cross line is located at a first road segment position 1311, at a position where the road segment is visible to camera 100. System 10 may be configured to determine that a moving vehicle 120 crosses the cross line 135 based on the image sequence. Each image event data may be the corresponding number of moving vehicles 120 that cross the cross line at the first road segment position 1311.

[0052] In some embodiments, system 10 is configured to determine a distance offset 139 between a detector geographical location 112 and the first road segment position 1311. Based on the distance offset 139, a corresponding time offset may be estimated. The time offset may be applied to the image event data set or the accident event data set.

[0053] In some embodiments, system 10 may be configured to identify a second road segment 132 from at least one segmented image. The second road segment 132 may intersect the first road segment 131 at an intersection 130. In this case, the first road segment position 1311 may be defined as the position where the first road segment 131 enters the intersection 130. The second road segment position 1321 may be defined as the position where the second road segment 131 enters the intersection 130.

[0054] In some embodiments, the system may be configured to transform at least one segmented image into a two-dimensional top view of the traffic condition. The top view may be rotated so that a matching process may be performed for multiple rotation positions. Each rotation position may generate a matching parameter. In other words, this configuration enables the matching process to take into account the rotation position.

[0055] In some embodiments, transforming at least one segmented image into a two-dimensional top view is performed by using a homography matrix. First, vanishing points are identified in the segmented image. For example, horizontal and vertical vanishing points may be found from markings on road segments or building walls in the vertical direction. The homography matrix may be found from the vanishing points using geometric operations. The homography matrix may also be determined by using a neural network. The homography matrix is applied to at least one segmented image to transform the at least one segmented image into a two-dimensional top view. The homography matrix may be decomposed into various components using, for example, direct linear transformation or singular value decomposition. If the images have different dimensions, the individual components will provide information about the rotation. By including various rotations in the matching process, the rotation that best fits the accident event data may be determined. In this way, the field of view of the camera may also be determined.

[0056] In some embodiments, the system may be configured to determine the geographical location of one or both of the corresponding road segment locations 1311, 1321. The system may also be configured to determine the camera geographical location 102 based on the geographical locations of the corresponding road segment locations 1311, 1321. In some embodiments, the system may be configured to determine the orientation of the camera 100 based on the geographical locations of the corresponding road segment locations 1311, 1321.

[0057] In some embodiments, the system may be configured to estimate the absolute distance between the road segment locations 1311, 1321 and the camera geographical location 102 by using a reference object in the field of view of the camera 100 and performing image analysis on at least one segmented image. The distance from the camera optical center to the object can be determined in terms of the number of pixels. In combination with the position data related to the accident event data, it can be inferred how far the camera is from the reference object and the road segment location.

[0058] In some embodiments, the system may be configured to determine an additional image event data set based on an image sequence. The additional image event data set may contain multiple additional image event data. Each additional image event data may indicate the corresponding number of moving vehicles 120 passing through the additional cross line 136 during the imaging time interval T IMG The additional cross line 136 is defined using the image sequence to laterally span the second road segment 131 and be located at the second road segment location 1321. The system may be configured to obtain an additional accident event data set from the database 160. The additional accident event data set may contain multiple additional accident event data. Each additional accident event data indicates the corresponding number of moving vehicles 120 passing through the additional vehicle detector 151 during the traffic event time interval T TE The additional vehicle detector 151 is located at the additional detector geographical location 152.

[0059] Further, the system may be configured to perform a matching process on the additional image event data set and the additional accident event data set. The matching process may generate resulting additional matching parameters. The additional matching parameters represent a measure of the degree to which the additional image event data set and the additional accident event data set correspond to each other.

[0060] Based on additional matching parameters, the system can be configured to determine that an event associated with an additional image event dataset is an event associated with an additional accident event dataset. Based on determining that the event associated with the additional image event dataset is an event associated with the additional accident event dataset, the system can be configured to determine that the camera geographical location 102 is associated with an additional detector geographical location 152. In such a configuration, image event data and accident event data can be obtained from multiple road segments in an intersection. Subsequently, the additional image event data and the additional accident event data can be used during the matching process.

[0061] In some embodiments, the system can be configured to identify corresponding vehicle attributes associated with a moving vehicle 120. The identified vehicle attributes can be any one or more of vehicle color, type of vehicle such as sedan, truck, bus, speed of the vehicle, and direction of vehicle movement. The system can be configured to determine that the moving vehicle 120 having the vehicle attributes crosses an intersection line 135. When the system obtains an accident event dataset from the database 160, it can be configured to obtain accident event data associated with the vehicle having the vehicle attributes. The system can be configured to include the vehicle attributes of the image event data and the vehicle attributes of the accident event data in the matching process.

[0062] Go to Figure 2a and Figure 2b and continue to refer to Figure 1 which will illustrate a method for determining the camera geographical location 102 of the camera 100. The method includes an obtaining step 202, whereby an image sequence recorded by the camera 100 is obtained. The image sequence depicts a traffic condition involving multiple moving vehicles 120.

[0063] In a determining step 204, a plurality of events associated with the moving vehicle 120 are determined based on the image sequence.

[0064] In a further determining step 206, an image event dataset is determined based on the plurality of events. The image event dataset can include one or more image event data. Each image event data indicates a corresponding number of events that occurred during a corresponding imaging time interval T IMG period.

[0065] In an obtaining step 208, an accident event dataset is obtained from the database 160. The accident event dataset can include one or more accident event data. Each accident event data indicates a corresponding number of events that occurred during a corresponding accident time interval T TEThe corresponding number of events that occurred during the period. The events of the accident event data are events detected by the traffic event detector 111 located at the detector geographical location 112. In some embodiments, each accident event data may be the corresponding number of vehicles 120 passing along the first road through the traffic event detector 111 located at the detector geographical location 112.

[0066] In the matching process step 210, a matching process is performed on the image event data set and the accident event data set. The matching process generates matching parameters. The matching parameters represent a measure of the degree to which the image event data and the accident event data correspond to each other.

[0067] In the matching determination step 212, based on the matching parameters, it is determined that the event associated with the image event data set is the event associated with the accident event data set.

[0068] In the location determination step 214, based on the result of the matching determination step 212, it is determined that the geographical location 102 is associated with the detector geographical location 112.

[0069] The imaging time interval T can be selected IMG and the accident time interval T TE , such that it refers to, for example, a one-hour period at a specific time of the day.

[0070] As Figure 2a shown, in some embodiments, the method may include a distance offset determination step 209, in which a distance offset 139 is determined between the detector geographical location 112 and the first road segment location 1311. Based on the distance offset 139, a corresponding time offset can be estimated. The time offset can be applied to the image event data set or the accident event data set.

[0071] The determination step 204 may include performing image segmentation on at least one image in the image sequence recorded by the camera 100. From at least one segmented image, the first road segment 131 can be identified. The cross line 135 can be defined using the image sequence to laterally span the first road segment 131 and be located at the first road segment location 1311 where the road segment is visible to the camera 100. It can be determined using the image sequence that the moving vehicle 120 crosses the cross line 135.

[0072] In various embodiments, the method may include identifying a second road segment 132 from at least one segmented image. The second road segment 132 may intersect the first road segment 131 at the intersection 130. In this case, the first road segment location 1311 may be defined as the location where the first road segment 131 enters the intersection 130. The second road segment location 1321 may be defined as the location where the second road segment 131 enters the intersection 130.

[0073] The matching step 210 may include transforming at least one segmented image into a two-dimensional top view of the traffic situation. The top view may be rotated so that the matching step 210 is performed for multiple rotation positions. Each rotation position results in resulting matching parameters. The matching parameters are a measure of the degree to which the image event data set and the accident event data set correspond to each other.

[0074] The position determination step 214 may include determining the geographical locations of the corresponding road segment positions 1311, 1321. Then, the camera geographical location 102 may be determined based on the geographical locations of the corresponding road segment positions 1311, 1321.

[0075] The position determination step 214 may include determining the orientation of the camera 100 based on the geographical locations of the corresponding road segment positions 1311, 1321.

[0076] The method may include an additional determination step 236, in which an additional image event data set is determined based on the image sequence. The additional image event data set may contain a plurality of additional image event data. Each additional image event data indicates the corresponding number of moving vehicles 120 that cross the additional cross line 136 during the imaging time interval T IMG The additional cross line 136 is defined using the image sequence to laterally span the second road segment 131 and be located at the second road segment position 1321.

[0077] The method may include an additional obtaining step 238, in which an additional accident event data set is obtained from the database 160. The additional accident event data set may contain a plurality of additional accident event data. Each additional accident event data indicates the corresponding number of moving vehicles 120 that pass by the additional vehicle detector 151 during the traffic event time interval T TE The additional vehicle detector 151 is located at the additional detector geographical location 152.

[0078] The method may include an additional matching step 240, in which a matching process is performed on the additional image event data set and the additional accident event data set. Then, the matching process may result in resulting additional matching parameters. The additional matching parameters represent a measure of the degree to which the additional image event data set and the additional accident event data set correspond to each other.

[0079] The method may include an additional matching determination step 242, in which, based on the additional matching parameters, it is determined that the event associated with the additional image event data set is the event associated with the additional accident event data set.

[0080] The method may include an additional position determination step 244, in which it is determined 244 that the camera geographical location 102 is associated with the additional detector geographical location 152.

[0081] The method may include identifying corresponding vehicle attributes associated with the moving vehicle 120. Determining that the moving vehicle 120 crosses the intersection line 135 may include determining that the moving vehicle 120 having the identified vehicle attributes crosses the intersection line 135.

[0082] The obtaining step 208 may include obtaining accident event data associated with a vehicle having the identified vehicle attributes.

[0083] The matching step 210 may include matching the vehicle attributes of the image event data with the vehicle attributes of the accident event data.

[0084] Those skilled in the art will recognize that the present invention is in no way limited to the above preferred embodiments. On the contrary, many modifications and variations are possible within the scope of the appended claims.

[0085] Moreover, by studying the drawings, the disclosure, and the appended claims, those skilled in the art can understand and realize variations of the disclosed embodiments when practicing the claimed invention.

Claims

1. A method for determining a camera geographic location of a camera, the method comprising: obtaining a sequence of images recorded by the camera, the sequence of images depicting a traffic situation involving a plurality of moving vehicles, determining a plurality of events associated with the moving vehicle based on the sequence of images, Based on the determined plurality of events, an image event data set is determined, wherein each image event data in the image event data set indicates a corresponding imaging time interval T. IMG The corresponding number of events that occurred during the period, An accident event dataset is obtained from a database, wherein each accident event data in the accident event dataset indicates an accident detected by a traffic event detector located at a geographic location of the detector at a corresponding accident time interval T. TE The corresponding number of events that occurred during the period, performing a matching process on the image event data set and the accident event data set to generate a matching parameter representing a measure of the degree to which the image event data and the accident event data correspond to each other, determining, based on the matching parameter, that the event associated with the image event dataset is an event associated with the accident event dataset, Based on determining that the event associated with the image event dataset is the event associated with the accident event dataset, determining that the camera geographic location is associated with the detector geographic location.

2. The method according to claim 1, wherein: Determining a plurality of events associated with the moving vehicle includes: performing image segmentation on at least one image in the sequence of images recorded by the camera, identifying a first road segment from the at least one segmented image, defining an intersection line using the sequence of images, laterally spanning the first road segment and at a first road segment position where the road segment is visible to the camera, The image sequence is used to determine that a moving vehicle crosses the intersection line.

3. The method according to claim 2, wherein: Each accident event data is a corresponding number of moving vehicles passing the traffic event detector located at the detector's geographical location along the first road segment.

4. The method according to claim 3, comprising: determining a distance offset between the detector's geographic location and the first road segment location, The matching process includes estimating a time offset corresponding to the distance offset and applying the time offset to the image event dataset or the accident event dataset.

5. The method according to claim 2, comprising: identifying a second road segment from the at least one segmented image, the second road segment intersecting the first road segment at an intersection, and defining the first road segment position as the position where the first road segment enters the intersection, and The second road segment position is defined as a position where the second road segment enters the intersection.

6. The method according to claim 5, wherein: Determining that the camera geographic location is associated with the detector geographic location includes: determining the geographical location of the corresponding road segment location, The camera geographic location is determined based on the geographic location of the corresponding road segment location.

7. The method according to claim 6, wherein: The matching process includes: transforming the at least one segmented image into a two-dimensional top view of the traffic condition, The top view is rotated and the matching process is performed for a plurality of rotational positions, generating a resulting matching parameter for each rotational position, wherein the matching parameter is a measure of the degree to which the image event dataset and the accident event dataset correspond to each other.

8. The method according to claim 7, wherein: Determining that the camera geographic location is associated with the detector geographic location includes determining an orientation of the camera based on the geographic location of the corresponding road segment location.

9. The method according to claim 6, comprising: An additional image event data set is determined based on the image sequence, wherein each additional image event data in the additional image event data set indicates a time interval T IMG a corresponding number of moving vehicles crossing an additional crossing line during the period, wherein the additional crossing line is defined using the image sequence to laterally span the second road segment and be located at the second road segment position, An additional accident event data set is obtained from the database, wherein each additional accident event data in the additional accident event data set indicates an accident that occurred during the traffic event time interval T TE a corresponding number of moving vehicles passing by additional vehicle detectors located at the additional detectors' geographical locations during the period, performing a matching process on the additional image event dataset and the additional accident event dataset to produce a resulting additional matching parameter, the additional matching parameter representing a measure of the degree to which the additional image event dataset and the additional accident event dataset correspond to each other, determining, based on the additional matching parameter, that the event associated with the additional image event dataset is the event associated with the additional accident event dataset, Based on determining that the event associated with the additional image event dataset is the event associated with the additional accident event dataset, determining that the camera geographic location is associated with the additional detector geographic location.

10. The method according to claim 2, comprising: identifying respective vehicle attributes associated with the mobile vehicle, and wherein: Determining that a moving vehicle crosses the intersection line includes determining that the moving vehicle having the identified vehicle attributes crosses the intersection line, Obtaining the accident event data set from the database comprises obtaining accident event data associated with a vehicle having the identified vehicle attributes, and wherein: The matching process includes matching the vehicle attributes of the image event data with the vehicle attributes of the accident event data.

11. The method according to claim 10, wherein: The specified vehicle attribute is any of the following: Vehicle color, The type of vehicle, such as car, truck, bus, The speed of the vehicle, The direction of the vehicle's movement.

12. A system for determining a camera geographic location of a camera, comprising a processing circuit, the processing circuit being configured to: obtaining a sequence of images recorded by a camera, the sequence of images depicting a traffic situation involving a plurality of moving vehicles, determining a plurality of events associated with the moving vehicle based on the sequence of images, Based on the determined plurality of events, an image event data set is determined, wherein each image event data in the image event data set indicates a corresponding imaging time interval T. IMG The corresponding number of events that occurred during the period, An accident event dataset is obtained from a database, wherein each accident event data in the accident event dataset indicates an accident detected by a traffic event detector located at a geographic location of the detector at a corresponding accident time interval T. TE The corresponding number of events that occurred during the period, performing a matching process on the image event data set and the accident event data set to generate a resulting matching parameter representing a measure of the degree to which the image event data and the accident event data correspond to each other, determining, based on the matching parameter, that the event associated with the image event dataset is an event associated with the accident event dataset, Based on determining that the event associated with the image event dataset is the event associated with the accident event dataset, determining that the camera geographic location is associated with the detector geographic location.

13. A non-transitory computer-readable storage medium having instructions stored thereon to cause the system according to claim 12 to perform the steps according to claim 1.

Images