CN109309811B - High-altitude parabolic detection system and method based on computer vision - Google Patents
High-altitude parabolic detection system and method based on computer vision Download PDFInfo
- Publication number
- CN109309811B CN109309811B CN201811015244.1A CN201811015244A CN109309811B CN 109309811 B CN109309811 B CN 109309811B CN 201811015244 A CN201811015244 A CN 201811015244A CN 109309811 B CN109309811 B CN 109309811B
- Authority
- CN
- China
- Prior art keywords
- altitude parabolic
- moving target
- track
- moving
- image data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
- H04N7/181—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast for receiving images from a plurality of remote sources
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N5/00—Details of television systems
- H04N5/14—Picture signal circuitry for video frequency region
- H04N5/144—Movement detection
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a high-altitude parabolic detection system and a high-altitude parabolic detection method based on computer vision, which comprise the following steps: the video image acquisition unit is used for comprehensively monitoring the cell in real time and acquiring image data; a network transmission unit for transmitting image data; the moving target detection unit is used for receiving the image data from the network transmission unit, analyzing and processing the acquired image data and determining a moving target in the image data; the central processing unit is used for tracking the moving target, acquiring the characteristics and the moving track of the moving target, judging whether the moving target belongs to a high-altitude parabolic object or not, and acquiring a throwing position of the high-altitude parabolic object, and relates to the field of community security systems. The invention has wide application range, strong universality and good detection effect.
Description
Technical Field
The invention relates to the field of community security systems, in particular to a high-altitude parabolic detection system and method based on computer vision.
Background
The high-altitude throwing is called as 'pain over a city', the high-altitude throwing is paid much attention in the past, the high-altitude throwing is used as an urban civilization behavior, the social hazard brought by the high-altitude throwing is large, through scientific calculation and testing, 30 g of eggs thrown from a floor about 12 meters in height are thrown onto a body to cause a swelling bag, about 54 meters of eggs are thrown to break a skull of a person, about 75 meters of eggs are thrown to cause the person to die on the spot, and irreparable damage is caused. Because the implementation place of the non-civilized behavior has a plurality of high-altitude floors, few witnesses exist, the parabolic time is short, and even someone intentionally hides the shadow during parabolic movement, so that law responsibility of the parabolic user is difficult to be traced by law enforcement departments, and the events are rare and forbidden frequently. The high-altitude parabolic behavior destroys the living environment, endangers the life safety of people, infringes the legal rights and interests of people, and the management and treatment relationship of the high-altitude parabolic behavior is stable in social security.
At present, the monitoring of high-altitude parabolic behaviors mainly depends on methods such as casual patrol of spontaneous organization of residents, and warning slogans posted in places where high-altitude parabolic behaviors frequently occur; but the irregular patrol method cannot completely monitor the behavior of the high-altitude object, and consumes a great deal of manpower; the method of pasting the warning slogan is easy to be ignored, cannot arouse enough attention, and has very limited effect. On the premise, in order to further enhance the prevention of the illiterate behavior which is extremely harmful to the high-altitude parabolic behavior, a high-altitude parabolic detection system based on computer vision needs to be built, so that after the high-altitude parabolic event occurs, later evidence obtaining and legal responsibility of related personnel can be facilitated, and the high-altitude parabolic behavior is effectively prevented by mainly using civilized propaganda and assisting in monitoring and management.
Disclosure of Invention
The invention aims to overcome the defects of the background technology and provide a high-altitude parabolic detection system and method based on computer vision, which are convenient for post-event evidence obtaining.
The invention provides a high-altitude parabolic detection system based on computer vision, which comprises:
the video image acquisition unit is used for comprehensively monitoring the cell in real time and acquiring image data;
a network transmission unit for transmitting image data;
the moving target detection unit is used for receiving the image data from the network transmission unit, analyzing and processing the acquired image data and determining a moving target in the image data;
and the central processing unit is used for tracking the moving target, acquiring the characteristics and the moving track of the moving target, judging whether the moving target belongs to a high-altitude parabolic object or not and acquiring the throwing position of the high-altitude parabolic object.
On the basis of the technical scheme, the video image acquisition unit comprises a plurality of network cameras and a data stream processing unit, the network cameras are arranged at different positions in a cell, and the data stream processing unit is used for acquiring video data streams of the network cameras and decoding the video data streams into image data by adopting an H265 decoding technology.
On the basis of the technical scheme, the high-altitude parabolic detection system based on computer vision further comprises a video storage unit, wherein the video storage unit is used for caching the video stream data of the motion trail of the high-altitude parabolic target judged by the central processing unit.
On the basis of the technical scheme, the video storage unit compresses the image data belonging to the high-altitude parabolic target into video stream data by an H265 compression method.
On the basis of the technical scheme, the video image acquisition unit, the network transmission unit, the moving target detection unit and the central processing unit are used for data communication by using the buffer area.
The invention also provides a high-altitude parabolic detection method based on computer vision, which comprises the following steps:
1) collecting an image;
the method comprises the steps that a comprehensive real-time monitoring cell obtains parabolic video stream data, and the video stream data are decoded into image data;
2) detecting a moving target;
analyzing and processing the image data, and determining a moving target in the image data;
3) tracking a moving target;
tracking each moving target, and acquiring the characteristics and the moving track of the moving target;
4) judging high altitude parabolic;
and analyzing and judging whether the moving target belongs to a high-altitude parabolic object or not, and acquiring a throwing position of the high-altitude parabolic object.
On the basis of the technical scheme, in the step 2), the moving object detection step is as follows:
(1) storing a sample set for each pixel in a first frame of image data, wherein pixel values of adjacent points around the pixel are sampling values of the sample set;
(2) comparing the pixel value of the pixel point of each subsequent frame with the sampling value of the sample set, and if the difference value is smaller than a threshold value, considering that the two points are close;
(3) sequentially traversing all sample sets to obtain the number of the pixel points close to the sample sets, if the number is larger than a threshold value, considering the pixel point as a background point, and otherwise, considering the pixel point as a foreground point;
(4) and screening the obtained foreground points, screening out objects which do not meet the requirements according to the perimeter L and the area S of the objects, carrying out corrosion, expansion and connectivity analysis on the detection result to remove detection noise, finally obtaining a moving target with 7 × 7 pixels at minimum, and extracting the characteristics of the moving target.
On the basis of the technical scheme, in the step 3), the moving target tracking step is as follows:
(1) acquiring a central point of the moving object and a sectional drawing of the moving object in the step 2);
(2) initializing the environment of a Kalman filtering algorithm in a first frame, and defaulting all moving targets as targets to be tracked;
(3) then, each moving object of each frame is matched with the existing track through a Hungarian algorithm, and if the matching is successful, the moving objects are added into the corresponding track; if the matching fails, predicting the position of the track in the current frame through a Kalman filtering algorithm;
(4) if a certain track is not successfully matched with the moving object for a plurality of times, namely the number of track points obtained by continuous prediction through a Kalman filtering algorithm is greater than a specific value A, the track is considered to be finished;
(5) if the number of points of a certain track is larger than the threshold value B, the track is considered to be effective, otherwise, the track is considered to be ineffective.
On the basis of the above technical solution, in step 4), the conditions for determining the high altitude parabola are as follows:
(1) the value of the ordinate of the moving target trajectory is continuously increasing;
(2) the number M of the high-altitude parabolic motion target track points is greater than a preset threshold value T, namely M is greater than T;
(3) the high-altitude parabolic motion target track is ended from the boundary of the monitoring range;
(4) the parabolic motion makes an accelerating motion at the beginning part;
(5) in the falling process of the high-altitude parabolic motion target, the difference value of the ordinate is not always too small.
On the basis of the technical scheme, in the step 4), the step of acquiring the throwing position belonging to the high-altitude parabolic target comprises the following steps:
(1) for each track, fitting all points on the track by using a least square method, using the frame number as an independent variable, using the numerical value in the y direction as a dependent variable, and fitting to obtain a function;
(2) and according to the function, the current point is stored as a throwing track until the difference between the numerical values of the two continuous points in the y direction is less than a threshold value A.
Compared with the prior art, the invention has the following advantages:
(1) the high-altitude parabolic detection system based on computer vision fully utilizes computer vision technology, detects the actual situation of high-altitude parabolic generation, strengthens the security management of a cell, improves the cell environment, has better detection effect on a moving object of a small target, can adapt to the situation that the target is temporarily shielded or missed, and the like, can adapt to the situation that the background environment slowly changes, can remove part of non-parabolic moving targets in the tracking process, and has wide application range, strong universality and good detection effect.
(2) The high-altitude parabolic detection system based on computer vision divides the system in a modularized way, links all modules by using buffer areas, fully reduces the coupling among all modules, can reduce the loss of data and improves the stability and robustness of the system.
(3) The high-altitude parabolic detection method based on computer vision detects small target moving objects in real time through the obtained videos of the monitoring scene, then tracks the moving targets, can remove a part of non-parabolic moving targets, judges whether the detected moving targets are high-altitude parabolas or not by performing curve fitting on the moving tracks of the moving targets and combining high-altitude parabolic movement rules.
Drawings
FIG. 1 is a diagram of a design architecture of a computer-vision high altitude parabolic detection system in accordance with an embodiment of the present invention;
FIG. 2 is a data buffering diagram of a producer-consumer synchronous control model according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of the operational data flow of a computer-vision high altitude parabolic detection method according to an embodiment of the present invention;
FIG. 4 is a detailed flow diagram of a ViBe background modeling moving object detection method based on OpenCL according to an embodiment of the present invention;
FIG. 5 is a two-dimensional ViBe model matching diagram of an embodiment of the invention;
fig. 6 is a detailed flow diagram of a kalman filtering moving target tracking method based on high-altitude parabolic motion prior knowledge according to an embodiment of the present invention.
Detailed Description
The invention is described in further detail below with reference to the figures and the embodiments.
Referring to fig. 1, an embodiment of the present invention provides a high altitude parabolic detection system based on computer vision, including:
the video image acquisition unit is used for comprehensively monitoring the cell in real time and acquiring image data;
a network transmission unit for transmitting image data;
the moving target detection unit is used for receiving the image data from the network transmission unit, analyzing and processing the acquired image data and determining a moving target in the image data;
and the central processing unit is used for tracking the moving target, acquiring the characteristics and the moving track of the moving target, judging whether the moving target belongs to a high-altitude parabolic object or not and acquiring the throwing position of the high-altitude parabolic object.
The high-altitude parabolic detection system based on computer vision fully utilizes computer vision technology, detects the actual situation of high-altitude parabolic generation, strengthens the security management of a cell, improves the cell environment, has better detection effect on a moving object of a small target, can adapt to the situation that the target is temporarily shielded or missed, and the like, can adapt to the situation that the background environment slowly changes, can remove part of non-parabolic moving targets in the tracking process, and has wide application range, strong universality and good detection effect.
On the basis of the technical solution of the above embodiment, further, the video image acquisition unit includes a plurality of network cameras and a data stream processing unit, the plurality of network cameras are installed at different positions in the cell, and the data stream processing unit is configured to acquire video data streams of the network cameras and decode the video data streams into image data by using an H265 decoding technology; the arrangement and installation of the monitoring cameras utilize the network cameras to be installed below the monitoring building, the cameras are installed at a position 30 meters away from the building, the installation height is 3 meters, under the condition, the proper number of the cameras is utilized, each camera is installed according to a proper installation angle, the proportion of the actual size of an object to the size of an article in a monitoring picture is ensured to be imaged, namely if the actual size of the object is N1N 2(cm), the size of the object shot in the video is N1N 2 (pixels), and the network transmission unit is used for connecting and transmitting monitoring data; acquiring a video data stream, namely acquiring a compressed video data stream through an IP address, a port number, a user name and a password of a camera; and (5) decoding the data stream, and decoding the obtained video data stream into pictures of one frame by using an H265 decoding technology.
On the basis of the technical solution of the above embodiment, further, the computer vision-based high-altitude parabolic detection system further includes a video storage unit, where the video storage unit is configured to cache video stream data of a motion trajectory determined by the central processing unit to belong to a high-altitude parabolic target; the scene of the high-altitude object throwing can be reproduced, and the post-investigation is convenient.
On the basis of the technical solution of the above embodiment, further, the video storage unit compresses the image data belonging to the high-altitude parabolic target into video stream data by an H265 compression method; when the high-altitude parabolic object appears, compressing the monitoring picture cached by the system for a period of time from the picture to the video through an H265 compression method to store the high-altitude parabolic key video, and searching and playing back the monitoring picture according to the time, the position and other information of the parabolic object, so as to determine the starting point of the high-altitude parabolic object.
On the basis of the technical scheme of the embodiment, further, data communication is performed among the video image acquisition unit, the network transmission unit, the moving object detection unit and the central processing unit by using a buffer area; each module of the system is used as an independent sub-thread, the middle of each module is connected through a buffer queue, the buffer queue is used for thread synchronous control, and a control mode adopts a producer-consumer model. For example, the synchronous control between a video image acquisition unit and a moving target detection unit sends an image acquisition signal to a system, an image acquisition thread is awakened, a monitoring video stream is obtained and decoded, an acquired video original picture is hung at the tail of a buffer area queue, and the related semaphore of a critical area is changed; when the image data exists in the buffer area, the moving object detection thread is awakened, and the picture is taken out from the head of the queue for further processing. This mode allows for short time tolerances between the various modules of the system. But this time is also limited and the setting of the size of the buffer determines the size of the tolerance time. If the processing speed of a certain thread of the system is too slow, the data of the related buffer area is accumulated more, when the processing speed is close to the total size of the buffer area, the system suspends the thread of the previous module and feeds back a pipeline speed-reducing instruction to prevent the buffer area from overflowing and the system from crashing. By the scheme, the coupling between the modules is fully reduced, the loss of data can be reduced, and the stability and the robustness of the system are improved.
Referring to fig. 2 to 6, an embodiment of the present invention further provides a high altitude parabolic detection method based on computer vision, including the following steps:
1) collecting an image;
the method comprises the steps that a comprehensive real-time monitoring cell obtains parabolic video stream data, and the video stream data are decoded into image data;
2) detecting a moving target;
analyzing and processing the image data, and determining a moving target in the image data;
3) tracking a moving target;
tracking each moving target, and acquiring the characteristics and the moving track of the moving target;
4) judging high altitude parabolic;
and analyzing and judging whether the moving target belongs to a high-altitude parabolic object or not, and acquiring a throwing position of the high-altitude parabolic object.
The high-altitude parabolic detection method based on computer vision detects small target moving objects in real time through the obtained videos of the monitoring scene, then tracks the moving targets, can remove a part of non-parabolic moving targets, judges whether the detected moving targets are high-altitude parabolas or not by performing curve fitting on the moving tracks of the moving targets and combining high-altitude parabolic movement rules.
On the basis of the technical solution of the above embodiment, further, in step 2), the moving object detecting step is:
(1) storing a sample set for each pixel in a first frame of image data, wherein pixel values of adjacent points around the pixel are sampling values of the sample set;
(2) comparing the pixel value of the pixel point of each subsequent frame with the sampling value of the sample set, and if the difference value is smaller than a threshold value, considering that the two points are close;
(3) sequentially traversing all sample sets to obtain the number of the pixel points close to the sample sets, if the number is larger than a threshold value, considering the pixel point as a background point, and otherwise, considering the pixel point as a foreground point;
(4) and screening the obtained foreground points, screening out objects which do not meet the requirements according to the perimeter L and the area S of the objects, carrying out corrosion, expansion and connectivity analysis on the detection result to remove detection noise, finally obtaining a moving target with 7 × 7 pixels at minimum, and extracting the characteristics of the moving target.
The moving target detection is implemented by carrying out graying and median filtering preprocessing on an original picture, and then carrying out corrosion, expansion and connectivity analysis on a detection result by using a moving target detection method based on ViBe background modeling of OpenCL to remove detection noise, so that a moving target with 7 × 7 pixels at minimum can be finally obtained, and the characteristics of the moving target are extracted. The flow chart is shown in fig. 3.
The implementation of the ViBe algorithm is mainly divided into three modules: initializing a background model, matching the model and updating the background model.
First, the background model is initialized using only the first picture or several pictures of the input video sequence. And randomly selecting the pixel value of the point or the pixel value of the neighborhood as a model for each pixel point in the picture. For example, at (X, y) pixel point X (X, y) in the image, N is located in its 8 neighborhood by the following method8And (X) randomly selecting N samples as model initialization parameters.
M(X)＝{vi(y)|y∈N8(X)},i＝1,2,...,N
Then, model matching is performed on the input image, and a moving object is detected by calculating the similarity between the pixel value of the input image and the background model. Suppose the current input frame is XtThe pixel value at the pixel point (X, y) is Xt(x, y). As shown in fig. 4, define a formula XtSphere S with (x, y) as center and R as radiusR(Xt(x,y))，SR(Xt(X, y)) means all of and Xt(x, y) set of points at distances less than R, falling at S by M (X)R(Xt(X, y)) the number m of samples in the region represents Xt(x, y) similarity to the background model M (X), for a given threshold Tm: if m is<TmThen, X is judgedt(x, y) is foreground; if m.gtoreq.TmThen determine Xt(x, y) is background. As shown in the following formula, dist (X)t(x,y),Mi(X)) represents Xt(x, y) and Mi(X).
Finally, in order to adapt to the background change of the model in the monitoring scene, so that the model can better describe the background of the current moment, the model needs to be updated. The ViBe algorithm is based on the current frame pixel value XtTo update the background model M (X) at time t-1t-1) Each background point has the same probability 1/phi to update the probability of the model, and the pixel value space propagation characteristic is utilized to update the model for the neighborhood points with the same probability 1/phi. If the probability that a sample in the model at time t is not updated in a continuous period of time is (N-1)/N, the probability that the sample is not changed after dt times is calculated as follows:
the ViBe moving object detection is mainly the calculation operation of the pixels of the image, and in order to enable the system to have a faster response speed, the parallel improvement of the image processing ViBe algorithm can be carried out. The OpenCL programming method is used for carrying out parallelization processing on the algorithm, the powerful parallel computing capacity of the GPU is fully utilized and is cooperated with the CPU, hardware is used more efficiently to complete large-scale and high-parallelism computing in the ViBe algorithm, and the system acceleration function is achieved. OpenCL programming utilizes the characteristics of a GPU to perform multi-thread parallel processing to realize a ViBe algorithm, and the algorithm after the parallel processing has obvious advantages in processing high-resolution images.
On the basis of the technical solution of the above embodiment, further, in step 3), the moving target tracking step is:
(1) acquiring a central point of the moving object and a sectional drawing of the moving object in the step 2);
(2) initializing the environment of a Kalman filtering algorithm in a first frame, and defaulting all moving targets as targets to be tracked;
(3) then, each moving object of each frame is matched with the existing track through a Hungarian algorithm, and if the matching is successful, the moving objects are added into the corresponding track; if the matching fails, predicting the position of the track in the current frame through a Kalman filtering algorithm;
(4) if a certain track is not successfully matched with the moving object for a plurality of times, namely the number of track points obtained by continuous prediction through a Kalman filtering algorithm is greater than a specific value A, the track is considered to be finished;
(5) if the number of points of a certain track is larger than the threshold value B, the track is considered to be effective, otherwise, the track is considered to be ineffective.
The moving target tracking predicts the position of the moving target in the next frame by the extracted characteristics of the detected moving target and by using a Kalman filtering moving target tracking method based on high-altitude parabolic motion prior knowledge, determines a target matching area, and finally realizes the matching of the moving target by a Hungarian matching method, wherein the flow schematic diagram is shown in FIG. 5.
The Kalman filter basically comprises three steps of: firstly, initializing a Kalman filter, assigning the position and the speed of a target to the Kalman filter, and recording the time of a current frame; then, state estimation is carried out, Kalman prediction is carried out by utilizing the time interval between the current frame and the previous frame, the position and the motion state of the target of the current frame are predicted, the optimal matching in the area is searched, the position and the motion state value of the real target are obtained, the current time is recorded at the same time, the priori knowledge of high-altitude parabolic motion is innovatively added, and if the track continuous non-falling time t occursnumT is greater than a threshold value T, i.e. Tnum>T, the moving target is represented as non-parabolic motion, and the moving target is directly removed as represented by the following formula; and finally, updating the state, wherein the state of the Kalman filter is updated by using the real state vector as the input of the Kalman filter. The kalman filter formula is mainly five, as follows:
kalman equation of prediction
The state equation is as follows:
covariance prediction equation:
kalman correction equation
Filter gain of kalman:
modifying the state vector:
modifying the covariance matrix:
the Hungarian algorithm is a process of searching an amplification path of a bipartite graph formed by points to be matched to obtain maximum matching, and the process is mainly divided into three steps.
(1) A matching graph M of the bipartite graph G is empty;
(2) finding out an augmentation path P of the bipartite graph G, and obtaining a matching graph M' of a larger bipartite graph to replace M through the exclusive OR operation of the augmentation path P and the original bipartite graph G;
(3) and (5) repeating the step (2) and continuously searching the augmentation path. And (5) stopping the step (2) when all the augmented paths are searched, and completing the matching.
On the basis of the technical solution of the above embodiment, further, in step 4), the conditions for determining the high altitude parabola are as follows:
(1) the value of the ordinate of the moving target trajectory is continuously increasing;
(2) the number M of the high-altitude parabolic motion target track points is greater than a preset threshold value T, namely M is greater than T;
(3) the high-altitude parabolic motion target track is ended from the boundary of the monitoring range;
(4) the parabolic motion makes an accelerating motion at the beginning part;
(5) in the falling process of the high-altitude parabolic motion target, the difference value of the ordinate is not always too small.
And the high-altitude parabolic motion judgment provides a high-altitude parabolic motion judgment method based on curve fitting through the high-altitude parabolic motion trajectory characteristics, the high-altitude parabolic motion is downward and accelerated, the parabolic motion exists in a video for a period of time, the parabolic motion is terminated from the monitoring range boundary, whether a moving target is the high-altitude parabolic motion is determined, and the high-altitude parabolic motion realizes parabolic motion position determination through curve fitting. The judgment accuracy is determined by the judgment condition of the high-altitude parabolic object, and the following parabolic features are mainly utilized as the main conditions for judging the parabolic object according to the motion characteristics of the high-altitude parabolic object:
(1) the motion track of the high-altitude parabola always moves downwards, so that the value of the ordinate of the motion target track is continuously increased;
(2) the number M of the high-altitude parabolic motion target track points is required to meet a set threshold value T, namely M is greater than T, the motion target moves in a monitoring range, the motion of the motion target can move for a certain time, and therefore the number of the detected track points is limited;
(3) the high-altitude parabolic motion target track is ended from the boundary of the monitoring range, parabolic motion cannot stay in a certain place, the monitoring range can be out of the certain boundary, and meanwhile, the boundary area cannot be an upper boundary area and only can be from a left boundary area, a right boundary area and a lower boundary area;
(4) the parabolic motion basically performs accelerated motion at the beginning part, but is influenced by the environment and the like, so that the whole parabolic motion inevitably generates partial points of a detected target without deceleration and uniform speed, and the proportional threshold is set to remove the moving target which is always in uniform speed or deceleration;
(5) in the falling process of the high-altitude parabolic motion target, the difference value of the vertical coordinate is not too small all the time, so that the noise of almost coincident vertical coordinates can be removed.
On the basis of the technical solution of the above embodiment, further, in step 4), the step of obtaining the throwing position belonging to the high-altitude parabolic target includes:
(1) for each track, fitting all points on the track by using a least square method, using the frame number as an independent variable, using the numerical value in the y direction as a dependent variable, and fitting to obtain a function;
(2) and according to the function, the current point is stored as a throwing track until the difference between the numerical values of the two continuous points in the y direction is less than a threshold value A.
Therefore, the throwing point of the high-altitude parabolic object can be tracked, the person responsible for the accident can be found out, the evidence can be taken after the accident and the legal responsibility of related personnel can be traced, the reoccurrence of the high-altitude parabolic event can be prevented, the life safety of people is guaranteed, the legal right of people is maintained, and the stability of social security is improved.
Various modifications and variations of the embodiments of the present invention may be made by those skilled in the art, and they are also within the scope of the present invention, provided they are within the scope of the claims of the present invention and their equivalents.
What is not described in detail in the specification is prior art that is well known to those skilled in the art.
Claims (7)
1. A high-altitude parabolic detection method based on computer vision is characterized by comprising the following steps:
1) collecting an image;
the method comprises the steps that a comprehensive real-time monitoring cell obtains parabolic video stream data, and the video stream data are decoded into image data;
2) detecting a moving target;
analyzing and processing the image data, and determining a moving target in the image data;
3) tracking a moving target;
tracking each moving target, and acquiring the characteristics and the moving track of the moving target;
4) judging high altitude parabolic;
analyzing and judging whether the moving target belongs to a high-altitude parabolic object or not, and acquiring a throwing position of the high-altitude parabolic object;
in the step 2), the moving target detection step is as follows:
(1) storing a sample set for each pixel in a first frame of image data, wherein pixel values of adjacent points around the pixel are sampling values of the sample set;
(2) comparing the pixel value of the pixel point of each subsequent frame with the sampling value of the sample set, and if the difference value between the pixel value of each subsequent frame and the sampling value of the sample set is less than a threshold value, determining that the two points are close;
(3) sequentially traversing all sample sets to obtain the number of the pixel points close to the sample sets, if the number is larger than a threshold value, considering the pixel point as a background point, and otherwise, considering the pixel point as a foreground point;
(4) screening the obtained foreground points, screening out objects which do not meet the requirements according to the perimeter L and the area S of the objects, carrying out corrosion, expansion and connectivity analysis on the detection result to remove detection noise, finally obtaining a moving target with 7 × 7 pixels at minimum, and extracting the characteristics of the moving target;
the moving target detection step is realized by using a Vibe algorithm based on an OpenCL programming method;
in step 3), the moving target tracking step is as follows:
(1) acquiring a central point of the moving object and a sectional drawing of the moving object in the step 2);
(2) initializing the environment of a Kalman filtering algorithm in a first frame, and defaulting all moving targets as targets to be tracked;
(3) then, each moving object of each frame is matched with the existing track through a Hungarian algorithm, and if the matching is successful, the moving objects are added into the corresponding track; if the matching fails, predicting the position of the track in the current frame through a Kalman filtering algorithm;
(4) if a certain track is not successfully matched with the moving object for a plurality of times, namely the number of track points obtained by continuous prediction through a Kalman filtering algorithm is greater than a specific value A, the track is considered to be finished;
(5) if the number of points of a certain track is greater than a threshold value B, the track is considered to be effective, otherwise, the track is considered to be invalid;
in the step 4), the step of obtaining the throwing position of the high-altitude parabolic target comprises the following steps:
(1) for each track, fitting all points on the track by using a least square method, using the frame number as an independent variable, using the numerical value in the y direction as a dependent variable, and fitting to obtain a function;
(2) and according to the function, the current point is stored as the track throwing until the difference between the numerical values of the two continuous points in the y direction is smaller than a threshold value A.
2. The high altitude parabolic detection method based on computer vision as claimed in claim 1, characterized in that: in step 4), the conditions for determining the high altitude parabola are as follows:
(1) the value of the ordinate of the moving target trajectory is continuously increasing;
(2) the number M of the high-altitude parabolic motion target track points is greater than a preset threshold value T, namely M is greater than T;
(3) the high-altitude parabolic motion target track is ended from the boundary of the monitoring range;
(4) the parabolic motion makes an accelerating motion at the beginning part;
(5) in the falling process of the high-altitude parabolic motion target, the difference value of the ordinate is not always too small.
3. A computer vision-based high altitude parabolic detection system for implementing the computer vision-based high altitude parabolic detection method of claim 1, comprising:
the video image acquisition unit is used for comprehensively monitoring the cell in real time and acquiring image data;
a network transmission unit for transmitting image data;
the moving target detection unit is used for receiving the image data from the network transmission unit, analyzing and processing the acquired image data and determining a moving target in the image data;
and the central processing unit is used for tracking the moving target, acquiring the characteristics and the moving track of the moving target, judging whether the moving target belongs to a high-altitude parabolic object or not and acquiring the throwing position of the high-altitude parabolic object.
4. A high altitude parabolic detection system based on computer vision as claimed in claim 3 wherein: the video image acquisition unit comprises a plurality of network cameras and a data stream processing unit, the network cameras are arranged at different positions in a cell, and the data stream processing unit is used for acquiring video data streams of the network cameras and decoding the video data streams into image data by adopting an H265 decoding technology.
5. A high altitude parabolic detection system based on computer vision as claimed in claim 4 wherein: the high-altitude parabolic detection system based on computer vision further comprises a video storage unit, wherein the video storage unit is used for caching the video stream data of the motion trail judged to belong to the high-altitude parabolic target by the central processing unit.
6. A high altitude parabolic detection system based on computer vision as claimed in claim 5 wherein: the video storage unit compresses image data belonging to the high-altitude parabolic target into video stream data by an H265 compression method.
7. A high altitude parabolic detection system based on computer vision as claimed in claim 3 wherein: and the video image acquisition unit, the network transmission unit, the moving target detection unit and the central processing unit carry out data communication by utilizing a buffer area.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811015244.1A CN109309811B (en) | 2018-08-31 | 2018-08-31 | High-altitude parabolic detection system and method based on computer vision |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811015244.1A CN109309811B (en) | 2018-08-31 | 2018-08-31 | High-altitude parabolic detection system and method based on computer vision |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109309811A CN109309811A (en) | 2019-02-05 |
| CN109309811B true CN109309811B (en) | 2021-02-19 |
Family
ID=65224317
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811015244.1A Active CN109309811B (en) | 2018-08-31 | 2018-08-31 | High-altitude parabolic detection system and method based on computer vision |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109309811B (en) |
Families Citing this family (36)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111753574A (en) * | 2019-03-26 | 2020-10-09 | 顺丰科技有限公司 | Throw area positioning method, device, equipment and storage medium |
| CN110275042B (en) * | 2019-05-07 | 2021-07-13 | 深圳市零壹移动互联系统有限公司 | High-altitude parabolic detection method based on computer vision and radio signal analysis |
| CN110738686B (en) * | 2019-10-12 | 2022-12-02 | 四川航天神坤科技有限公司 | Static and dynamic combined video man-vehicle detection method and system |
| CN111079663B (en) * | 2019-12-19 | 2022-01-11 | 深圳云天励飞技术股份有限公司 | High-altitude parabolic monitoring method and device, electronic equipment and storage medium |
| CN111260693B (en) * | 2020-01-20 | 2023-07-28 | 北京中科晶上科技股份有限公司 | High-altitude parabolic detection method |
| CN111382692A (en) * | 2020-03-05 | 2020-07-07 | 深学科技(杭州)有限公司 | High altitude parabolic detection method, device and computer readable storage medium |
| CN111488799B (en) * | 2020-03-13 | 2020-12-15 | 安徽小眯当家信息技术有限公司 | Falling object identification method and system based on image identification |
| CN111523397B (en) * | 2020-03-31 | 2023-11-24 | 深圳市奥拓电子股份有限公司 | Intelligent lamp post visual identification device, method and system and electronic equipment thereof |
| CN111476163B (en) * | 2020-04-07 | 2022-02-18 | 浙江大华技术股份有限公司 | High-altitude parabolic monitoring method and device and computer storage medium |
| CN111401311A (en) * | 2020-04-09 | 2020-07-10 | 苏州海赛人工智能有限公司 | High-altitude parabolic recognition method based on image detection |
| CN111553257A (en) * | 2020-04-26 | 2020-08-18 | 上海天诚比集科技有限公司 | High-altitude parabolic early warning method |
| CN111553256A (en) * | 2020-04-26 | 2020-08-18 | 上海天诚比集科技有限公司 | High-altitude parabolic early warning identification method based on object track identification |
| CN111553274A (en) * | 2020-04-28 | 2020-08-18 | 青岛聚好联科技有限公司 | High-altitude parabolic detection method and device based on trajectory analysis |
| CN111654664B (en) * | 2020-05-08 | 2022-04-08 | 浙江大华技术股份有限公司 | High-altitude parabolic detection method and system, computer equipment and storage medium |
| CN111723654B (en) * | 2020-05-12 | 2023-04-07 | 中国电子系统技术有限公司 | High-altitude parabolic detection method and device based on background modeling, YOLOv3 and self-optimization |
| CN111627049B (en) * | 2020-05-29 | 2023-09-26 | 北京中科晶上科技股份有限公司 | Method and device for determining high-altitude parabolic object, storage medium and processor |
| CN113808409B (en) * | 2020-06-17 | 2023-02-10 | 华为技术有限公司 | Road safety monitoring method, system and computer equipment |
| CN111539388B (en) * | 2020-06-22 | 2020-10-09 | 湖南联智科技股份有限公司 | High-altitude parabolic monitoring method |
| CN111931599B (en) * | 2020-07-20 | 2023-04-18 | 浙江大华技术股份有限公司 | High altitude parabolic detection method, equipment and storage medium |
| CN111898511A (en) * | 2020-07-23 | 2020-11-06 | 北京以萨技术股份有限公司 | High-altitude parabolic detection method, device and medium based on deep learning |
| CN112016414A (en) * | 2020-08-14 | 2020-12-01 | 熵康(深圳)科技有限公司 | Method and device for detecting high-altitude parabolic event and intelligent floor monitoring system |
| CN112288991A (en) * | 2020-09-08 | 2021-01-29 | 江苏澳联科技开发有限公司 | Intelligent alarm system for falling object |
| CN112213716B (en) * | 2020-09-21 | 2024-07-16 | 浙江大华技术股份有限公司 | Object positioning method and device and electronic equipment |
| CN112258573B (en) * | 2020-10-16 | 2022-11-15 | 腾讯科技（深圳）有限公司 | Method and device for acquiring throwing position, computer equipment and storage medium |
| CN112347915B (en) * | 2020-11-05 | 2024-03-12 | 重庆览辉信息技术有限公司 | High-altitude throwing object distinguishing system |
| CN112365524B (en) * | 2020-11-10 | 2024-03-29 | 宁波博登智能科技有限公司 | High-altitude parabolic real-time alarm system based on time sequence image |
| CN112418069B (en) * | 2020-11-19 | 2021-05-11 | 中科智云科技有限公司 | High-altitude parabolic detection method and device, computer equipment and storage medium |
| CN112488024B (en) * | 2020-12-08 | 2024-08-23 | 平安国际智慧城市科技股份有限公司 | High-altitude parabolic tracing method and device, computer equipment and storage medium |
| CN112489069A (en) * | 2020-12-12 | 2021-03-12 | 江西洪都航空工业集团有限责任公司 | Vehicle window parabolic detection method based on real-time video |
| CN112733690B (en) * | 2020-12-31 | 2024-02-20 | 北京易华录信息技术股份有限公司 | High-altitude parabolic detection method and device and electronic equipment |
| CN113052878B (en) * | 2021-03-30 | 2024-01-02 | 北京中科通量科技有限公司 | Multipath high-altitude parabolic detection method and system for edge equipment in security system |
| CN113269046B (en) * | 2021-04-28 | 2022-04-22 | 深圳市海清视讯科技有限公司 | High-altitude falling object identification method and system |
| CN113255697B (en) * | 2021-06-01 | 2024-07-26 | 南京图菱视频科技有限公司 | High-precision high-altitude parabolic detection system and method in complex scene |
| CN113409360A (en) * | 2021-06-29 | 2021-09-17 | 深圳市商汤科技有限公司 | High altitude parabolic detection method and device, equipment and computer storage medium |
| CN115086531A (en) * | 2022-07-08 | 2022-09-20 | 浙江上视科技有限公司 | Intelligent data intelligent visual platform for smart park |
| CN115937267B (en) * | 2023-03-03 | 2023-10-24 | 北京灵赋生物科技有限公司 | Target track tracking method based on multi-frame video |
-
2018
- 2018-08-31 CN CN201811015244.1A patent/CN109309811B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN109309811A (en) | 2019-02-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109309811B (en) | High-altitude parabolic detection system and method based on computer vision | |
| CN109872341B (en) | High-altitude parabolic detection method and system based on computer vision | |
| CN108629299B (en) | Long-time multi-target tracking method and system combining face matching | |
| Hou et al. | Detection of the mobile object with camouflage color under dynamic background based on optical flow | |
| CN111260693B (en) | High-altitude parabolic detection method | |
| US7982774B2 (en) | Image processing apparatus and image processing method | |
| CN111401311A (en) | High-altitude parabolic recognition method based on image detection | |
| CN109101944B (en) | Real-time video monitoring method for identifying garbage thrown into river channel | |
| CN106571014A (en) | Method for identifying abnormal motion in video and system thereof | |
| CN109711318B (en) | Multi-face detection and tracking method based on video stream | |
| US20120120238A1 (en) | Two layer video motion detection | |
| CN102034240A (en) | Method for detecting and tracking static foreground | |
| US20200145623A1 (en) | Method and System for Initiating a Video Stream | |
| CN111898511A (en) | High-altitude parabolic detection method, device and medium based on deep learning | |
| WO2021139049A1 (en) | Detection method, detection apparatus, monitoring device, and computer readable storage medium | |
| AU2015203666A1 (en) | Methods and systems for controlling a camera to perform a task | |
| US20160210759A1 (en) | System and method of detecting moving objects | |
| CN113537107A (en) | Face recognition and tracking method, device and equipment based on deep learning | |
| CN109615641B (en) | Multi-target pedestrian tracking system and tracking method based on KCF algorithm | |
| CN103187083A (en) | Storage method and system based on time domain video fusion | |
| Tao et al. | Real-time detection and tracking of moving object | |
| CN111784750A (en) | Method, device and equipment for tracking moving object in video image and storage medium | |
| Schreiber et al. | GPU-based non-parametric background subtraction for a practical surveillance system | |
| CN116630365A (en) | Parabolic detection method and device, electronic equipment and storage medium | |
| WO2023042337A1 (en) | Image processing system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |