A method, system and device for detecting objects thrown from a high altitude

Abstract

The present invention discloses a method, system and device for detecting objects dropped from a high altitude. First, a moving target to be detected in a monitoring video is determined, that is, a target that may be an object dropped from a high altitude. However, considering that there may be interference targets such as flying insects and night lights among these targets that do not belong to objects dropped from a high altitude, the present application further determines whether the number of inflection points in the motion trajectory of each moving target to be detected is less than a first preset number. If the number of inflection points in the motion trajectory of the moving target to be detected is less than the first preset number, this indicates that the falling motion of the moving target is more consistent with the free fall motion. Therefore, the moving target to be detected is determined to be a target dropped from a high altitude, thereby avoiding determining irregularly moving targets such as flying birds and night lights as targets dropped from a high altitude, reducing the probability of false detection, improving the accuracy of detecting events dropped from a high altitude, and reducing the workload of manual screening.

Description

Technical Field

[0001] The present invention relates to the field of monitoring, and in particular to a method, system and device for detecting objects thrown from a high altitude. Background Art

[0002] In recent years, tragic events caused by high-altitude objects have occurred frequently, causing serious harm to people's life safety and property safety. Therefore, the management and monitoring of high-altitude objects have become increasingly important. In order to realize the automatic monitoring of high-altitude objects, it is necessary to establish a high-altitude object detection system to automatically detect and record high-altitude objects. In the prior art, after capturing the surveillance video captured by the camera, by extracting the moving target in the surveillance video captured by the camera, it is determined whether the longitudinal moving distance of the moving target in the surveillance video captured by the camera is greater than a certain threshold. When it is greater than a certain threshold, it is determined that the moving target has a throwing event. However, this method is easy to judge that irregularly moving targets such as flying insects and night lights have also been thrown events, and the probability of false detection is high. The accuracy of the detection of throwing events is low, which increases the workload of manual screening. Summary of the invention

[0003] The purpose of the present invention is to provide a method, system and device for detecting objects thrown from high altitudes, so as to avoid judging irregularly moving targets such as flying birds and night lights as objects thrown from high altitudes, thereby reducing the probability of false detection, improving the accuracy of detecting objects thrown from high altitudes, and reducing the workload of manual screening.

[0004] In order to solve the above technical problems, the present invention provides a method for detecting objects thrown from a high altitude, comprising:

[0005] Determine the moving target to be detected in the surveillance video;

[0006] Acquire the motion trajectories of all the moving targets to be detected according to the monitoring video;

[0007] Determining the number of inflection points in the motion trajectory of each of the motion targets to be detected according to the motion trajectory of each of the motion targets to be detected;

[0008] Determining whether the number of inflection points in the motion trajectory of each of the moving targets to be detected is less than a first preset number;

[0009] If it is less than, it is determined that the moving target to be detected whose number of inflection points in the motion trajectory is less than the first preset number is a high-altitude parabolic target.

[0010] Preferably, determining the number of inflection points in the motion trajectory of each of the motion targets to be detected according to the motion trajectory of each of the motion targets to be detected includes:

[0011] A functional relationship formula for fitting a trajectory fitting curve of the moving target to be detected according to the motion trajectory of the moving target to be detected;

[0012] Calculate the number of inflection points of the trajectory fitting curve of the moving target to be detected according to the functional relationship;

[0013] The number of inflection points of the trajectory fitting curve of the moving object to be detected is used as the number of inflection points in the motion trajectory of the moving object to be detected.

[0014] Preferably, before determining the moving target to be detected in the surveillance video, the method further includes:

[0015] Determine the number of all moving objects in the surveillance video;

[0016] Determine whether the number of all moving targets in the surveillance video is less than a second preset number;

[0017] If it is less than, the process proceeds to the step of determining the moving target to be detected in the surveillance video.

[0018] Preferably, before entering the step of determining the moving target to be detected in the surveillance video, the method further includes:

[0019] Determining whether the number of all moving targets in the monitored area in the monitoring video is less than a third preset number;

[0020] If it is less than, the process proceeds to the step of determining the moving target to be detected in the surveillance video.

[0021] Preferably, determining a moving target to be detected in a surveillance video includes:

[0022] Obtaining the surveillance video obtained after video acquisition of the surveillance area;

[0023] Determine a moving target in the surveillance video, and determine whether a falling distance of the moving target is greater than a preset threshold;

[0024] If the falling distance of the moving target is greater than the preset threshold, the moving target is determined as a moving target to be detected.

[0025] Preferably, after determining that the number of inflection points in the motion trajectory of the moving target to be detected is less than a first preset number, the method further includes:

[0026] Determine, according to the motion trajectory of the moving target to be detected, that the number of inflection points in the motion trajectory is greater than the first preset number, that the first longitudinal movement distance of the moving target to be detected from appearance to falling distance in the monitoring video reaches the preset threshold value, and the second longitudinal movement distance of the moving target to be detected from appearance to disappearance in the monitoring video;

[0027] calculating a ratio of the first longitudinal movement distance to the second longitudinal movement distance;

[0028] Determining whether the ratio is less than a preset ratio;

[0029] If it is less than, the step of determining that the moving target to be detected whose number of inflection points in the motion trajectory is less than the first preset number is a high-altitude parabolic target is entered.

[0030] Preferably, determining the number of all moving targets in the surveillance video includes:

[0031] The number of all moving targets in the surveillance video is determined by Gaussian background modeling or VIBE algorithm.

[0032] Preferably, after determining that the number of inflection points in the motion trajectory of the moving target to be detected is less than a first preset number, the method further includes:

[0033] Inputting a surveillance video of a moving target to be detected, in which the number of inflection points in the motion trajectory is less than a first preset number, into a pre-trained deep classification model;

[0034] Using the deep classification model to determine whether the moving target to be detected is a preset high-altitude parabolic target;

[0035] If so, it is determined that the moving target to be detected that meets the preset high-altitude target is a high-altitude parabolic target.

[0036] The present invention also provides a high-altitude object throwing detection system, comprising:

[0037] A moving target determination unit to be detected, used to determine the moving target to be detected in the monitoring video;

[0038] A motion trajectory acquisition unit, used to acquire the motion trajectories of all the moving targets to be detected according to the monitoring video;

[0039] An inflection point number determining unit, used to determine the number of inflection points in the motion trajectory of each of the motion targets to be detected according to the motion trajectory of each of the motion targets to be detected;

[0040] The inflection point number judgment unit is used to judge whether the number of inflection points in the motion trajectory of each of the to-be-detected moving targets is less than a first preset number. If so, the to-be-detected moving target whose number of inflection points in the motion trajectory is less than the first preset number is determined to be a high-altitude parabolic target.

[0041] The present invention also provides a device for detecting objects thrown from a high altitude, comprising:

[0042] Memory for storing computer programs;

[0043] A processor is used to implement the steps of the method for detecting objects thrown from a high altitude as described above when executing the computer program.

[0044] The present invention discloses a method, system and device for detecting objects dropped from a high altitude. First, a moving target to be detected in a monitoring video is determined, that is, a target that may be an object dropped from a high altitude. However, considering that there may be interference targets such as flying insects and night lights among these targets that do not belong to objects dropped from a high altitude, the present application further determines whether the number of inflection points in the motion trajectory of each moving target to be detected is less than a first preset number. If the number of inflection points in the motion trajectory of the moving target to be detected is less than the first preset number, this indicates that the falling motion of the moving target is more consistent with the free fall motion. Therefore, the moving target to be detected is determined to be a target dropped from a high altitude, thereby avoiding determining irregularly moving targets such as flying birds and night lights as targets dropped from a high altitude, reducing the probability of false detection, improving the accuracy of detecting events dropped from a high altitude, and reducing the workload of manual screening. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the prior art and the drawings required for use in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0046] Figure 1 A flowchart of a method for detecting objects thrown from a high altitude provided by the present invention;

[0047] Figure 2 A schematic diagram of the structure of a high-altitude object throwing detection system provided by the present invention;

[0048] Figure 3 The present invention provides a schematic structural diagram of a device for detecting objects thrown from a high altitude. DETAILED DESCRIPTION

[0049] The core of the present invention is to provide a method, system and device for detecting objects thrown from high altitudes, which avoids judging irregularly moving targets such as flying birds and night lights as objects thrown from high altitudes, reduces the probability of false detection, improves the accuracy of detecting objects thrown from high altitudes, and reduces the workload of manual screening.

[0050] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments 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.

[0051] Please refer to Figure 1 , Figure 1 The present invention provides a flow chart of a method for detecting objects thrown from a high altitude.

[0052] A method for detecting objects thrown from a high altitude, comprising:

[0053] S1: Determine the moving target to be detected in the surveillance video;

[0054] Taking into account that moving targets other than high-altitude throwing objects may appear in the surveillance video, if all moving targets are carefully judged, it will increase unnecessary calculations and reduce the efficiency of detection. Therefore, in this embodiment, the moving targets to be detected in the surveillance video will be determined first. The moving targets to be detected here are moving targets that may be high-altitude throwing objects targets that have been determined through preliminary screening. However, the moving targets to be detected determined through this step still need to be further accurately screened to identify high-altitude throwing objects targets.

[0055] S2: Obtain the motion trajectories of all moving targets to be detected based on the surveillance video;

[0056] In the prior art, after the longitudinal falling distance of the moving target reaches a threshold, it is determined to be a high-altitude parabolic target. However, for some irregularly moving non-high-altitude parabolic targets, the result obtained by only detecting the longitudinal falling distance is one-sided and cannot make an accurate judgment. In order to further accurately identify the high-altitude parabolic, it is necessary to conduct a comprehensive analysis of its motion trajectory. Therefore, in this embodiment, in order to make a judgment based on the motion trajectory of the moving target to be detected, it is necessary to first obtain the motion trajectory of the moving target to be detected. The motion trajectory of the moving target to be detected can be extracted by obtaining the video screen from the appearance to the disappearance of the moving target to be detected in the monitoring video. For example, the motion trajectory of the moving target to be detected is determined by target tracking. After determining the moving target to be detected, the moving target to be detected is added to the tracking module for tracking, and the tracking result is obtained to obtain the motion trajectory of the moving target to be detected, and further judgment is made on whether it is a high-altitude parabolic according to the motion trajectory.

[0057] S3: determining the number of inflection points in the motion trajectory of each of the motion targets to be detected according to the motion trajectory of each of the motion targets to be detected, and the concavity and convexity of the motion trajectory of the motion target to be detected changes at the inflection points;

[0058] Considering that flying birds, night lights and other moving targets that are easily misdetected as high-altitude parabolic targets have the most significant difference from high-altitude parabolic targets in that their longitudinal direction of movement frequently changes during the movement process, in this embodiment, the concept of inflection point is introduced, that is, the point where the convexity of the function of the curve corresponding to the motion trajectory of the moving target to be detected changes. For example, at the inflection point, the function of the curve corresponding to the motion trajectory of the moving target to be detected changes from a concave function to a convex function, or from a convex function to a concave function. The number of inflection points in the motion trajectory of each moving target to be detected is determined to facilitate subsequent analysis of whether the motion of the moving target to be detected conforms to free fall motion, and whether it is a high-altitude parabolic target is judged based on the result of whether it conforms to free fall motion.

[0059] S4: Determine whether the number of inflection points in the motion trajectory of each moving target to be detected is less than a first preset number;

[0060] S5: If it is less than, it is determined that the moving target to be detected whose number of inflection points in the motion trajectory is less than the first preset number is a high-altitude parabolic target.

[0061] Considering that for moving targets such as flying birds and night lights that are easily misdetected as high-altitude parabolic targets, the number of inflection points in the function corresponding to the curve of their motion trajectory is usually large, while for high-altitude parabolic targets with lighter mass such as clothing and plastic bags, there may be fewer inflection points in the function corresponding to the curve of their motion trajectory when falling. Therefore, in this embodiment, a first preset number is set. The first preset number can be a number value determined by technicians based on the statistical average of the number of inflection points in the motion trajectory of high-altitude parabolic targets with lighter mass such as clothing and plastic bags and the statistical average of the number of inflection points in the motion trajectory of moving targets such as flying birds and night lights that are easily misdetected as high-altitude parabolic targets, which helps to distinguish between high-altitude parabolic targets and non-high-altitude parabolic targets. This application does not limit the method for determining the first preset number.

[0062] Specifically, by judging whether the number of inflection points in the motion trajectory of the motion target to be detected is greater than a first preset number, it is determined whether the motion target to be detected is a high-altitude parabolic target. When the number of inflection points in the motion trajectory of the motion target to be detected is less than the first preset number, it means that its falling pattern is more consistent with free fall motion. Therefore, the motion target to be detected whose number of inflection points in the motion trajectory is less than the first preset number is determined to be a high-altitude parabolic target.

[0063] To sum up, in this embodiment, the moving targets to be detected in the monitoring video are first determined, that is, targets that may be objects thrown from high altitudes. However, considering that there may be interfering targets such as flying insects and night lights among these targets that are not objects thrown from high altitudes, the present application will also determine whether the number of inflection points in the motion trajectory of each moving target to be detected is less than a first preset number. If the number of inflection points in the motion trajectory of the moving target to be detected is less than the first preset number, this indicates that the falling motion of the moving target is more consistent with the free fall motion. Therefore, the moving target to be detected is determined to be a target of objects thrown from high altitudes, avoiding the determination of irregularly moving moving targets such as flying birds and night lights as targets of objects thrown from high altitudes, reducing the probability of false detection, improving the accuracy of detecting objects thrown from high altitudes, and reducing the workload of manual screening.

[0064] Based on the above embodiments:

[0065] As a preferred embodiment, determining the number of inflection points in the motion trajectory of each moving target to be detected according to the motion trajectory of each moving target to be detected includes:

[0066] A functional relationship of fitting a trajectory fitting curve of the moving target to be detected according to the moving trajectory of the moving target to be detected;

[0067] Calculate the number of inflection points of the trajectory fitting curve of the moving target to be detected according to the functional relationship;

[0068] The number of inflection points of the trajectory fitting curve of the moving target to be detected is taken as the number of inflection points in the motion trajectory of the moving target to be detected.

[0069] In order to more accurately determine the number of inflection points in the motion trajectory of the moving target to be detected, in this embodiment, the motion trajectory of the moving target to be detected from appearance to disappearance in the monitoring video is first obtained according to the monitoring video. The motion trajectory obtained here is a plurality of discrete trajectory points. The trajectory curve of the moving target to be detected is fitted according to all the trajectory points of the moving target to be detected, and the functional relationship of the trajectory curve is obtained;

[0070] p(x)=p1x n +p2x n-1 +p3x n-2 +…+p n x+p n+1

[0071] a=f(x,y,n);

[0072] Among them, x is the horizontal coordinate of the trajectory point, y is the vertical coordinate of the trajectory point, n is the polynomial order, a is the independent variable coefficient after the trajectory curve is fitted, f(x, y, n) is the polynomial fitting function, p(x) is the Nth degree polynomial about x, p iis the coefficient of each term in the polynomial, 1≤i≤n+1. Then calculate the number of inflection points.

[0073] The value of the dependent variable is obtained, and the number of inflection points of the curve of the polynomial fitting function is solved, that is, the second-order derivative of the function is calculated, and the number of points where the second-order derivative is 0 and the second-order derivatives on the left and right sides have different signs, and the third-order derivative is not 0, as well as the number of points where the second-order derivative does not exist, is used as the number of inflection points in the motion trajectory of the motion target to be detected. The number of inflection points in the motion trajectory of the motion target to be detected is determined, and the number of inflection points determined by mathematical calculation is more accurate and reliable.

[0074] As a preferred embodiment, before determining the moving target to be detected in the surveillance video, the method further includes:

[0075] Determine the number of all moving objects in the surveillance video;

[0076] Determine whether the number of all moving targets in the surveillance video is less than a second preset number;

[0077] If it is less than, the process proceeds to the step of determining the moving target to be detected in the surveillance video.

[0078] Considering that on some snowy or windy days, the camera lens of the surveillance video will shake violently, resulting in a serious reduction in the accuracy of obtaining the moving target, and the number of moving targets obtained from the surveillance video is abnormally large. In this case, it is impossible to detect high-altitude objects according to the surveillance video at this time. Therefore, in this embodiment, before determining the moving target to be detected in the surveillance video, the number of all moving targets in the surveillance video will be determined first, and it will be determined whether the number is less than the second preset threshold. The second preset threshold here is set by the technician according to the number of all moving targets identified in the surveillance video in the shaking scene such as snowy or windy days. If the number of all moving targets in the surveillance video is not less than the second preset threshold, it means that the number of moving targets in the current scene is abnormal, and it is a shaking scene such as snowy or windy days, and it is impossible to detect high-altitude objects. If the number of all moving targets in the surveillance video is less than the second preset threshold, it means that the number of moving targets in the current scene is normal, and the current scene is not a shaking scene, so it can enter the step of determining the moving target to be detected in the surveillance video. This embodiment adds automatic recognition of shaking scenes to avoid invalid high-altitude object detection in shaking scenes.

[0079] As a preferred embodiment, before entering the step of determining the moving target to be detected in the surveillance video, the method further includes:

[0080] Determining whether the number of all moving targets in the monitored area in the monitoring video is less than a third preset number;

[0081] If it is less than, the process proceeds to the step of determining the moving target to be detected in the surveillance video.

[0082] The monitoring video includes the area to be monitored and the area not to be monitored. The area to be monitored refers to the area with a high probability of high-altitude dropping, such as a high-rise building, which is artificially demarcated by the technical staff, while the area not to be monitored refers to the area other than the area to be monitored in the monitoring video, and the probability of high-altitude dropping in these areas is low, such as the sky. Considering that in some cases, the number of moving targets in the area to be monitored may be normal, but the total number of moving targets in the monitoring video may be abnormal. For example, there is a red flag fluttering in the wind, or a group of flying birds in the area not to be monitored, but this is not a shaking scene, and the area to be monitored can still be detected for high-altitude dropping. In order to solve the above problem, in this embodiment, a third preset number is set. The third preset threshold here is set by the technical staff according to the number of moving targets identified in the area to be monitored in the monitoring video in shaking scenes such as snowy days or strong winds. By judging whether the number of all moving targets in the monitored area in the surveillance video is less than a third preset number, whether the current scene is a shaking scene can be further identified. If the number of all moving targets in the monitored area in the surveillance video is less than the third preset number, the step of determining the moving targets to be detected in the surveillance video is entered, thereby further improving the accuracy of identifying the shaking scene.

[0083] As a preferred embodiment, determining a moving target to be detected in a surveillance video includes:

[0084] Obtaining the surveillance video obtained after video acquisition of the surveillance area;

[0085] Determine the moving target in the surveillance video and judge whether the falling distance of the moving target is greater than a preset threshold;

[0086] If the falling distance of the moving target is greater than a preset threshold, the moving target is determined as a moving target to be detected.

[0087] In this embodiment, a preset threshold is set to select moving targets that may be targets of high-altitude throwing objects from all moving targets as moving targets to be detected. The falling distance here refers to the distance that an object falls from top to bottom. The implementation method is simple and can quickly screen out moving targets that are obviously not targets of high-altitude throwing objects from a large number of moving targets. The preset threshold here can be, but is not limited to, set by technicians based on the height of the coverage area of ​​the area to be monitored or the height of common high-altitude throwing behaviors.

[0088] As a preferred embodiment, after determining that the number of inflection points in the motion trajectory of the moving target to be detected is less than a first preset number, the method further includes:

[0089] Determine, according to the motion trajectory of the moving target to be detected, a first longitudinal movement distance from the appearance to the falling distance of the moving target to be detected in the monitoring video that reaches a preset threshold value and a second longitudinal movement distance from the appearance to the disappearance of the moving target to be detected in the monitoring video when the number of inflection points in the motion trajectory is greater than a first preset number;

[0090] Calculating a ratio of the first longitudinal movement distance to the second longitudinal movement distance;

[0091] Determine whether the ratio is less than a preset ratio;

[0092] If it is less than, the step of determining that the moving target to be detected whose number of inflection points in the motion trajectory is less than the first preset number is a high-altitude parabolic target is entered.

[0093] Considering that for high-altitude parabolic targets, the longitudinal moving distance from appearance to disappearance in the monitoring video is usually large, while for non-high-altitude parabolic targets that are easily misjudged, such as night lights, they may appear in the monitoring video and suddenly disappear, that is, the longitudinal moving distance from appearance to disappearance is short. Therefore, in this embodiment, in order to further accurately screen out high-altitude parabolic targets from moving targets whose number of inflection points is less than the first preset number, the first longitudinal moving distance from appearance to falling distance of the moving target to be detected in the monitoring video with the number of inflection points in the motion trajectory greater than the first preset number reaching the preset threshold and the second longitudinal moving distance from appearance to disappearance of the moving target to be detected in the monitoring video will be determined according to the motion trajectory of the moving target to be detected. Since the monitoring video is composed of video frames, the first longitudinal moving distance from appearance to falling distance of the moving target to be detected in the monitoring video reaching the preset threshold is greater than or equal to the preset threshold. Calculate the ratio of the first longitudinal moving distance to the second longitudinal moving distance. If this ratio is smaller, it means that the longitudinal moving distance of the moving target from appearance to disappearance is larger, which means that the ratio of the falling distance of the target to the complete falling trajectory when the target meets the falling threshold is smaller, that is, it disappears after a certain distance of stable falling, rather than the target disappearing when it reaches the falling threshold, so as to screen out invalid targets such as flashing lights with a short longitudinal distance. The larger the ratio, the smaller the longitudinal moving distance of the moving target from appearance to disappearance, which means that the moving target suddenly disappears during the movement, which may be a night light or a flying bird. By comparing the calculated ratio with the preset ratio, if the calculated ratio is less than the preset ratio, it is determined to be a high-altitude parabolic target.

[0094] As a preferred embodiment, determining the number of all moving targets in the surveillance video includes:

[0095] The number of all moving targets in the surveillance video is determined by Gaussian background modeling or VIBE (Visual Background Extractor) algorithm. Gaussian background modeling is a background representation method based on the statistical information of pixel samples. It uses statistical information such as the probability density of a large number of sample values ​​of pixels over a long period of time to represent the background, and then uses statistical differences to judge target pixels. It can model complex dynamic backgrounds. The VIBE algorithm is also a background modeling method. The algorithm uses neighborhood pixels to create a background model and detects the foreground by comparing the background model with the current input pixel value. It has the advantages of simple concept, easy implementation, and high computational efficiency.

[0096] As a preferred embodiment, after determining that the number of inflection points in the motion trajectory of the moving target to be detected is less than a first preset number, the method further includes:

[0097] Inputting a surveillance video of a moving target to be detected, in which the number of inflection points in the motion trajectory is less than a first preset number, into a pre-trained deep classification model;

[0098] Use the deep classification model to determine whether the moving target to be detected is a preset high-altitude parabolic target;

[0099] If so, the moving target to be detected that meets the preset high-altitude target is determined to be a high-altitude parabolic target.

[0100] In this embodiment, in order to further screen the interference target, if the number of inflection points in the motion trajectory of the moving target to be detected is less than the first preset number, the moving target to be detected is classified by the deep classification model. The deep classification model is pre-trained with interference targets such as flying birds, rain and snow, high-altitude parabolic targets such as paper balls, garbage bags, bottles, cartons, clothes, flower pots and other targets to learn the difference between high-altitude parabolic targets and non-high-altitude parabolic targets. A small picture of the moving target to be detected can be obtained through the monitoring video, and put into the deep classification model to determine whether the moving target to be detected is a preset high-altitude parabolic target, such as paper balls, garbage bags, bottles, cartons, clothes, flower pots, etc. If the moving target to be detected is a preset high-altitude parabolic target, the moving target to be detected that meets the preset high-altitude target is determined to be a high-altitude parabolic target, so that the high-altitude parabolic target can be reported or the user can be alerted later. If the moving target to be detected is not a preset high-altitude parabolic target, it will not be reported.

[0101] Please refer to Figure 2 , Figure 2 The present invention provides a structural schematic diagram of a high-altitude object throwing detection system.

[0102] The present invention also provides a high-altitude object throwing detection system, comprising:

[0103] A moving target to be detected determining unit 21, used to determine a moving target to be detected in a surveillance video;

[0104] A motion trajectory acquisition unit 22 is used to acquire the motion trajectories of all moving targets to be detected according to the monitoring video;

[0105] An inflection point number determination unit 23, used to determine the number of inflection points in the motion trajectory of each of the motion targets to be detected according to the motion trajectory of each of the motion targets to be detected, where the concavity and convexity of the motion trajectory of the motion target to be detected changes at the inflection points;

[0106] The inflection point number judgment unit 24 is used to judge whether the number of inflection points in the motion trajectory of each moving target to be detected is less than a first preset number. If less than, the moving target to be detected whose number of inflection points in the motion trajectory is less than the first preset number is determined to be a high-altitude parabolic target.

[0107] For the relevant introduction of the high-altitude object throwing detection system, please refer to the above embodiment, and this application will not go into details here.

[0108] Please refer to Figure 3 , Figure 3 The present invention provides a schematic structural diagram of a device for detecting objects thrown from a high altitude.

[0109] The present invention also provides a device for detecting objects thrown from a high altitude, comprising:

[0110] A memory 31, used for storing computer programs;

[0111] The processor 32 is used to implement the steps of the method for detecting objects thrown from a high altitude as described above when executing a computer program.

[0112] For the relevant introduction of the high-altitude object throwing detection device, please refer to the above embodiment, and this application will not go into details here.

[0113] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0114] It should also be noted that, in this specification, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0115] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for detecting objects thrown from a high altitude, characterized in that: include: Determine a moving target to be detected in a surveillance video; wherein, the determining of the moving target to be detected in the surveillance video comprises: obtaining a surveillance video obtained after performing video acquisition on the surveillance area; determining the moving target in the surveillance video, and determining whether the falling distance of the moving target is greater than a preset threshold; if the falling distance of the moving target is greater than the preset threshold, determining the moving target as the moving target to be detected; Acquire the motion trajectories of all the moving targets to be detected according to the monitoring video; Determining the number of inflection points in the motion trajectory of each of the motion targets to be detected according to the motion trajectory of each of the motion targets to be detected; Determine whether the number of inflection points in the motion trajectory of each of the moving targets to be detected is less than a first preset number; wherein the first preset number is a number value determined by the statistical average of the number of inflection points in the motion trajectory of a light-weight high-altitude parabolic target and the statistical average of the number of inflection points in the motion trajectory of a moving target that is falsely detected as a high-altitude parabolic target and used to distinguish between high-altitude parabolic targets and non-high-altitude parabolic targets; Wherein, after determining that the number of inflection points in the motion trajectory of the moving target to be detected is less than a first preset number, the method further includes: determining, according to the motion trajectory of the moving target to be detected, a first longitudinal moving distance of the moving target to be detected whose number of inflection points in the motion trajectory is greater than the first preset number and whose distance from appearance to falling in the monitoring video reaches the preset threshold value, and a second longitudinal moving distance of the moving target to be detected from appearance to disappearance in the monitoring video; calculating a ratio of the first longitudinal moving distance to the second longitudinal moving distance; and determining whether the ratio is less than a preset ratio; If it is less than, it is determined that the moving target to be detected whose number of inflection points in the motion trajectory is less than the first preset number is a high-altitude parabolic target.

2. The method for detecting objects thrown from a height as claimed in claim 1, characterized in that: Determining the number of inflection points in the motion trajectory of each of the motion targets to be detected according to the motion trajectory of each of the motion targets to be detected includes: A functional relationship formula for fitting a trajectory fitting curve of the moving target to be detected according to the motion trajectory of the moving target to be detected; Calculate the number of inflection points of the trajectory fitting curve of the moving target to be detected according to the functional relationship; The number of inflection points of the trajectory fitting curve of the moving object to be detected is used as the number of inflection points in the motion trajectory of the moving object to be detected.

3. The method for detecting objects thrown from a height as claimed in claim 1, characterized in that: Before determining the moving target to be detected in the surveillance video, the following steps are also included: Determine the number of all moving objects in the surveillance video; Determine whether the number of all moving targets in the surveillance video is less than a second preset number; If it is less than, the process proceeds to the step of determining the moving target to be detected in the surveillance video.

4. The method for detecting objects thrown from a height as claimed in claim 3, characterized in that: Before entering the step of determining the moving target to be detected in the surveillance video, the following steps are also included: Determining whether the number of all moving targets in the monitored area in the monitoring video is less than a third preset number; If it is less than, the process proceeds to the step of determining the moving target to be detected in the surveillance video.

5. The method for detecting objects thrown from a height as claimed in claim 3, characterized in that: Determining the number of all moving targets in the surveillance video includes: The number of all moving targets in the surveillance video is determined by Gaussian background modeling or VIBE algorithm.

6. The method for detecting objects thrown from a height as claimed in any one of claims 1 to 5, characterized in that: After determining that the number of inflection points in the motion trajectory of the moving target to be detected is less than a first preset number, the method further includes: Inputting a surveillance video of a moving target to be detected, in which the number of inflection points in the motion trajectory is less than a first preset number, into a pre-trained deep classification model; Using the deep classification model to determine whether the moving target to be detected is a preset high-altitude parabolic target; If so, it is determined that the moving target to be detected that meets the preset high-altitude target is a high-altitude parabolic target.

7. A high-altitude object detection system, characterized in that: include: A unit for determining a moving target to be detected is used to determine a moving target to be detected in a surveillance video; wherein, the determining of the moving target to be detected in the surveillance video comprises: obtaining a surveillance video obtained after video acquisition of the surveillance area; determining the moving target in the surveillance video, and judging whether the falling distance of the moving target is greater than a preset threshold; if the falling distance of the moving target is greater than the preset threshold, determining the moving target as the moving target to be detected; A motion trajectory acquisition unit, used to acquire the motion trajectories of all the moving targets to be detected according to the monitoring video; An inflection point number determining unit, used to determine the number of inflection points in the motion trajectory of each of the motion targets to be detected according to the motion trajectory of each of the motion targets to be detected; An inflection point number determination unit, used to determine whether the number of inflection points in the motion trajectory of each of the moving targets to be detected is less than a first preset number; The first preset number is a number value used to distinguish between high-altitude parabolic targets and non-high-altitude parabolic targets, which is determined by the statistical average of the number of inflection points of the light-weight high-altitude parabolic targets in their motion trajectories and the statistical average of the number of inflection points of the moving targets that are mistakenly detected as high-altitude parabolic targets in their motion trajectories; Wherein, after determining that the number of inflection points in the motion trajectory of the moving target to be detected is less than a first preset number, the method further includes: determining, according to the motion trajectory of the moving target to be detected, a first longitudinal moving distance of the moving target to be detected whose number of inflection points in the motion trajectory is greater than the first preset number and whose distance from appearance to falling in the monitoring video reaches the preset threshold value, and a second longitudinal moving distance of the moving target to be detected from appearance to disappearance in the monitoring video; calculating a ratio of the first longitudinal moving distance to the second longitudinal moving distance; and determining whether the ratio is less than a preset ratio; If it is less than, it is determined that the moving target to be detected whose number of inflection points in the motion trajectory is less than the first preset number is a high-altitude parabolic target.

8. A device for detecting objects thrown from a high altitude, characterized in that: include: Memory for storing computer programs; A processor is used to implement the steps of the method for detecting high-altitude objects as described in any one of claims 1 to 6 when executing the computer program.

Images