A method for abnormal target detection in traffic scenes based on rules and learning

Abstract

The present invention discloses a method for detecting abnormal targets in traffic scenes based on rules and learning, the method comprising: obtaining real-time collected traffic scene video data; performing statistics and calculations on the traffic scene video data to obtain suspected abnormal motion information of the target object; and determining the target of abnormal motion in the traffic scene according to the suspected abnormal motion information and traffic rules. The embodiment of the present invention performs statistics and calculations on the collected traffic scene video data to obtain suspected abnormal motion information of the target object, and then determines the target of abnormal motion in the traffic scene according to the suspected abnormal motion information and traffic rules, with a small amount of calculation, high real-time performance, and more accurate determination of abnormal motion targets in the traffic scene.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a method for detecting abnormal targets in traffic scenes based on rules and learning. Background Art

[0002] The existing method for detecting abnormal targets in traffic scenes is to use a generative adversarial network (GAN) to predict the next frame of four consecutive frames of video. During training, the generator is trained with normal video samples so that it can predict normal videos well; during testing, the error between the predicted frame generated by the generator for normal videos and the original video is small, while the generator cannot predict abnormal videos well, and there is a large error between its predicted frame and the original frame. Therefore, the defect of using the error between the predicted frame and the original frame to detect abnormal behavior is that the algorithm is designed on the assumption that the generator trained with limited normal samples can predict all normal samples well. However, it is impossible to collect all types of normal behaviors in reality, and the effect is not ideal for atypical normal samples. In addition, this type of method is an end-to-end abnormal behavior detection method based on deep learning, and the complex deep neural network structure makes the algorithm unable to meet the real-time requirements in practical applications.

[0003] Therefore, the existing technology still needs to be improved and developed. Summary of the invention

[0004] The technical problem to be solved by the present invention is that, in view of the above-mentioned defects of the prior art, a rule-based and learning-based method for detecting abnormal targets in traffic scenes is provided, aiming to solve the problems in the prior art of inaccurate determination of abnormal moving targets in traffic scenes, large amount of calculation, and inability to achieve real-time performance.

[0005] The technical solution adopted by the present invention to solve the problem is as follows:

[0006] In a first aspect, an embodiment of the present invention provides a method for detecting abnormal targets in traffic scenes based on rules and learning, wherein the method comprises:

[0007] Obtain real-time collected traffic scene video data;

[0008] Performing statistics and calculations on the traffic scene video data to obtain suspected abnormal motion information of the target object;

[0009] According to the suspected abnormal motion information and traffic rules, a target of abnormal motion in the traffic scene is determined.

[0010] In one implementation, the step of performing statistics and calculations on the traffic scene video data to determine suspected abnormal motion information of the target object includes:

[0011] Determine, based on the traffic video data, the moving speed of the target object in a preset time period, the moving direction of the target object in a preset time period, the moving trajectory of the target object in a preset time period, the current traffic scene of the target object, and the number of target objects in the current traffic scene;

[0012] According to the traffic scene, the number of target objects, and the movement speed, statistics and calculation are performed to determine a suspected abnormal movement speed;

[0013] According to the movement direction, the movement direction is judged to determine a suspected abnormal movement direction;

[0014] According to the motion trajectory, the motion trajectory is judged to determine a suspected abnormal motion trajectory.

[0015] In one implementation, the determining of the suspected abnormal movement speed according to the traffic scene, the number of target objects, and the movement speed includes:

[0016] According to the traffic scene, the movement speed and the number of target objects, a target object movement speed threshold is determined through statistics and calculations; wherein, when the traffic scene is a speed-limited section in an urban area, the movement speed is a first movement speed, and the number of target objects is a first target object number; when the traffic scene is a high-speed section with a low-speed limit, the movement speed is a second movement speed, and the number of target objects is a second target object number; when the traffic scene is a high-speed section with both high-speed and low-speed limits, the movement speed is a third movement speed, and the number of target objects is a third target object number;

[0017] The suspected abnormal motion speed is determined according to the target object motion speed threshold.

[0018] In one implementation, the step of determining the target object movement speed threshold according to the traffic scene, the movement speed, and the number of target objects includes:

[0019] When the traffic scene is a speed-limited section in an urban area, determining a first threshold of the target object movement speed according to the first movement speed and the first number of target objects;

[0020] When the traffic scene is a high-speed road section with a low-speed restriction, determining a second threshold value of the target object movement speed according to the second movement speed and the second number of target objects;

[0021] When the traffic scene is a high-speed road section that restricts both high and low speeds, a third target object movement speed threshold and a fourth target object movement speed threshold are determined according to the third movement speed and the third target object quantity, wherein the third target object movement speed threshold is less than the fourth target object movement speed threshold.

[0022] In one implementation, the determining of the suspected abnormal motion speed according to the target object motion speed threshold comprises:

[0023] When the traffic scene is a speed-limited section in an urban area, and when the first movement speed is greater than a first threshold of the movement speed of the target object, the first movement speed is determined to be a suspected abnormal movement speed;

[0024] When the traffic scene is a high-speed road section with a low-speed restriction, and when the second movement speed is less than a second threshold of the target object movement speed, the second movement speed is determined to be a suspected abnormal movement speed;

[0025] When the traffic scene is a high-speed section that restricts both high and low speeds, and when the third motion speed is less than a third threshold value of the target object motion speed or when the third motion speed is greater than a fourth threshold value of the target object motion speed, the third motion speed is determined to be a suspected abnormal motion speed.

[0026] In one implementation, the step of discriminating the movement direction according to the movement direction to determine the suspected abnormal movement direction includes:

[0027] Get the maximum average motion direction change value v of the target object max ;

[0028] Get the average movement direction change value of the target object

[0029] When the movement direction is greater than When , the movement direction is a suspected abnormal movement direction; wherein β is a parameter greater than 0.8 and less than 1, and is adjusted according to the current traffic scene.

[0030] In one implementation, the step of distinguishing the motion trajectory according to the motion trajectory to determine a suspected abnormal motion trajectory includes:

[0031] According to the motion trajectory, obtain the maximum motion direction change value direction of the target object max ;

[0032] When direction max >γ, the motion trajectory is a suspected abnormal motion trajectory; wherein γ is a preset motion direction threshold.

[0033] In one implementation, the step of determining the abnormal motion target according to the suspected abnormal motion information and traffic rules includes:

[0034] Obtain the prior values ​​set in the system according to traffic rules;

[0035] An abnormal motion target is determined according to the suspected abnormal motion speed, the suspected abnormal motion direction, the suspected abnormal motion trajectory and the prior value.

[0036] In a second aspect, an embodiment of the present invention further provides a device for detecting abnormal targets in traffic scenes based on rules and learning, wherein the device comprises:

[0037] A traffic scene video data acquisition unit, used to acquire real-time collected traffic scene video data;

[0038] A suspected abnormal motion information acquisition unit, used to perform statistics and calculations on the traffic scene video data to obtain suspected abnormal motion information of the target object;

[0039] The traffic scene abnormal motion target determination unit is used to determine the target of abnormal motion in the traffic scene according to the suspected abnormal motion information and traffic rules.

[0040] In a third aspect, an embodiment of the present invention further provides an intelligent terminal comprising a memory and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by one or more processors, wherein the one or more programs include a method for detecting abnormal targets in traffic scenes based on rules and learning as described in any one of the above.

[0041] In a fourth aspect, an embodiment of the present invention further provides a non-temporary computer-readable storage medium. When the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the rule-based and learning-based abnormal target detection method in traffic scenes as described in any one of the above.

[0042] Beneficial effects of the present invention: The embodiment of the present invention first obtains real-time collected traffic scene video data; then performs statistics and calculations on the traffic scene video data to obtain suspected abnormal motion information of the target object; finally, the target of abnormal motion in the traffic scene is determined according to the suspected abnormal motion information and traffic rules; It can be seen that the embodiment of the present invention obtains the suspected abnormal motion information of the target object by performing statistics and calculations on the collected traffic scene video data, and then determines the target of abnormal motion in the traffic scene according to the suspected abnormal motion information and traffic rules, with a small amount of calculation, high real-time performance is achieved, and the determination of abnormal motion targets in the traffic scene is more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0044] Figure 1 Schematic diagram of a method for detecting abnormal targets in traffic scenes based on rules and learning provided by an embodiment of the present invention

[0045] Figure 2 A principle block diagram of a device for detecting abnormal targets in traffic scenes based on rules and learning is provided in an embodiment of the present invention.

[0046] Figure 3 A block diagram of the internal structure of an intelligent terminal provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0047] The present invention discloses a method for detecting abnormal targets in traffic scenes based on rules and learning. In order to make the purpose, technical solution and effect of the present invention clearer and more specific, the present invention is further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0048] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "said", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present invention refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0049] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as generally understood by those skilled in the art in the art to which the present invention belongs. It should also be understood that terms such as those defined in general dictionaries should be understood to have meanings consistent with the meanings in the context of the prior art, and will not be interpreted with idealized or overly formal meanings unless specifically defined as herein.

[0050] In the prior art, the method for detecting abnormal targets in traffic scenes is to use a generative adversarial network (GAN) to predict the next frame of four consecutive frames of video. During training, the generator is trained with normal video samples so that it can predict normal videos well; during testing, the error between the predicted frame generated by the generator for normal videos and the original video is small, while the generator cannot predict abnormal videos well, and there is a large error between its predicted frame and the original frame. Therefore, the error between the predicted frame and the original frame is used to detect abnormal behavior. The defect of this method is that the algorithm is designed on the assumption that the generator trained with limited normal samples can predict all normal samples well. However, it is impossible to collect all types of normal behaviors in reality, and the effect is not ideal for atypical normal samples. In addition, this type of method is an end-to-end abnormal behavior detection method based on deep learning. The complex deep neural network structure makes the algorithm unable to meet the real-time requirements in practical applications. Therefore, the prior art determines the abnormal moving targets in traffic scenes inaccurately, the amount of calculation is large, and the real-time problem cannot be achieved.

[0051] In order to solve the problems of the prior art, this embodiment provides a method for detecting abnormal targets in traffic scenes based on rules and learning. Through the method in this embodiment, the abnormal target detection system in traffic scenes based on rules and learning first obtains the traffic scene video data collected by the camera in real time, including picture data and audio and video data. The traffic scene video data can see whether the target object is in motion or stationary, the speed of the target object, the direction of the target object, and the trajectory of the target object. Then, the traffic scene video data is counted and calculated to obtain the suspected abnormal motion information of the target object, that is, to obtain the information that the target object may have abnormal motion. In order to further determine the accurate abnormal motion target of the traffic scene, it is also necessary to add traffic rules to assist in judgment, that is, on the basis of the suspected abnormal motion information, the target of the abnormal motion of the traffic scene is finally determined according to the traffic rules. The method in this embodiment has a small amount of calculation, achieves high real-time performance, and is more accurate in determining the abnormal motion target of the traffic scene.

[0052] Example

[0053] Abnormal target detection technology in traffic scenes is an important technical means to judge traffic violations such as speeding and dangerous driving behaviors. Automatic detection of abnormal targets can effectively improve the efficiency of such administrative enforcement and save a lot of manpower. In practical applications, manual confirmation and administrative enforcement can be carried out based on the results obtained by the automatic detection system. Conventional abnormal target detection in traffic scenes is often based on the "learning algorithm", that is, the result obtained by the algorithm is used as the judgment result. However, the identification of abnormal targets in actual application scenarios is a very complex problem. It not only changes with the changes in the scene, but also changes with the adjustment of the speed limit regulations in the scene. In addition, traffic management regulations are also one of the key factors that directly affect the identification of abnormal targets. The computational errors in the estimation of the speed, direction of movement, trajectory, etc. of moving targets in traffic scenes also restrict the accuracy of abnormal target identification. Therefore, the present invention proposes that since most vehicles will abide by traffic rules, the principle of "most normal, few abnormal" can be used to identify abnormal targets in traffic scenes, and traffic rules can be used for auxiliary identification at the same time, that is, the "intersection" of suspected abnormal moving targets and suspected illegal moving targets is used as the final abnormal moving target. Through the abnormal target detection method for traffic scenes based on rules and learning in this embodiment, the real-time collected traffic scene video data is first obtained. The targets photographed by the camera have the characteristics of "large near and small far". Even if they are targets with exactly the same physical scale, the imaging size of the target close to the camera will be relatively large, while the imaging size of the target far from the camera will be relatively small. Therefore, it is necessary to divide the imaging area of ​​the surveillance camera in such a way that the area of ​​a single area close to the surveillance camera is larger, and the area of ​​a single area far from the surveillance camera is smaller. Then, the traffic scene video data is counted and calculated to obtain the suspected abnormal motion information of the target object; that is, the movement speed, movement direction, movement trajectory, etc. of the moving target in the traffic scene are calculated. The suspected abnormal motion information of the target object can be preliminarily obtained through the motion information of the above traffic scene. In order to accurately judge the target of abnormal motion in the traffic scene, a traffic rule judgment item is added, that is, the target of abnormal motion in the traffic scene is determined based on the suspected abnormal motion information and the traffic rules. The method in this embodiment has a small amount of calculation, achieves high real-time performance, and is more accurate in determining the target of abnormal motion in the traffic scene.

[0054] Exemplary Methods

[0055] This embodiment provides a method for detecting abnormal targets in traffic scenes based on rules and learning, which can be applied to communication intelligent terminals. Figure 1 As shown, the method includes:

[0056] Step S100, obtaining real-time collected traffic scene video data;

[0057] Specifically, after the system is installed in a certain traffic application scenario, the system will collect traffic video data in real time through the camera configured by the system. The collection methods include magnetic frequency collection and wave-break collection. Magnetic frequency collection measures the flow rate after triggering the video, while wave-break collection measures the speed after triggering the video. When the system obtains the real-time collected traffic scene video data, it can complete the information and prepare for subsequent data analysis.

[0058] This embodiment provides a method for detecting abnormal targets in traffic scenes based on rules and learning, which can be applied to communication intelligent terminals. Figure 1 As shown, the method includes:

[0059] Step S200: performing statistics and calculations on the traffic scene video data to obtain suspected abnormal motion information of the target object;

[0060] In this embodiment, the system operation method is run on the client side or server. In practice, after the system obtains the traffic scene video data, it can perform statistics and calculations on the client side or server. For example, in practice, the video data of the traffic scene includes the movement speed of the traffic scene target, the movement direction of the traffic scene target, and the movement trajectory of the traffic scene target. By placing the above information on the client side or server for statistics, and then according to a calculation method, it can also be a learning algorithm based on a neural network, etc., to obtain suspected abnormal motion information of the target object.

[0061] In order to obtain the suspected abnormal motion information of the target object, the statistics and calculations on the traffic scene video data to determine the suspected abnormal motion information of the target object include the following steps:

[0062] Step S201, determining the moving speed of the target object in a preset time period, the moving direction of the target object in a preset time period, the moving trajectory of the target object in a preset time period, the current traffic scene of the target object, and the number of target objects in the current traffic scene according to the traffic video data;

[0063] Step S202: According to the traffic scene, the number of target objects and the movement speed, statistics and calculation are performed to determine a suspected abnormal movement speed;

[0064] Step S203: according to the movement direction, the movement direction is judged to determine the suspected abnormal movement direction;

[0065] Step S204: According to the motion trajectory, the motion trajectory is judged to determine a suspected abnormal motion trajectory.

[0066] Specifically, after the system acquires the traffic video data, it can determine the target object's speed in a preset time period, the target object's direction in a preset time period, the target object's trajectory in a preset time period, the target object's current traffic scene, and the number of target objects in the current traffic scene; because in the actual operation process, the target object's suspected abnormal motion information is obtained separately, for example, according to the traffic scene, the number of target objects, the movement speed is counted and calculated to determine the suspected abnormal movement speed; according to the movement direction, the movement direction is judged to determine the suspected abnormal movement direction; according to the movement trajectory, the movement trajectory is judged to determine the suspected abnormal movement trajectory. Through multi-dimensional operations on the target object's suspected abnormal motion information, comprehensive target object suspected abnormal motion information is obtained.

[0067] In order to determine the suspected abnormal movement speed, the movement speed is counted and calculated according to the traffic scene, the number of target objects, and the determination of the suspected abnormal movement speed includes the following steps: according to the traffic scene, the movement speed and the number of target objects, a target object movement speed threshold is determined through statistics and calculation; wherein, when the traffic scene is a speed limit section in an urban area, the movement speed is a first movement speed, and the number of target objects is the first target object number; when the traffic scene is a high-speed section with low speed restrictions, the movement speed is a second movement speed, and the number of target objects is the second target object number; when the traffic scene is a high-speed section with both high and low speed restrictions, the movement speed is a third movement speed, and the number of target objects is the third target object number; according to the target object movement speed threshold, the suspected abnormal movement speed is determined.

[0068] In this embodiment, after the system obtains the traffic video data, it can determine the target object's movement speed in a preset time period, the target object's current traffic scene, and the number of target objects in the current traffic scene; due to the different traffic scenes where the movement speed and the number of target objects are located, when the traffic scene is a speed-limited section in an urban area, the movement speed is the first movement speed, and the number of target objects is the first number of target objects; when the traffic scene is a high-speed section with a low speed limit, the movement speed is the second movement speed, and the number of target objects is the second number of target objects; when the traffic scene is a high-speed section with both high and low speed limits, the movement speed is the third movement speed, and the number of target objects is the third number of target objects; a threshold value for determining suspected abnormal moving targets is calculated by a statistical method, and a suspected abnormal moving target is determined based on a comparison between the movement speed and the movement speed threshold. In another embodiment, a threshold value for determining suspected abnormal moving targets is calculated based on the movement speed, the target object's current traffic scene, and the number of target objects in the current traffic scene, and a suspected abnormal moving target is determined based on a comparison between the movement speed and the movement speed threshold. The learning and training algorithm may adopt a neural network or any possible algorithm model in practice.

[0069] In order to determine the target object motion speed threshold, the method of determining the target object motion speed threshold according to the traffic scene, the motion speed and the number of target objects includes the following steps: when the traffic scene is a speed limit section in an urban area, determining a first target object motion speed threshold according to the first motion speed and the first target object number; when the traffic scene is a high-speed section with low speed restrictions, determining a second target object motion speed threshold according to the second motion speed and the second target object number; when the traffic scene is a high-speed section with both high and low speed restrictions, determining a third target object motion speed threshold and a fourth target object motion speed threshold according to the third motion speed and the third target object number, wherein the third target object motion speed threshold is less than the fourth target object motion speed threshold.

[0070] Specifically, when the traffic scene is a speed-limited section in an urban area, that is, when the excessively high speed is abnormal, the system analyzes the speed of all moving targets in different areas within a time period t before the current moment, and estimates the speed of all moving targets; using the formula Determine the movement speed s. For a certain area, n represents the number of all moving targets in the area within the time period t, and n s Represents the number of moving targets with a speed less than or equal to the moving speed s within the area within a period of time t. Formula It means if If it is exactly equal to the suspected abnormal motion target determination threshold α, then the speed value s is called the speed threshold. The system determines all moving targets with speed values greater than s within the region at the current moment as suspected abnormal motion targets. If it is not exactly equal to but closest to the suspected abnormal motion target determination threshold α, then the speed value s is also called the speed threshold.

[0071] When the traffic scenario is a highway section with a low-speed limit, that is, it is applied to a traffic scenario where too low a moving speed is abnormal (commonly found in highway sections with a minimum speed limit). The system analyzes the moving speeds of all moving targets in different regions within the t time period before the current moment, and estimates the moving speeds of all moving targets; using the formula to determine the speed s. For a certain region, n represents the number of all moving targets within the region in the t time period, and n g represents the number of moving targets within the region in the t time period with a speed greater than or equal to the moving speed g. The formula means that if is exactly equal to or closest to the suspected abnormal motion target determination threshold α, then the speed value g is called the speed threshold.

[0072] When the traffic scenario is a highway section with both a high-speed limit and a low-speed limit, that is, it is applied to a traffic scenario that restricts the maximum speed and needs to simultaneously identify ultra-low speed or stationary moving targets. At this time, two determination thresholds α1 and α2 are required. The system analyzes the moving speeds of all moving targets in different regions within the t time period before the current moment, and estimates the moving speeds of all moving targets; using the formula to determine the speed s. For a certain region, n represents the number of all moving targets within the region in the t time period, and n p represents the number of moving targets within the region in the t time period with a speed greater than or equal to the moving speed p; n q represents the number of moving targets within the region in the t time period with a speed less than or equal to the moving speed q. The size relationship between p and q is p < q. The formula indicates is exactly equal to or closest to the suspected abnormal motion target determination threshold α1, and the formula indicates is exactly equal to or closest to the suspected abnormal motion target determination threshold α2. The speed values p and q are respectively called the low speed threshold and the high speed threshold.

[0073] In order to determine the suspected abnormal movement speed, the determination of the suspected abnormal movement speed according to the target object movement speed threshold includes: when the traffic scene is a speed limit section in an urban area, and when the first movement speed is greater than the first threshold of the target object movement speed, the first movement speed is determined to be the suspected abnormal movement speed; when the traffic scene is a high-speed section with a low speed limit, and when the second movement speed is less than the second threshold of the target object movement speed, the second movement speed is determined to be the suspected abnormal movement speed; when the traffic scene is a high-speed section with both high and low speed limits, and when the third movement speed is less than the third threshold of the target object movement speed or when the third movement speed is greater than the fourth threshold of the target object movement speed, the third movement speed is determined to be the suspected abnormal movement speed.

[0074] Specifically, in this embodiment, when the traffic scene is a speed-limited section in an urban area, after the system determines that the movement speed s is the movement speed threshold, the system determines that all movement targets in the area at the current moment whose movement speed is greater than s are suspected abnormal movement targets. When the traffic scene is a high-speed section with a low-speed limit, after the system determines that the movement speed g is the speed threshold, the system determines that all movement targets in the area at the current moment whose speed value is less than g are suspected abnormal movement targets. When the traffic scene is a high-speed section with both high-speed and low-speed limits, that is, when the system determines that the movement speed p and the movement speed q are the low-speed threshold and the high-speed threshold respectively, the system determines that all movement targets in the area at the current moment whose movement speed is less than p or all movement targets in the area at the current moment whose movement speed is greater than q are suspected abnormal movement targets.

[0075] In order to determine the suspected abnormal motion direction, the motion direction is judged according to the motion direction, and the suspected abnormal motion direction is determined by the following steps: obtaining the maximum average motion direction change value v of the target object max ; Get the average moving direction change value of the target object When the movement direction is greater than When , the movement direction is a suspected abnormal movement direction; wherein β is a parameter greater than 0.8 and less than 1, and is adjusted according to the current traffic scene.

[0076] Specifically, the average movement direction change of all single targets in different regions within the t time period before the current moment is analyzed, and the average movement direction change of all moving targets is estimated based on the movement directions of the targets at different times. Let the average movement direction change value of the jth target in a region within the t time period be v j ; The average movement direction change value of all targets in the area within the t time period is The maximum average movement direction change value v of all targets in the area within the time period t maxThe average change in motion direction of a single target is calculated by dividing the target's movement in the analysis area into k time segments, and calculating the absolute value of the difference in motion direction between the current time segment and the previous time segment starting from the second time segment; since a total of k-1 absolute values ​​can be obtained, the average of these k-1 values ​​is the average change in motion direction of the current target.

[0077] Then, the average movement direction change value in the area at the current moment is greater than The target is determined as a moving target with a suspected abnormal moving direction. β is a parameter greater than 0.8 and less than 1. β can be adjusted according to the current traffic scene.

[0078] In order to determine the suspected abnormal motion trajectory, the motion trajectory is judged according to the motion trajectory, and the determination of the suspected abnormal motion trajectory includes the following steps: according to the motion trajectory, obtaining the maximum motion direction change value direction of the target object max ; When direction max >γ, the motion trajectory is a suspected abnormal motion trajectory; wherein γ is a preset motion direction threshold.

[0079] Specifically, based on the motion trajectory, it is assumed that the motion directions and average motion direction change values ​​of all single targets in different areas in the time period t before the current moment have been estimated. After the system obtains the motion trajectory of the target object, it is assumed that the maximum motion direction change value of a target estimated at the current moment in the area is direction max Then, if the target to be analyzed in the area at the current moment has met the judgment condition of suspected abnormal motion direction, if direction max >γ, the target is identified as a moving target with a suspected abnormal motion trajectory. γ is a preset threshold, which is set according to the current traffic scenario.

[0080] This embodiment provides a method for detecting abnormal targets in traffic scenes based on rules and learning, which can be applied to communication intelligent terminals. Figure 1 As shown, the method includes:

[0081] Step S300: Determine the target of abnormal motion in the traffic scene according to the suspected abnormal motion information and traffic rules.

[0082] In practice, the system may have inaccurate problems when judging abnormal motion targets in traffic scenes based on suspected abnormal motion information. Therefore, this embodiment uses the "intersection" of suspected abnormal motion targets and suspected illegal motion targets as the final abnormal motion target solution. That is, each area in the traffic scene is processed as follows: first, a suspected abnormal motion target determination threshold is set, and the suspected abnormal motion target is determined based on the threshold. Then, the suspected illegal motion target is determined based on traffic rules, and finally the "intersection" of the two is used as the final abnormal motion target, that is, the target that meets both the suspected abnormal motion information target and the suspected illegal motion target is used as the abnormal motion target, which improves the accuracy of the judgment.

[0083] In order to determine the target of abnormal motion in the traffic scene, the step of determining the abnormal motion target according to the suspected abnormal motion information and traffic rules includes the following steps:

[0084] Step S301, obtaining a priori values ​​set in the system according to traffic rules;

[0085] Step S302: determining an abnormal motion target according to the suspected abnormal motion speed, the suspected abnormal motion direction, the suspected abnormal motion trajectory and the prior value.

[0086] In practice, traffic rules include: motor vehicles and non-motor vehicles must drive on the right side of the road; if a road has a dedicated lane, only the specified vehicles are allowed to pass in the dedicated lane, and other vehicles are not allowed to enter the dedicated lane. When there is a traffic police on-site command, you should follow the traffic police's command; on roads without traffic signals, you should pass under the principle of ensuring safety and smooth traffic; the following situations do not allow overtaking: (i) the vehicle in front is turning left, turning around, or overtaking; (ii) there is a possibility of meeting oncoming vehicles; (iii) the vehicle in front is a police car, fire truck, ambulance, or engineering rescue vehicle performing an emergency task; when road maintenance vehicles and engineering vehicles are operating, their driving routes and directions are not restricted by traffic signs and markings without affecting the passage of passing vehicles, and passing vehicles and personnel should pay attention to avoid them. Sprinklers, road sweepers and other motor vehicles should operate in accordance with safety operation standards; if it does not affect the passage of other vehicles, they may not be restricted by vehicle lane separation, but they may not drive in the opposite direction; in addition, there are also regulations on slowing down near schools. In reality, the system may determine that the target object is suspected abnormal motion information, but it is not abnormal motion according to traffic rules. Therefore, the embodiment of the present invention sets the prior value of traffic rules in the system, and then in the actual traffic application scenario, the abnormal motion target is determined according to the suspected abnormal motion speed, the suspected abnormal motion direction, the suspected abnormal motion trajectory and the prior value set by the traffic rules. At this time, the abnormal motion target determined has a high accuracy rate.

[0087] Exemplary Devices

[0088] like Figure 2 As shown in, an embodiment of the present invention provides a traffic scene abnormal target detection device based on rules and learning, the device comprising: a traffic scene video data acquisition unit 401, a suspected abnormal motion information acquisition unit 402, and a traffic scene abnormal motion target determination unit 403; the traffic scene video data acquisition unit 401 is used to acquire real-time collected traffic scene video data;

[0089] The suspected abnormal motion information acquisition unit 402 is used to perform statistics and calculations on the traffic scene video data to obtain the suspected abnormal motion information of the target object;

[0090] The traffic scene abnormal motion target determination unit 403 is used to determine the target of the abnormal motion in the traffic scene according to the suspected abnormal motion information and traffic rules.

[0091] Based on the above embodiments, the present invention further provides an intelligent terminal, whose principle block diagram can be shown as follows: Figure 3 As shown. The intelligent terminal includes a processor, a memory, a network interface, a display screen, and a temperature sensor connected through a system bus. Among them, the processor of the intelligent terminal is used to provide computing and control capabilities. The memory of the intelligent terminal includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the intelligent terminal is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for detecting abnormal targets in traffic scenes based on rules and learning is implemented. The display screen of the intelligent terminal can be a liquid crystal display screen or an electronic ink display screen, and the temperature sensor of the intelligent terminal is pre-set inside the intelligent terminal to detect the operating temperature of the internal device.

[0092] Those skilled in the art will understand that Figure 3 The schematic diagram in the figure is only a block diagram of a part of the structure related to the scheme of the present invention, and does not constitute a limitation on the smart terminal to which the scheme of the present invention is applied. The specific smart terminal may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0093] In one embodiment, a smart terminal is provided, comprising a memory and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by one or more processors, and the one or more programs include instructions for performing the following operations:

[0094] Obtain real-time collected traffic scene video data;

[0095] Performing statistics and calculations on the traffic scene video data to obtain suspected abnormal motion information of the target object;

[0096] According to the suspected abnormal motion information and traffic rules, a target of abnormal motion in the traffic scene is determined.

[0097] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0098] In summary, the present invention discloses a method for detecting abnormal targets in traffic scenes based on rules and learning, an intelligent terminal, and a storage medium. The method includes: obtaining real-time collected traffic scene video data; performing statistics and calculations on the traffic scene video data to obtain suspected abnormal motion information of the target object; and determining the target of abnormal motion in the traffic scene according to the suspected abnormal motion information and traffic rules. The embodiment of the present invention performs statistics and calculations on the collected traffic scene video data to obtain suspected abnormal motion information of the target object, and then determines the target of abnormal motion in the traffic scene according to the suspected abnormal motion information and traffic rules. The amount of calculation is small, high real-time performance is achieved, and the determination of abnormal motion targets in the traffic scene is more accurate.

[0099] It should be understood that the present invention discloses a method for detecting abnormal targets in traffic scenes based on rules and learning. It should be understood that the application of the present invention is not limited to the above examples. For ordinary technicians in this field, it can be improved or transformed according to the above description. All these improvements and transformations should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A method for detecting abnormal targets in traffic scenes based on rules and learning, characterized in that: The method comprises: Acquire real-time collected traffic scene video data; the collection methods include magnetic frequency collection and band collection, wherein magnetic frequency collection is to measure the flow after triggering the video, and the band collection is to measure the speed after triggering the video; Performing statistics and calculations on the traffic scene video data to obtain suspected abnormal motion information of the target object; Determine the target of abnormal movement in the traffic scene according to the suspected abnormal movement information and traffic rules; The performing statistics and calculation on the traffic scene video data to determine the suspected abnormal motion information of the target object includes: Determine, based on the traffic scene video data, the moving speed of the target object in a preset time period, the moving direction of the target object in a preset time period, the moving trajectory of the target object in a preset time period, the current traffic scene of the target object, and the number of target objects in the current traffic scene; According to the traffic scene, the number of target objects, and the movement speed, statistics and calculation are performed to determine a suspected abnormal movement speed; According to the movement direction, the movement direction is judged to determine a suspected abnormal movement direction; According to the motion trajectory, the motion trajectory is judged to determine a suspected abnormal motion trajectory; According to the movement speed, the current traffic scene of the target object and the number of target objects in the current traffic scene, a suspected abnormal moving target determination threshold is calculated through a learning and training algorithm, and the suspected abnormal moving target is determined based on a comparison between the movement speed and the movement speed threshold; the learning and training algorithm adopts a neural network; The determining of the suspected abnormal movement speed according to the traffic scene, the number of target objects, and the movement speed includes: According to the traffic scene, the movement speed and the number of target objects, a target object movement speed threshold is determined through statistics and calculation; wherein, when the traffic scene is a speed-limited section in an urban area, the movement speed is a first movement speed, and the number of target objects is a first target object number; When the traffic scene is a high-speed road section with a low-speed restriction, the movement speed is a second movement speed, and the number of target objects is a second number of target objects; When the traffic scene is a high-speed road section with both high-speed and low-speed restrictions, the movement speed is a third movement speed, and the number of target objects is a third number of target objects; Determining a suspected abnormal motion speed according to the target object motion speed threshold; Determining the target object movement speed threshold according to the traffic scene, the movement speed and the number of target objects includes: When the traffic scene is a speed-limited section in an urban area, determining a first threshold of the target object movement speed according to the first movement speed and the first number of target objects; When the traffic scene is a high-speed road section with a low-speed restriction, determining a second threshold value of the target object movement speed according to the second movement speed and the second number of target objects; When the traffic scene is a high-speed road section that limits both high speed and low speed, determining a third target object movement speed threshold and a fourth target object movement speed threshold according to the third movement speed and the third number of target objects, wherein the third target object movement speed threshold is less than the fourth target object movement speed threshold; Determining the suspected abnormal motion speed according to the target object motion speed threshold comprises: When the traffic scene is a speed-limited section in an urban area, and when the first movement speed is greater than a first threshold of the movement speed of the target object, the first movement speed is determined to be a suspected abnormal movement speed; When the traffic scene is a high-speed road section with a low-speed restriction, and when the second movement speed is less than a second threshold of the target object movement speed, the second movement speed is determined to be a suspected abnormal movement speed; When the traffic scene is a high-speed road section that restricts both high and low speeds, and when the third motion speed is less than a third threshold of the target object motion speed or when the third motion speed is greater than a fourth threshold of the target object motion speed, the third motion speed is determined to be a suspected abnormal motion speed; The step of judging the movement direction according to the movement direction to determine the suspected abnormal movement direction includes: Get the maximum average motion direction change value v of the target object max ; Get the average movement direction change value of the target object When the movement direction is greater than When , the movement direction is a suspected abnormal movement direction; wherein β is a parameter greater than 0.8 and less than 1, and is adjusted according to the current traffic scene; The step of judging the motion trajectory according to the motion trajectory to determine a suspected abnormal motion trajectory includes: According to the motion trajectory, obtain the maximum motion direction change value direction of the target object max ; When direction max >γ, the motion trajectory is a suspected abnormal motion trajectory; wherein γ is a preset motion direction threshold; Determining the abnormal motion target according to the suspected abnormal motion information and traffic rules includes: Obtain the prior values ​​set in the system according to traffic rules; Determine an abnormal motion target according to the suspected abnormal motion speed, the suspected abnormal motion direction, the suspected abnormal motion trajectory and the prior value; When the traffic scene is a speed-limited section in an urban area, determining a first threshold of the target object movement speed according to the first movement speed and the first number of target objects includes: like If it is exactly equal to or closest to the suspected abnormal motion target determination threshold α, then the speed value s is the first threshold of the target object's motion speed; Where n represents the number of first target objects in any area within a time period of t, and n s Represents the number of moving targets whose first moving speed is less than or equal to the speed value s within the area within the time period t; When the traffic scene is a high-speed road section with a low-speed restriction, determining a second threshold of the target object movement speed according to the second movement speed and the second number of target objects includes: like If it is exactly equal to or closest to the suspected abnormal motion target determination threshold α, then the speed value g is the second threshold of the target object motion speed; Among them, n represents the number of second target objects in any area within the time period t, and n g Represents the number of moving targets whose second moving speed is greater than or equal to the speed value g within the time period t in the area; When the traffic scene is a high-speed road section that limits both high speed and low speed, determining a third target object movement speed threshold and a fourth target object movement speed threshold according to the third movement speed and the third number of target objects, wherein the third target object movement speed threshold is less than the fourth target object movement speed threshold, comprises: like If it is exactly equal to or closest to the suspected abnormal motion target determination threshold α1, then the speed value p is the third threshold of the target object motion speed; like If it is exactly equal to or closest to the suspected abnormal motion target determination threshold α2, then the speed value q is the fourth threshold of the target object motion speed; Among them, n represents the number of third target objects in any area within a period of time t, and n p represents the number of moving targets with a third moving speed greater than or equal to the moving speed p within the region within a time period t; n q represents the number of moving targets whose third moving speed is less than or equal to the moving speed q within the time period t in the area; The step of obtaining the average motion direction change value v of the target object comprises: The movement of the target in the analysis area is divided into k time segments. Starting from the second time segment, the absolute value of the difference in movement direction between the current time segment and the previous time segment is calculated, and a total of k-1 absolute values ​​are obtained. The average of the k-1 absolute values ​​is the average movement direction change value of the current target.

2. An intelligent terminal, characterized in that: The device comprises a memory and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by one or more processors, including for executing the method according to claim 1.

3. A non-transitory computer-readable storage medium, characterized in that: When the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the method as claimed in claim 1.

Images