Classification method and device of image acquisition equipment, electronic equipment and storage medium
By classifying and verifying the background and foreground motion trajectory of the monitoring images acquired by the image acquisition device, the problem of low classification accuracy of image acquisition devices in the prior art is solved, and more efficient and accurate equipment classification is achieved.
Patent Information
- Application Number
- CN202310599587.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-23
- Publication Date
- 2024-11-26
AI Technical Summary
In the prior art, the classification accuracy of image acquisition equipment is not high, especially when the monitoring area is large and the number of equipment is large, resulting in low classification retrieval efficiency and easy to misjudgment.
By obtaining the background image of the monitoring image collected by the image acquisition device, classifying and verifying the device according to the foreground motion trajectory type, outputting classification confirmation reminder information to instruct users to confirm the classification type, and improving classification accuracy.
It improves the classification accuracy of image acquisition equipment, reduces misjudgment, and enhances the efficiency and accuracy of classification retrieval.
Smart Images

Figure CN119027705A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of security monitoring technology, and in particular to a classification method, device, electronic equipment and storage medium for image acquisition equipment. Background Art
[0002] Video surveillance has been widely used in places such as road traffic, shopping malls, stations, and factories. It can use multiple image acquisition devices in the monitoring system to monitor different areas of the monitoring place. When you need to check the situation of a certain monitoring area, you can retrieve the image acquisition devices deployed in the area and view the monitoring data of these image acquisition devices.
[0003] At present, image acquisition devices can be classified and retrieved according to their names or customized labels. However, this method requires manual pre-marking of each type of image acquisition device, or pre-defining a format for the name of each type of image acquisition device, so that the monitoring data of a certain type of image acquisition device can be retrieved and viewed. This results in low efficiency of classification retrieval, especially when the monitoring area is large and the number of image acquisition devices in the monitoring system is large. This problem will be more obvious.
[0004] In order to solve the above problems, in the related technology, the monitoring images collected by the image acquisition device can be compared with the image features, and the monitoring images with similar image features can be automatically classified and filed to improve the classification efficiency. However, the accuracy of this classification method is not high. For example, the counters, cashier counters and service counters in shopping malls have similar counter features. Through automatic classification and filing, the image acquisition devices monitoring the cashier counters and service counters can be easily misjudged as the image acquisition devices monitoring the counters, resulting in inaccurate classification of the image acquisition devices. Summary of the invention
[0005] The present invention provides a classification method, device, electronic device and storage medium for image acquisition equipment, which are used to solve the problem of low accuracy of classification of image acquisition equipment in the prior art.
[0006] The present invention provides a classification method for image acquisition devices, comprising:
[0007] Acquire a first monitoring image captured by an image acquisition device to be classified, and cluster the image acquisition devices to be classified based on a background picture of the first monitoring image to obtain a first classification result;
[0008] For each first image acquisition device of the first classification type in the first classification result, determining a first trajectory type corresponding to the first classification type based on the first monitoring image corresponding to the first image acquisition device;
[0009] In the case where there is a second image acquisition device whose trajectory type is different from the first trajectory type among the first image acquisition devices, determining an image acquisition device whose classification is questionable from among the second image acquisition devices, and outputting classification confirmation reminder information; the classification confirmation reminder information is used to instruct to confirm the classification type of the image acquisition device whose classification is questionable;
[0010] The trajectory type is used to characterize the motion trajectory type of the moving object in the foreground image of the monitoring image acquired by the image acquisition device.
[0011] According to a classification method for an image acquisition device provided by the present invention, determining a first trajectory type corresponding to the first classification type based on the first monitoring image corresponding to the first image acquisition device includes:
[0012] determining a third trajectory type of each image acquisition device based on the first monitoring image corresponding to each image acquisition device in the first image acquisition devices, and clustering all the third trajectory types to obtain a trajectory type clustering result;
[0013] Taking each trajectory type in the trajectory type clustering result as a first target trajectory type, and determining a first proportion corresponding to the first target trajectory type; the first proportion is used to represent the proportion of image acquisition devices containing the first target trajectory type in the first image acquisition device in the first image acquisition device;
[0014] When the first proportion is greater than a first proportion threshold, the first target trajectory type is determined as a first trajectory type corresponding to the first classification type.
[0015] According to a classification method for image acquisition devices provided by the present invention, clustering the image acquisition devices to be classified based on the background picture of the first monitoring image to obtain a first classification result includes:
[0016] Classifying the image acquisition device to be classified according to the type of the moving object in the first monitoring image to obtain a second classification result;
[0017] For each third image acquisition device of the second classification type in the second classification result, the background picture of the first monitoring image corresponding to the third image acquisition device is obtained and the background features of the background picture are extracted; the third image acquisition devices are clustered based on the background features to obtain the first classification result.
[0018] A classification method for image acquisition devices provided by the present invention further includes:
[0019] Determine first label information of each first classification type in the first classification result, and add the first label information to the first image acquisition device of the first classification type;
[0020] The step of determining the image acquisition device whose classification is questionable from the second image acquisition device comprises:
[0021] Determine a first target label from the first label information corresponding to the first classification type;
[0022] Determine a fourth image acquisition device that includes the first target label among the image acquisition devices to be classified;
[0023] Taking each trajectory type of the second image acquisition device as a second target trajectory type, determining a second proportion corresponding to the second target trajectory type; the second proportion is used to represent the proportion of the image acquisition devices in the fourth image acquisition device that include the second target trajectory type in the fourth image acquisition device;
[0024] When the second proportion is not greater than a second proportion threshold, the image acquisition device containing the second target trajectory type in the second image acquisition device is determined as the image acquisition device with questionable classification.
[0025] According to a classification method for an image acquisition device provided by the present invention, determining first label information of each first classification type in the first classification result includes:
[0026] For each first image acquisition device of the first classification type in the first classification result, extract a common term from the name of each image acquisition device in the first image acquisition device, and when the number of image acquisition devices in the first image acquisition device that include the common term is greater than a first quantity threshold, determine the common term as the first label information corresponding to the first classification type; and / or,
[0027] For each first image acquisition device of the first classification type in the first classification result, the geographic location information of each image acquisition device in the first image acquisition devices is clustered, and each type of geographic location information obtained by clustering is used as target geographic location information. When the number of image acquisition devices containing the target geographic location information in the first image acquisition devices is greater than a second quantity threshold, the target geographic location information is determined as the first label information corresponding to the first classification type.
[0028] A classification method for image acquisition devices provided by the present invention further includes:
[0029] When a target image acquisition device is detected, a second monitoring image acquired by the target image acquisition device is acquired; the target image acquisition device includes a newly added image acquisition device or an image acquisition device whose position information has been changed;
[0030] Acquire target background features of the second monitoring image, and determine a fourth trajectory type of the target image acquisition device based on the second monitoring image;
[0031] searching a second target label that matches the target background feature and the fourth trajectory type from a classification information table; the classification information table is used to store the correspondence between background features, trajectory types and label information;
[0032] The second label information of the target image acquisition device is obtained, and when the second label information does not match the second target label, label abnormality prompt information is output.
[0033] According to a classification method for an image acquisition device provided by the present invention, after outputting the classification confirmation reminder information, the method further includes:
[0034] In case of detecting a first instruction to confirm classification, classifying the image acquisition device whose classification is in doubt into the first classification type;
[0035] When a second instruction to adjust the classification is detected, a target classification type is determined, and the image acquisition device whose classification is questionable is classified into the target classification type.
[0036] The present invention also provides a classification device for an image acquisition device, comprising:
[0037] A classification module, used to obtain a first monitoring image acquired by an image acquisition device to be classified, and cluster the image acquisition devices to be classified based on a background image of the first monitoring image to obtain a first classification result;
[0038] a trajectory type determination module, configured to determine, for each first image acquisition device of the first classification type in the first classification result, a first trajectory type corresponding to the first classification type based on the first monitoring image corresponding to the first image acquisition device;
[0039] a classification verification module, configured to, when there is a second image acquisition device whose trajectory type is different from the first trajectory type in the first image acquisition device, determine the image acquisition device whose classification is in doubt from the second image acquisition device, and output classification confirmation reminder information; the classification confirmation reminder information is used to instruct to confirm the classification type of the image acquisition device whose classification is in doubt;
[0040] The trajectory type is used to characterize the motion trajectory type of the moving object in the foreground image of the monitoring image acquired by the image acquisition device.
[0041] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, a classification method for image acquisition devices as described above is implemented.
[0042] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the classification method of the image acquisition device as described in any one of the above is implemented.
[0043] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the classification method of the image acquisition device as described in any one of the above is implemented.
[0044] The image acquisition device classification method, device, electronic device and storage medium provided by the present invention first cluster the image acquisition devices to be classified based on the background picture of the first monitoring image collected by the image acquisition device to be classified to obtain the first classification result of the image acquisition device to be classified; then, for each first image acquisition device of the first classification type in the first classification result, determine the type of motion trajectory of the moving object in the foreground picture of the corresponding first monitoring image, and obtain the first trajectory type corresponding to the first classification type; for different monitoring scenes, different types of motion trajectories can be presented in the foreground picture, and when there is a second image acquisition device with a trajectory type different from the first trajectory type in the first image acquisition device, it indicates that the second image acquisition device may not belong to the first classification type and needs further verification, then further determine the image acquisition device with doubtful classification from the second image acquisition device and output classification confirmation reminder information, and instruct the user to further confirm the classification type of the image acquisition device with doubtful classification through the classification confirmation reminder information. In this way, by further using the motion trajectory type of the foreground moving object of each type of first image acquisition device classified based on the background picture to verify whether the classification result is correct, the image acquisition device with incorrect classification can be verified, thereby improving the accuracy of image acquisition device classification. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0046] Figure 1is a flow chart of a classification method for an image acquisition device provided by an embodiment of the present invention;
[0047] Figure 2 is a schematic diagram of a motion trajectory in which the trajectory type is unidirectional in an embodiment of the present invention;
[0048] Figure 3 is a schematic diagram of a motion trajectory in which the trajectory type is bidirectional in an embodiment of the present invention;
[0049] Figure 4 is a schematic diagram of a motion trajectory in which the trajectory type is multi-directional in an embodiment of the present invention;
[0050] Figure 5 is a schematic diagram of a motion trajectory in which the trajectory type is undirected in an embodiment of the present invention;
[0051] Figure 6 is a schematic diagram of a motion trajectory in which the trajectory type is wandering in an embodiment of the present invention;
[0052] Figure 7 is a schematic diagram of a motion trajectory in which the trajectory type is stationary in an embodiment of the present invention;
[0053] Figure 8 is a structural schematic diagram of a classification device of an image acquisition device provided by an embodiment of the present invention;
[0054] Fig. 9 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0055] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0056] It should be noted that the serial numbers assigned to the objects described in the present invention, such as "first", "second", etc., are only used to distinguish the objects described and do not have any order or technical meaning.
[0057] In a monitoring system, the image acquisition devices in the monitoring system can be automatically classified and archived according to the screen features of the monitoring images collected by each image acquisition device through feature comparison, and image acquisition devices with similar screens can be archived together to achieve classification of image acquisition devices. However, similar screen features may exist in different monitoring scenarios. For example, the counters, cashier counters, and service counters in shopping malls have similar counter-type features. Through automatic classification and archiving, the image acquisition devices of these places will be classified into one category. In this way, when it is necessary to view the monitoring images of the image acquisition devices of the monitoring counters, the monitoring images of the image acquisition devices of the cashier counters and service counters will be retrieved and viewed at the same time, resulting in the viewing of the monitoring images not being the required scene of the place. Therefore, how to ensure the accurate classification of image acquisition devices that are automatically archived and classified based on screen feature comparison is of great significance to the retrieval and viewing of image acquisition devices.
[0058] On this basis, the present invention has found through research that for different monitoring scenes, in the foreground picture of the monitoring image collected by the image acquisition device, the moving objects can present different types of motion trajectories. For example, in a shopping mall, the motion trajectory of people in the counter scene is mainly wandering and directionless, while the motion trajectory of people in the service desk scene is mainly stationary. In this way, the counter scene and the service desk scene with similar picture features can be distinguished by using the type of people's motion trajectory, so as to distinguish the image acquisition devices in these two scenes, and improve the accuracy of image acquisition device classification.
[0059] Based on this, an embodiment of the present invention provides a classification method for image acquisition devices, which can cluster and classify image acquisition devices based on the static background of the monitoring image, and use the dynamic foreground trajectory of the monitoring image to perform classification verification on the image acquisition devices after clustering and classification, to determine whether the clustering and classification of the image acquisition devices is correct, thereby realizing automatic classification of image acquisition devices to be classified in the monitoring system, and improving the accuracy of image acquisition device classification through verification.
[0060] In the embodiment of the present invention, the image acquisition device may include at least one of the following: a camera, a video camera, a still camera, and other devices with a camera function. Among them, other devices with a camera function may include a mobile phone and a tablet computer, etc., but are not limited thereto. It is understandable that in the same monitoring system, one or more different types of image acquisition devices may be included, such as including multiple cameras, or including at least one camera and at least one still camera, etc.
[0061] Combine the following Figure 1-Figure 7The classification method of the image acquisition device of the present invention is described. The classification method of the image acquisition device can be applied to electronic devices such as terminal devices or servers. Among them, the terminal device may include a mobile phone, a computer, a vehicle-mounted device, a tablet computer, a wearable device, etc.; the server may include an independent server, a cluster server or a cloud server, etc. The classification method of the image acquisition device can also be applied to a classification device of an image acquisition device set in an electronic device such as a terminal device or a server, and the classification device of the image acquisition device can be implemented by software, hardware or a combination of both.
[0062] Figure 1 A schematic diagram showing a flow chart of a classification method for an image acquisition device provided by an embodiment of the present invention is shown as an example. Figure 1 As shown, the classification method of the image acquisition device may include the following steps 110 to 130.
[0063] Step 110: Acquire a first monitoring image captured by an image acquisition device to be classified, and cluster the image acquisition devices to be classified based on a background image of the first monitoring image to obtain a first classification result.
[0064] When classifying the image acquisition devices to be classified in the monitoring system, the background separation of the first monitoring image acquired by the image acquisition device to be classified can be performed to obtain the background picture of the first monitoring image, and then the image features of the background picture can be extracted to obtain the background features of the first monitoring image. Then, the background features of the first monitoring image are clustered according to the similarity by feature comparison to obtain the first classification result. For example, for the first monitoring images of any two image acquisition devices in the image acquisition devices to be classified, if the similarity of the background features of the two is greater than the similarity threshold, it is considered that the background features of the two are similar or the same, and the two image acquisition devices can be clustered into the same category.
[0065] Among them, in the process of background separation of the first monitoring image captured by any one of the image acquisition devices to be classified, the moving object in the first monitoring image can be removed, the background picture can be retained, and then the first monitoring image captured by the image acquisition device at different time periods can be used to complete the picture of the position where the moving object is removed, so as to obtain the complete background picture of the image acquisition device. The background picture can reflect the background environment of the area monitored by the image acquisition device. The monitoring images of the same place have similar background environments. In this way, the background environment of the monitoring area where the image acquisition device to be classified is located can be used to classify the image acquisition device to be classified. By removing the moving object from the monitoring image, the interference of the moving object can be avoided, and the accuracy of clustering the image acquisition device to be classified using background features can be improved.
[0066] Exemplarily, the image acquisition devices to be classified in the monitoring system can be deployed in multiple different locations to monitor different areas. Based on this, the image acquisition devices to be classified can be classified according to their geographical location types, and then the image acquisition devices of each geographical location type in the classification results are clustered based on the background features of the first monitoring image they collected to obtain a first classification result. The geographical location type can be determined based on the scope of the place identified by the map data. For example, the geographical location type can include at least one of a scenic spot type, a shopping mall type, a hospital type, and a community type, but is not limited thereto.
[0067] Exemplarily, the image acquisition devices to be classified may be classified according to the types of moving objects in the first surveillance image captured by the image acquisition devices to be classified, and then each type of image acquisition device in the classification result may be clustered based on the background features of the first surveillance image captured by the image acquisition devices to obtain a first classification result. The moving object type may include at least one of no moving object, pedestrians, vehicles, and both pedestrians and vehicles.
[0068] In this way, the image acquisition devices to be classified can be classified at multiple levels, thereby improving the accuracy of classification and the efficiency of image acquisition device retrieval and viewing.
[0069] Step 120: for each first image acquisition device of the first classification type in the first classification result, determine a first trajectory type corresponding to the first classification type based on a first monitoring image corresponding to the first image acquisition device.
[0070] The trajectory type is used to characterize the motion trajectory type of the moving object in the foreground image of the monitoring image acquired by the image acquisition device.
[0071] Exemplarily, for each first image acquisition device of the first classification type, the motion trajectory type of the moving object in the first monitoring image can be determined based on the first monitoring image corresponding to each image acquisition device in the first image acquisition device, and the motion trajectory type can be determined as the trajectory type of the image acquisition device. Then, the trajectory types of all image acquisition devices in the first image acquisition device can be clustered with the same trajectory type to obtain the classification result of the trajectory type. Then, the number of each trajectory type can be determined, and the trajectory type with the largest number can be determined as the first trajectory type corresponding to the first classification type; or the ratio of the number of each trajectory type to the total number of trajectory types can be calculated respectively to obtain the first proportion of each trajectory type, and the trajectory type with a first proportion greater than the first proportion threshold can be determined as the first trajectory type corresponding to the first classification type. It can be understood that the first trajectory type can be one or more.
[0072] The motion trajectory type of the moving object may include at least one of unidirectional, bidirectional, multidirectional, non-directional, wandering and stationary.
[0073] Step 130: When there is a second image acquisition device whose trajectory type is different from the first trajectory type in the first image acquisition device, determine the image acquisition device whose classification is questionable from the second image acquisition devices, and output classification confirmation reminder information.
[0074] The classification confirmation reminder information is used to instruct the user to confirm the classification type of the image acquisition device whose classification is in doubt. According to the classification confirmation reminder information, the user can confirm whether the classification type of the image acquisition device whose classification is in doubt is correct; if the confirmation is correct, the image acquisition device whose classification is in doubt is classified into the first classification type; otherwise, a new classification type can be established for the image acquisition device whose classification is in doubt, or it can be classified into the correct classification type among the existing classification types.
[0075] After determining the first trajectory type corresponding to the first classification type, the trajectory type of each image acquisition device in the first image acquisition device can be compared with the first trajectory type. If the trajectory type of a certain image acquisition device is the same as the first trajectory type, it means that the image acquisition device belongs to the first classification type and the classification is correct; if the trajectory type of a certain image acquisition device is different from the first trajectory type, it means that the classification of the image acquisition device may be wrong and needs further confirmation. In this way, a second image acquisition device in the first image acquisition device whose trajectory type is different from the first trajectory type can be determined.
[0076] In the case that there is a second image acquisition device whose trajectory type is different from the first trajectory type in the first image acquisition device, the second image acquisition device can be directly determined as an image acquisition device with questionable classification. Alternatively, the first label information corresponding to the first classification type can be used to verify the labels of each image acquisition device in the second image acquisition device to determine the image acquisition device with questionable classification in the second image acquisition device.
[0077] For example, the image acquisition device that does not contain the first label information in the second image acquisition device can be determined as an image acquisition device with questionable classification. Alternatively, a fourth image acquisition device containing the first label information can be determined in the image acquisition device to be classified, and the fourth image acquisition device can be taken as an image acquisition device of the same classification type, and the trajectory types of these image acquisition devices can be combined to obtain the classification trajectory type corresponding to the classification type; then, the image acquisition device in the second image acquisition device whose trajectory type is different from the classification trajectory type can be determined as an image acquisition device with questionable classification, or, the proportion of image acquisition devices of each trajectory type in the second image acquisition device in the fourth image acquisition device can be determined, and a proportion threshold can be preset. If the proportion is less than or equal to the proportion threshold, it is considered that the image acquisition device of this trajectory type does not belong to the classification type, and the image acquisition device of this trajectory type is determined as an image acquisition device with questionable classification.
[0078] The classification method of image acquisition devices provided by the embodiment of the present invention first clusters the image acquisition devices to be classified based on the background picture of the first monitoring image collected by the image acquisition device to be classified, and obtains the first classification result of the image acquisition device to be classified; then, for each first image acquisition device of the first classification type in the first classification result, the motion trajectory type of the moving object in the foreground picture of the corresponding first monitoring image is determined, and the first trajectory type corresponding to the first classification type is obtained; for different monitoring scenes, different types of motion trajectories may be presented in the foreground picture, and when there is a second image acquisition device with a trajectory type different from the first trajectory type in the first image acquisition device, it indicates that the second image acquisition device may not belong to the first classification type and needs further verification, then the image acquisition device with doubtful classification is further determined from the second image acquisition device and the classification confirmation reminder information is output, and the classification confirmation reminder information is used to instruct the user to further confirm the classification type of the image acquisition device with doubtful classification. In this way, for each type of first image acquisition device classified based on the background picture, by further using the motion trajectory type of its foreground moving object to verify whether the classification result is correct, the image acquisition device with incorrect classification can be verified, thereby improving the accuracy of image acquisition device classification.
[0079] based on Figure 1A classification method for image acquisition devices of the corresponding embodiment, in an exemplary embodiment, determining a first trajectory type corresponding to a first classification type based on a first monitoring image corresponding to a first image acquisition device may include: determining a third trajectory type for each image acquisition device based on the first monitoring image corresponding to each image acquisition device in the first image acquisition device, and clustering all third trajectory types to obtain a trajectory type clustering result; taking each trajectory type in the trajectory type clustering result as a first target trajectory type, and determining a first proportion corresponding to the first target trajectory type; when the first proportion is greater than a first proportion threshold, determining the first target trajectory type as the first trajectory type corresponding to the first classification type; wherein the first proportion is used to characterize the proportion of image acquisition devices in the first image acquisition device that include the first target trajectory type in the first image acquisition device.
[0080] For example, taking the image acquisition device as a camera, assume that after the cameras to be classified are classified through step 110, the first classification results obtained are classification type A and classification type B. For classification type A, assuming that the corresponding first camera is m cameras, the trajectory type of the camera can be determined according to the first monitoring image corresponding to each camera to obtain m trajectory types, that is, each camera corresponds to one trajectory type; then the m trajectory types can be clustered with the same trajectory type to determine the number of each trajectory type, and the ratio of the number to m is calculated to obtain the first proportion of each trajectory type, and the trajectory type whose first proportion is greater than the first proportion threshold is determined as the first trajectory type corresponding to classification type A. It can be understood that the first trajectory type can be one or more. For example, the m trajectory types include m1 trajectory types G1 and m2 trajectory types G2, and the first proportion threshold is M. If and Then it can be determined that the first trajectory type corresponding to the A classification type is trajectory type G1. Wherein, m1+m2=m.
[0081] For the B classification type, the same processing method as the A classification type can be used to obtain the first trajectory type corresponding to the B classification type.
[0082] Exemplarily, in the embodiment of the present invention, for any image acquisition device, the trajectory type of the image acquisition device can be determined according to the motion trajectory of the moving object in the surveillance image acquired by the image acquisition device. The trajectory type can be classified into at least one of unidirectional, bidirectional, multidirectional, non-directional, wandering and stationary according to the overall movement of the moving object.
[0083] Specifically, taking the image acquisition device as a camera as an example, for any camera, all motion trajectories of moving objects in the surveillance image acquired by the camera within a set time period can be obtained, and these motion trajectories can be merged through trajectory fitting. After the merger, the number of all motion trajectories can be counted, and the direction of each motion trajectory can be determined. When the number of motion trajectories greater than a set threshold among these motion trajectories is all facing the same direction, the direction is determined as the trajectory direction of the camera screen, and the number of types of trajectory directions is determined at the same time. Then, the trajectory type of the camera is determined according to the trajectory direction and the number of types of trajectory directions.
[0084] Specifically, for any camera, if each motion trajectory after fitting moves in the same direction and the maximum angle between the motion trajectories is less than 45 degrees, it can be determined that the trajectory type of the camera is unidirectional. Figure 2 The schematic diagram of the motion trajectory of a unidirectional trajectory type is shown as an example. Figure 2 As shown in the figure, there is only one direction of motion trajectory in the camera image, that is, each motion trajectory moves in the same direction. This type of trajectory is mainly suitable for scenes such as pedestrian gate entry and vehicle one-way lane.
[0085] For any camera, if the number of motion trajectories moving in two directions in the fitted motion trajectory is greater than the set threshold, and the angle between the two directions is greater than 90 degrees, it can be determined that the trajectory type of the camera is bidirectional. Figure 3 The schematic diagram of the motion trajectory of the bidirectional trajectory type is shown as an example. Figure 3 As shown, the camera image includes two directions of motion trajectories, the angle between the two directions is 180 degrees, that is, the directions are opposite, forming a bidirectional motion trajectory. This trajectory is mainly suitable for scenes such as pedestrian roads and two-way lanes.
[0086] For any camera, if the number of motion trajectories moving in more than two directions in the fitted motion trajectory is greater than the set threshold, and the angle between each direction is greater than 45 degrees, then the trajectory type of the camera can be determined to be multi-directional. Figure 4 The schematic diagram of the multi-directional motion trajectory is shown as an example. Figure 4 As shown in FIG. 1 , the camera image includes motion trajectories in four directions, and the angles between the four directions are all greater than 45 degrees, forming a multi-directional motion trajectory. This trajectory is mainly suitable for scenes such as multi-fork intersections and indoor roads.
[0087] For any camera, if the motion trajectory after fitting still has no unified direction, for example, if the directions of the motion trajectory after fitting are greater than 4, it is considered that there is no unified direction, and the trajectory type of the camera can be determined to be undirected. Figure 5The schematic diagram of the trajectory type of the motion trajectory is shown as an example, referring to Figure 5 As shown in FIG. 1 , the camera image includes motion trajectories in more than four directions, and there is no unified direction between the motion trajectories, forming a non-directional motion trajectory. This trajectory is mainly suitable for scenes without fixed pedestrian roads such as pedestrian plazas and playgrounds.
[0088] For any camera, if the fitted motion trajectory moves back and forth in the picture within the first preset time period to form a wandering trajectory, it can be determined that the trajectory type of the camera is wandering. Figure 6 The schematic diagram of the motion trajectory of the trajectory type of wandering is shown as an example, referring to Figure 6 As shown in the figure, the camera image includes a trajectory moving back and forth in one direction, forming a wandering trajectory. This trajectory is mainly suitable for scenes such as shopping malls, exhibition halls, scenic spots, and driving school examination rooms.
[0089] For any camera, if there are trajectories greater than a set threshold in the fitted motion trajectory that do not move in the camera image within a second preset time period, that is, most of the trajectories in the camera image are stationary, then the trajectory type of the camera can be determined to be stationary. Figure 7 The schematic diagram of the motion trajectory of the static trajectory type is shown as an example. Figure 7 As shown in the figure, the circle trajectory represents the static trajectory point. If the motion trajectory in the camera image is all static trajectory points, the trajectory type of the camera is static. This trajectory type is mainly suitable for scenes such as office buildings, cinemas, conference halls and parking lots.
[0090] according to Figure 2 to Figure 7 From the examples, it can be understood that the unidirectional, bidirectional and multidirectional trajectory types usually correspond to scenes moving along the road; the undirected, wandering and stationary trajectory types usually correspond to scenes without road constraints, such as movement trajectories in squares, shopping malls, offices, etc.
[0091] Understandably, Figure 2 to Figure 7 Take the camera as an example to explain, for any image acquisition device, you can follow Figure 2 to Figure 7 Based on this, for each image acquisition device in the first image acquisition device, according to Figure 2 to Figure 7 The principle of determining the trajectory type in the example may be to determine the third trajectory type of each image acquisition device according to the first monitoring image corresponding to the image acquisition device.
[0092] based on Figure 1In a method for classifying image acquisition devices of the corresponding embodiment, in an exemplary embodiment, the image acquisition devices to be classified may be first classified according to the type of moving objects in the first monitoring image captured by the image acquisition device to be classified, and then each type of image acquisition devices in the classification result may be clustered and classified according to the background features, thereby forming multiple sub-categories of image acquisition devices under each type of moving object. Specifically, clustering the image acquisition devices to be classified based on the background picture of the first monitoring image to obtain a first classification result may include: classifying the image acquisition devices to be classified according to the type of moving objects in the first monitoring image to obtain a second classification result; for each third image acquisition device of the second classification type in the second classification result, obtaining the background picture of the first monitoring image corresponding to the third image acquisition device and extracting the background features of the background picture, and clustering the third image acquisition devices based on the background features to obtain the first classification result.
[0093] The moving object type may include at least one of no moving object, pedestrians, vehicles, and both pedestrians and vehicles. The second classification type is the moving object type.
[0094] For example, in the process of clustering the third image acquisition devices based on background features, the geographical location types of the third image acquisition devices can be combined, and the image acquisition devices of each geographical location type in the third image acquisition devices can be clustered according to the background features to obtain the first classification result, thereby forming multiple subcategories under each major category. In this way, the image acquisition devices to be classified can be classified at multiple levels to improve the accuracy of classification and the efficiency of image acquisition device retrieval and viewing.
[0095] According to the classification method of image acquisition devices in the above-mentioned embodiments, for each first classification type in the first classification result, the number of image acquisition devices in some first classification types may be small, and when confirming the trajectory type corresponding to the first classification type based on the motion trajectory of these image acquisition devices, it may lead to inaccurate determination of the trajectory type. Based on this, in an exemplary embodiment of the present invention, after classifying the image acquisition devices to be classified to obtain the first classification result, the image acquisition devices of each classification type in the first classification result can be labeled, and label information belonging to the classification type is added to the image acquisition devices of each classification type, and the first classification result is verified based on the label information.
[0096] Specifically, the classification method of the image acquisition device also includes: determining the first label information of each first classification type in the first classification result, and adding the first label information to the first image acquisition device of the first classification type. Correspondingly, determining the image acquisition device whose classification is questionable from the second image acquisition device may include: determining the first target label from the first label information corresponding to the first classification type; determining the fourth image acquisition device containing the first target label among the image acquisition devices to be classified; taking each trajectory type of the second image acquisition device as the second target trajectory type, and determining the second proportion corresponding to the second target trajectory type; when the second proportion is not greater than the second proportion threshold, determining the image acquisition device containing the second target trajectory type in the second image acquisition device as the image acquisition device whose classification is questionable; wherein the second proportion is used to characterize the proportion of the image acquisition devices containing the second target trajectory type in the fourth image acquisition device.
[0097] The first tag information may include a name tag and / or a place tag. The first target tag may be at least one tag in the first tag information. A tag may be randomly selected from the first tag information as the first target tag, or each tag in the first tag information may be used as the first target tag in sequence. It is also possible to set a priority for each tag in the first tag information, and use the tag with the highest priority as the first target tag. For example, when the name tag is included in the first tag information, the name tag is used as the first target tag, and when the name tag is not included in the first tag information, the place tag is used as the first target tag.
[0098] Exemplarily, determining the first label information of each first classification type in the first classification result may include: for each first image acquisition device of the first classification type in the first classification result, extracting a common term from the names of each image acquisition device in the first image acquisition device, and when the number of image acquisition devices in the first image acquisition device that contains the common term is greater than a first quantity threshold, determining the common term as the first label information corresponding to the first classification type; and / or, for each first image acquisition device of the first classification type in the first classification result, clustering the geographic location information of each image acquisition device in the first image acquisition device, using each type of geographic location information obtained by clustering as target geographic location information, and when the number of image acquisition devices in the first image acquisition device that contains the target geographic location information is greater than a second quantity threshold, determining the target geographic location information as the first label information corresponding to the first classification type.
[0099] Among them, the common terms as the name tags of the first classification type can be obtained by extracting keywords from the names of the image acquisition devices, that is, the common terms are the keywords shared by the names of the image acquisition devices in the first image acquisition device that are greater than the first quantity threshold. After obtaining the name tag corresponding to the first classification type, the name tag can be used to label the first image acquisition device corresponding to the first classification type. It is understandable that the image acquisition devices in the first image acquisition device that cannot extract terms will also be marked with the name tag.
[0100] The target geographic location information can be used as a place label of the first classification type, and can be determined based on the place range marked by the first image acquisition device in the map. Specifically, the map data can identify the place ranges such as business districts, communities, scenic spots, hospitals, etc. The image acquisition devices within the same place range have the same geographic location information. For each first image acquisition device of the first classification type in the first classification result, the first image acquisition device can be labeled again using the geographic location information of each image acquisition device in the first image acquisition device in the map. When the number of image acquisition devices in the first image acquisition device that is greater than the second number threshold has the same geographic location information, the geographic location information can be used as the place label of the first classification type, and the first image acquisition devices of the first classification type are labeled with a unified place label.
[0101] It is understandable that once the deployment location of the image acquisition device is determined, its geographical location information is determined, so the place label can be easily obtained. In this way, when the first image acquisition device is set with a name, the first image acquisition device can be marked with a name label, or the first image acquisition device can be marked with a name label and a place label at the same time; when the first image acquisition device is not set with a name, the first image acquisition device can be marked with a place label, thereby ensuring the labeling of the first image acquisition device. In this way, the corresponding first label information can be added to each classification type of image acquisition device in the image acquisition device to be classified.
[0102] Based on this, after labeling the image acquisition devices of each classification type, the first classification result can be verified using the first label information of each classification type.
[0103] For example, taking the image acquisition device as a camera, assuming that the cameras to be classified are divided into A classification type and B classification type through clustering of background features, and the proportion threshold used to determine the trajectory type is E%. For each classification type, if more than E% of the cameras in the classification type contain at least one of the same trajectory types, then the at least one trajectory type is determined as the first trajectory type corresponding to the classification type, that is, a classification type can have at least one first trajectory type. For cameras in the classification type that do not meet the first trajectory type, they can be marked as questionable cameras to be verified.
[0104] Assume that there are 12 cameras in category A. Figure 2 to Figure 7 The principle of the example is to analyze the motion trajectories of the surveillance images collected by these 12 cameras to obtain the trajectory type of each camera. For example, in the A classification type, 8 cameras have wandering trajectories, 2 cameras have omnidirectional trajectories, and the other 2 cameras have stationary trajectories. The proportion threshold is 30%. Then, the 4 cameras with non-wandering trajectories can be determined as the second camera and marked as suspicious cameras. Assuming that there are 20 cameras in the B classification type, similarly, the following can be used. Figure 2 to Figure 7 The principle of the example is to analyze the motion trajectories of the surveillance images captured by the 20 cameras to obtain the trajectory type of each camera. For example, in the B classification type, 8 cameras have wandering trajectories, 10 cameras have non-directional trajectories, and the trajectories of the remaining 2 cameras are respectively 1 static trajectory and 1 multi-directional trajectory. The proportion threshold is 30%. The two cameras with static trajectories and multi-directional trajectories can be determined as the second camera and marked as suspicious cameras.
[0105] For each classification type, such as classification type A, the marked questionable camera can be used as the final camera with questionable classification, and the corresponding classification confirmation reminder information can be output to remind the user to confirm whether to classify the camera with questionable classification into classification type A. Alternatively, the marked questionable camera can be further verified using the first label information corresponding to classification type A to determine the final camera with questionable classification.
[0106] Specifically, a label can be selected from the first label information of the A classification type as the first target label, and all cameras containing the first target label in the cameras to be classified are regarded as the same type of cameras, and the motion trajectory analysis is performed on their monitoring images to obtain the trajectory type corresponding to the camera type, and then the cameras that are different from the trajectory type in the marked questionable cameras are determined as the final questionable classification cameras. Among them, the cameras containing the first target label may correspond to multiple classification types, and the cameras with the same label have the same characteristics, such as the same scene in the monitoring area, which can be considered as the same type of camera. Then, the trajectory type of the cameras with a proportion greater than the threshold value among these cameras can be determined as the trajectory type of the camera of this type. In this way, by merging the cameras containing the first target label and performing a combined analysis on the trajectory types of each camera corresponding to the first target label, the number of cameras when determining the trajectory type corresponding to the same type of camera can be appropriately expanded, thereby improving the accuracy of the trajectory type of each type of camera.
[0107] For example, assuming that both classification type A and classification type B contain the label "shopping mall counter", the trajectory types of all cameras with this label can be merged, that is, the trajectory types of all cameras of classification type A and classification type B can be merged. Among the 32 cameras after the merger, 16 cameras have wandering trajectories, 12 cameras have omnidirectional trajectories, 3 cameras have stationary trajectories, and 1 camera has multi-directional trajectories. The wandering trajectories and omnidirectional trajectories both meet the condition of being greater than the proportion threshold of 30%. The cameras with omnidirectional trajectories among the questionable cameras originally marked in classification type A will be removed from doubt. Then, the 2 stationary trajectory cameras in classification type A and the multi-directional trajectory cameras and stationary trajectory cameras in classification type B are determined as the final cameras with questionable classification.
[0108] Based on the classification method of the image acquisition device of each embodiment described above, in an exemplary embodiment, after analyzing a large number of monitoring images collected by each image acquisition device in the monitoring system using the method of the embodiment described above and classifying each image acquisition device based on the analysis, the corresponding relationship between background features, trajectory types and label information can be obtained to form a classification information table. When there is a new image acquisition device in the monitoring system or the deployment position of a certain image acquisition device changes, the label of the new image acquisition device or the image acquisition device with changed position information can be verified according to the classification information table to see whether it is correct. Specifically, the classification method of the image acquisition device also includes: when the target image acquisition device is detected, obtaining a second monitoring image collected by the target image acquisition device; obtaining the target background features of the second monitoring image, and determining the fourth trajectory type of the target image acquisition device based on the second monitoring image; searching for a second target label matching the target background features and the fourth trajectory type from the classification information table; obtaining the second label information of the target image acquisition device, and outputting label abnormality prompt information when the second label information does not match the second target label. Among them, the target image acquisition device includes a new image acquisition device or an image acquisition device with changed position information; the classification information table is used to save the corresponding relationship between background features, trajectory types and label information.
[0109] For example, taking the image acquisition device as a camera, assuming that a new camera F is added to the monitoring system, the background features of the second monitoring image acquired by camera F can be extracted, and the trajectory type of camera F can be determined based on the second monitoring image. Then, the background features are matched with the classification information table to check whether the background features and the corresponding trajectory type exist in the classification information table. If so, the corresponding second target label is found according to the classification information table to determine whether the label information of camera F matches the second target label. If not, the label abnormality prompt information corresponding to camera F is output to prompt the user to confirm whether to add a new classification type for camera F or to classify camera F into the correct classification type, that is, the classification type corresponding to the second target label.
[0110] Exemplarily, an optional representation of the classification information table can be expressed as the following Table 1. The image acquisition device without motion trajectory does not record. For the case where the moving object types are both pedestrians and vehicles, the motion trajectories of pedestrians and vehicles can be analyzed separately and recorded in the corresponding data of pedestrians and vehicles.
[0111] Table 1
[0112]
[0113] It should be noted that in Table 1, the numbers after each feature value in the background feature column are only used to distinguish different feature values and do not represent specific values.
[0114] Based on the classification method of the image acquisition device of each of the above-mentioned embodiments, in an exemplary embodiment, after outputting the classification confirmation reminder information, the user's operation based on the classification confirmation reminder information can be received. For example, if it is confirmed that the current classification of the image acquisition device with doubtful classification is not abnormal, a first instruction to confirm the classification can be input to the electronic device; if it is believed that the current classification of the image acquisition device with doubtful classification is inappropriate and needs to be classified into the correct classification type, a second instruction to adjust the classification can be input to the electronic device. In this way, the user can make further determination on the classification of the image acquisition device with doubtful classification based on the classification verification result, thereby improving the accuracy of classification. Specifically, the classification method of the image acquisition device also includes: in the case of detecting the first instruction to confirm the classification, classifying the image acquisition device with doubtful classification into the first classification type; in the case of detecting the second instruction to adjust the classification, determining the target classification type, and classifying the image acquisition device with doubtful classification into the target classification type.
[0115] Exemplarily, when the second instruction to adjust the classification is detected, the target classification type can be determined based on the trajectory type of the image acquisition device whose classification is questionable. For example, when the trajectory type of the image acquisition device whose classification is questionable exists in the classification information table, the trajectory type of the image acquisition device whose classification is questionable can be determined as the target classification type, and the image acquisition device whose classification is questionable can be classified into the target classification type in the classification information table; when the trajectory type of the image acquisition device whose classification is questionable does not exist in the classification information table, the corresponding trajectory type can be automatically created in the classification information table, and the created trajectory type can be used as the target classification type.
[0116] Exemplarily, the second instruction to adjust the classification may be an instruction to create a classification type, based on which, when the second instruction to adjust the classification is detected, a new classification type may be created as a target classification type for the image acquisition device with questionable classification according to the second instruction. Alternatively, the target classification type to which the image acquisition device with questionable classification belongs may be specified in the existing classification type through the second instruction.
[0117] In the following, in combination with specific application scenarios, taking a case where the image acquisition device is a camera as an example, the classification method of the image acquisition device provided by the embodiment of the present invention is further illustrated.
[0118] Taking the mall monitoring system as an example, the mall counter screen and some service desk (such as cashier counter, front desk, etc.) screens have similar characteristics. After these cameras are classified according to the background characteristics of each camera in the monitoring system, the cameras of these screens will be considered to be of the same type. According to the classification method of the image acquisition device provided by the present invention, it can be determined that the pedestrian motion trajectories in the counter and service desk areas are not the same, that is, the trajectory types of these cameras are not the same. The trajectory types of the counter class are mainly wandering and undirected, while the trajectory of the service desk is mainly static. The camera monitoring the service desk will be considered to be a camera with doubtful classification, and a classification confirmation reminder message can be output, and the user confirms whether the camera monitoring the service desk and the camera monitoring the counter are classified into the same category. If it is confirmed to be classified into the same category, the service desk tag can be added to the wandering and undirected trajectory types, and the feature value of the corresponding background feature can be increased at the same time; if it is not classified into the same category, the service desk tag can be added to the static trajectory. In this way, the camera classified based on the background feature is further classified by trajectory type, and the camera with doubtful classification can be verified. After that, the camera with doubtful classification can be accurately classified, which improves the accuracy of camera classification.
[0119] When a new camera is added to the monitoring system or the location information of a camera is changed, the camera can be labeled and label information can be added. At this time, the classification information table can be used to compare the background characteristics and trajectory type of the camera to determine whether the label information of the camera matches the classification information table. If it does not match, the user is prompted that the label information of the camera is inaccurate, that is, the classified label is inaccurate and needs to be replaced. In this way, the classification information table can be used to automatically check whether the label classification of the newly added camera or the camera with changed location information is abnormal, thereby reducing the error rate of the label.
[0120] The image acquisition device classification method provided by the embodiment of the present invention classifies the image acquisition devices by mutual verification of the static background and dynamic foreground trajectories of the monitoring screen of the image acquisition device. When the image acquisition device is connected to the monitoring system, the monitoring system automatically clusters the image acquisition devices according to the background characteristics of the monitoring images acquired by each image acquisition device to obtain a first classification result; for each classification type of image acquisition device in the first classification result, keywords are automatically extracted from the names of these image acquisition devices, and these image acquisition devices are labeled and managed using the keywords; the first classification result can be classified and verified by the motion trajectory of the foreground moving object in the monitoring image to determine whether the current first classification result is correct, and the image acquisition devices whose classification is questionable are found for further confirmation, thereby achieving the effect of automatic classification, and the accuracy of image acquisition device classification is improved by verifying the classification results.
[0121] The classification method for image acquisition devices provided by the embodiment of the present invention can use the trajectory type of the moving object in the monitoring image to verify whether the classification label of the image acquisition device is correct. For the image acquisition device in question, the user can confirm whether to continue to be classified into the current classification type or to be classified into other classification types. Compared with the method of classifying and labeling only by background features, the verification function based on the trajectory type is added, and when the amount of data in the monitoring image is large enough, the wrong classification can be better identified, thereby reducing the error rate of the classification label of the image acquisition device.
[0122] The classification device of the image acquisition device provided by the present invention is described below. The classification device of the image acquisition device described below and the classification method of the image acquisition device described above can be referred to each other.
[0123] Figure 8 The schematic diagram of the structure of the classification device of the image acquisition device provided by the embodiment of the present invention is exemplarily shown. Figure 8 As shown, the classification device 800 of the image acquisition device may include: a classification module 810, which is used to obtain a first monitoring image acquired by the image acquisition device to be classified, and cluster the image acquisition devices to be classified based on the background picture of the first monitoring image to obtain a first classification result; a trajectory type determination module 820, which is used to determine, for each first image acquisition device of the first classification type in the first classification result, a first trajectory type corresponding to the first classification type based on the first monitoring image corresponding to the first image acquisition device; a classification verification module 830, which is used to determine the image acquisition device whose classification is questionable from the second image acquisition device when there is a second image acquisition device whose trajectory type is different from the first trajectory type in the first image acquisition device, and output classification confirmation reminder information; wherein the classification confirmation reminder information is used to indicate confirmation of the classification type of the image acquisition device whose classification is questionable; the trajectory type is used to characterize the motion trajectory type of the moving object in the foreground picture of the monitoring image acquired by the image acquisition device.
[0124] In an exemplary embodiment, the trajectory type determination module 820 may include: a first determination unit, used to determine the third trajectory type of each image acquisition device based on the first monitoring image corresponding to each image acquisition device in the first image acquisition device; a first clustering unit, used to cluster all third trajectory types to obtain a trajectory type clustering result; a second determination unit, used to take each trajectory type in the trajectory type clustering result as a first target trajectory type, and determine a first proportion corresponding to the first target trajectory type, wherein the first proportion is used to characterize the proportion of image acquisition devices in the first image acquisition device that contain the first target trajectory type in the first image acquisition device; and a third determination unit, used to determine the first target trajectory type as the first trajectory type corresponding to the first classification type when the first proportion is greater than a first proportion threshold.
[0125] In an exemplary embodiment, the classification module 810 may include: a classification unit, used to classify the image acquisition device to be classified according to the type of moving objects in the first monitoring image to obtain a second classification result; a feature extraction unit, used to obtain the background picture of the first monitoring image corresponding to the third image acquisition device and extract the background features of the background picture for each third image acquisition device of the second classification type in the second classification result; and a second clustering unit, used to cluster the third image acquisition devices based on the background features extracted by the feature extraction unit to obtain the first classification result.
[0126] In an exemplary embodiment, the classification device 800 of the image acquisition device may further include: a labeling module, used to determine first label information of each first classification type in the first classification result, and add first label information to the first image acquisition device of the first classification type.
[0127] Correspondingly, the classification verification module 830 may include: a fourth determination unit, used to determine the first target label from the first label information corresponding to the first classification type; a fifth determination unit, used to determine the fourth image acquisition device containing the first target label in the image acquisition device to be classified; a sixth determination unit, used to take each trajectory type of the second image acquisition device as the second target trajectory type, and determine the second proportion corresponding to the second target trajectory type, wherein the second proportion is used to characterize the proportion of the image acquisition devices containing the second target trajectory type in the fourth image acquisition device; and a seventh determination unit, used to determine the image acquisition device containing the second target trajectory type in the second image acquisition device as an image acquisition device with questionable classification when the second proportion is not greater than the second proportion threshold.
[0128] In an exemplary embodiment, the labeling module may include a first label determination unit and / or a second label determination unit. The first label determination unit is used to: for each first image acquisition device of the first classification type in the first classification result, extract common terms from the names of each image acquisition device in the first image acquisition device; when the number of image acquisition devices containing common terms in the first image acquisition device is greater than a first quantity threshold, determine the common terms as the first label information corresponding to the first classification type. The second label determination unit is used to: for each first image acquisition device of the first classification type in the first classification result, cluster the geographic location information of each image acquisition device in the first image acquisition device; use each type of geographic location information obtained by clustering as target geographic location information, and when the number of image acquisition devices containing target geographic location information in the first image acquisition device is greater than a second quantity threshold, determine the target geographic location information as the first label information corresponding to the first classification type.
[0129] In an exemplary embodiment, the classification device 800 of the image acquisition device may also include a label verification module; the label verification module may be used to: obtain a second monitoring image acquired by the target image acquisition device when the target image acquisition device is detected; obtain the target background features of the second monitoring image, and determine the fourth trajectory type of the target image acquisition device based on the second monitoring image; search for a second target label that matches the target background features and the fourth trajectory type from a classification information table; the classification information table is used to store the correspondence between background features, trajectory types, and label information; obtain the second label information of the target image acquisition device, and output label abnormality prompt information when the second label information does not match the second target label. The target image acquisition device includes a newly added image acquisition device or an image acquisition device whose position information has changed.
[0130] In an exemplary embodiment, the classification device 800 of the image acquisition device may also include: a first classification module, which is used to classify the image acquisition device with questionable classification into a first classification type when a first instruction to confirm the classification is detected; and a second classification module, which is used to determine the target classification type and classify the image acquisition device with questionable classification into the target classification type when a second instruction to adjust the classification is detected.
[0131] Fig. 9 An example of a structural diagram of an electronic device is shown in FIG. Fig. 9 As shown, the electronic device may include: a processor 910, a communication interface 920, a memory 930, and a communication bus 940. The processor 910, the communication interface 920, and the memory 930 communicate with each other through the communication bus 940. The processor 910 may call the logic instructions in the memory 930 to execute the classification method of the image acquisition device provided by any of the above method embodiments.
[0132] In addition, the logic instructions in the above-mentioned memory 930 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.
[0133] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the classification method of the image acquisition device provided in any of the above method embodiments.
[0134] On the other hand, the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the classification method for image acquisition devices provided by any of the above method embodiments.
[0135] Among them, the computer-readable storage medium includes a non-transitory computer-readable storage medium.
[0136] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.
[0137] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.
[0138] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A classification method for image acquisition devices, characterized in that: include: Acquire a first monitoring image captured by an image acquisition device to be classified, and cluster the image acquisition devices to be classified based on a background picture of the first monitoring image to obtain a first classification result; For each first image acquisition device of the first classification type in the first classification result, determining a first trajectory type corresponding to the first classification type based on the first monitoring image corresponding to the first image acquisition device; In the case that there is a second image acquisition device whose trajectory type is different from the first trajectory type among the first image acquisition devices, determining an image acquisition device whose classification is questionable from among the second image acquisition devices, and outputting classification confirmation reminder information; The classification confirmation reminder information is used to instruct to confirm the classification type of the image acquisition device whose classification is in doubt; The trajectory type is used to characterize the motion trajectory type of the moving object in the foreground image of the monitoring image acquired by the image acquisition device.
2. The classification method of image acquisition equipment according to claim 1, characterized in that: The determining, based on the first monitoring image corresponding to the first image acquisition device, a first trajectory type corresponding to the first classification type includes: determining a third trajectory type of each image acquisition device based on the first monitoring image corresponding to each image acquisition device in the first image acquisition devices, and clustering all the third trajectory types to obtain a trajectory type clustering result; Taking each trajectory type in the trajectory type clustering result as a first target trajectory type, and determining a first proportion corresponding to the first target trajectory type; the first proportion is used to represent the proportion of image acquisition devices containing the first target trajectory type in the first image acquisition device in the first image acquisition device; When the first proportion is greater than a first proportion threshold, the first target trajectory type is determined as a first trajectory type corresponding to the first classification type.
3. The classification method of image acquisition equipment according to claim 1, characterized in that: The clustering of the image acquisition devices to be classified based on the background picture of the first monitoring image to obtain a first classification result includes: Classifying the image acquisition device to be classified according to the type of the moving object in the first monitoring image to obtain a second classification result; For each third image acquisition device of the second classification type in the second classification result, the background picture of the first monitoring image corresponding to the third image acquisition device is obtained and the background features of the background picture are extracted; the third image acquisition devices are clustered based on the background features to obtain the first classification result.
4. The classification method of the image acquisition device according to any one of claims 1 to 3, characterized in that: Also includes: Determine first label information of each first classification type in the first classification result, and add the first label information to the first image acquisition device of the first classification type; The step of determining the image acquisition device whose classification is questionable from the second image acquisition device comprises: Determine a first target label from the first label information corresponding to the first classification type; Determine a fourth image acquisition device that includes the first target label among the image acquisition devices to be classified; Taking each trajectory type of the second image acquisition device as a second target trajectory type, determining a second proportion corresponding to the second target trajectory type; the second proportion is used to represent the proportion of the image acquisition devices in the fourth image acquisition device that include the second target trajectory type in the fourth image acquisition device; When the second proportion is not greater than a second proportion threshold, the image acquisition device containing the second target trajectory type in the second image acquisition device is determined as the image acquisition device with questionable classification.
5. The classification method of image acquisition equipment according to claim 4, characterized in that: The determining of first label information of each first classification type in the first classification result includes: For each first image acquisition device of the first classification type in the first classification result, extract a common term from the name of each image acquisition device in the first image acquisition device; if the number of image acquisition devices containing the common term in the first image acquisition device is greater than a first quantity threshold, determine the common term as the first label information corresponding to the first classification type; and / or, For each first image acquisition device of the first classification type in the first classification result, the geographic location information of each image acquisition device in the first image acquisition devices is clustered; each type of geographic location information obtained by clustering is used as target geographic location information, and when the number of image acquisition devices containing the target geographic location information in the first image acquisition devices is greater than a second quantity threshold, the target geographic location information is determined as the first label information corresponding to the first classification type.
6. The classification method of image acquisition equipment according to any one of claims 1 to 3, characterized in that: Also includes: When a target image acquisition device is detected, acquiring a second monitoring image acquired by the target image acquisition device; The target image acquisition device includes a newly added image acquisition device or an image acquisition device whose position information has been changed; Acquire target background features of the second monitoring image, and determine a fourth trajectory type of the target image acquisition device based on the second monitoring image; searching a second target label that matches the target background feature and the fourth trajectory type from a classification information table; the classification information table is used to store the correspondence between background features, trajectory types and label information; The second label information of the target image acquisition device is obtained, and when the second label information does not match the second target label, label abnormality prompt information is output.
7. The classification method of image acquisition equipment according to any one of claims 1 to 3, characterized in that: After outputting the classification confirmation reminder information, the method further includes: In case of detecting a first instruction to confirm classification, classifying the image acquisition device whose classification is questionable into the first classification type; When a second instruction to adjust the classification is detected, a target classification type is determined, and the image acquisition device whose classification is questionable is classified into the target classification type.
8. A classification device for an image acquisition device, characterized in that: include: A classification module, used to obtain a first monitoring image acquired by an image acquisition device to be classified, and cluster the image acquisition devices to be classified based on a background image of the first monitoring image to obtain a first classification result; a trajectory type determination module, configured to determine, for each first image acquisition device of the first classification type in the first classification result, a first trajectory type corresponding to the first classification type based on the first monitoring image corresponding to the first image acquisition device; a classification verification module, configured to, if there is a second image acquisition device whose trajectory type is different from the first trajectory type among the first image acquisition devices, determine an image acquisition device whose classification is questionable from among the second image acquisition devices, and output a classification confirmation reminder message; The classification confirmation reminder information is used to instruct to confirm the classification type of the image acquisition device whose classification is in doubt; The trajectory type is used to characterize the motion trajectory type of the moving object in the foreground image of the monitoring image acquired by the image acquisition device.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the classification method for the image acquisition device according to any one of claims 1 to 7 is implemented.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the classification method for an image acquisition device as claimed in any one of claims 1 to 7 is implemented.
Citation Information
Patent Citations
Detention wandering behavior recognition method and system based on target trajectory analysis
CN112633150A
Video scene classification method, system and device and storage medium
CN113673391A
Clustering method and device, equipment and computer storage medium
CN113962326A
Portrait clustering test method, system, equipment and medium
CN114332973A
Image matching using motion manifolds
US20130108177A1
