Perimeter intrusion identification method and system, electronic equipment and storage medium

Abstract

The invention provides a perimeter intrusion identification method and system, electronic equipment and a storage medium, and relates to the technical field of monitoring image processing. The perimeter intrusion identification method comprises the following steps: carrying out image preprocessing on a target boundary identification area and an extended identification area to obtain a target processing image; performing image detection and recognition on the target processing image, and performing image segmentation on the target processing image under the condition that the target body does not exist in the target processing image so as to obtain at least two sub-target processing images; performing image amplification processing on the at least two sub-target processing images to obtain at least two amplified sub-target processing images; and carrying out image detection and identification on the at least two amplified sub-target processing images again, and determining that a perimeter intrusion behavior exists under the condition that a target body exists in at least one amplified sub-target processing image. The peripheral area can be covered by arranging a single monitoring device, and the method has the advantages of being low in cost and high in recognition precision.

Description

Technical Field

[0001] The present invention relates to the technical field of monitoring image processing, and in particular to a perimeter intrusion identification method and system, electronic equipment and storage medium thereof. Background Art

[0002] Perimeter intrusion identification refers to the use of various technical means to detect and identify unauthorized persons or objects entering a specific area. Perimeter intrusion identification is an important measure for patrolling the perimeter security of key areas. It can effectively detect intrusions by people, objects or vehicles, thereby protecting the security of key areas.

[0003] At present, perimeter intrusion identification generally adopts infrared electronic fences, inspection robots or surveillance cameras. However, the inventors found that the infrared electronic fence identification method will also issue an alarm prompt when non-human living things (such as animals) pass through the fence area, resulting in false alarms and poor identification accuracy; inspection robots need to be regularly inspected and charged, etc., and the maintenance cost is high; the surveillance camera method requires multiple cameras to be deployed at the same time to cover the entire area, resulting in high equipment costs. Summary of the invention

[0004] The present invention provides a perimeter intrusion identification method and system, electronic equipment and storage medium, which are used to solve the problems of high cost and poor identification accuracy existing in the current perimeter intrusion identification method.

[0005] One aspect of the present invention provides a perimeter intrusion identification method, including: acquiring target image information collected by a monitoring device; setting a target boundary identification area according to the target image information; obtaining an extended identification area according to the target boundary identification area; performing image preprocessing on the target boundary identification area and the extended identification area to obtain a target processing image; performing image detection and identification on the target processing image to determine whether there is a target body in the target processing image; if there is no target body in the target processing image, performing image segmentation on the target processing image to obtain at least two sub-target processing images; performing image magnification processing on the at least two sub-target processing images to obtain at least two enlarged sub-target processing images; performing image detection and identification again on the at least two enlarged sub-target processing images, and if there is a target body in at least one enlarged sub-target processing image, determining that there is perimeter intrusion behavior.

[0006] According to some embodiments of the present invention, the perimeter intrusion identification method also includes: when there is a target object in the target processing image, determining a target object identification area where the target object is located; performing image magnification processing on the target object identification area to obtain an enlarged target object identification area; performing image detection and identification on the enlarged target object identification area again, and when there is a target object in the enlarged target object identification area, determining that there is perimeter intrusion behavior.

[0007] According to some embodiments of the present invention, when the monitoring device is an aerial surveillance dome camera, performing image magnification processing on at least two sub-target processed images includes: calling a zoom function of the aerial surveillance dome camera to magnify the sub-target processed images.

[0008] According to some embodiments of the present invention, when the monitoring device is a monocular camera, performing image magnification processing on at least two sub-target processed images includes: calling an image adjustment function in a computer vision library to magnify the area of ​​the sub-target processed images; or performing magnification processing on the sub-target processed images according to an image super-resolution technique to magnify the area of ​​the sub-target processed images.

[0009] One aspect of the present invention provides a perimeter intrusion identification system, including an image acquisition module, a region identification module and an image processing module. The image acquisition module acquires target image information collected by a monitoring device; the region identification module sets a target boundary identification region according to the target image information, and obtains an extended identification region according to the target boundary identification region; the image processing module performs image preprocessing on the target boundary identification region and the extended identification region to obtain a target processing image, performs image detection and identification on the target processing image, determines whether there is a target body in the target processing image, and if there is no target body in the target processing image, performs image segmentation on the target processing image to obtain at least two sub-target processing images, performs image enlargement processing on the at least two sub-target processing images to obtain at least two enlarged sub-target processing images, and performs image detection and identification on the at least two enlarged sub-target processing images again, and determines that there is perimeter intrusion behavior if there is a target body in at least one enlarged sub-target processing image.

[0010] According to some embodiments of the present invention, when a target object exists in a target processing image, the image processing module determines a target body identification area where the target object is located; the image processing module performs image magnification processing on the target body identification area to obtain an enlarged target body identification area; the image processing module performs image detection and identification on the enlarged target body identification area again, and when a target object exists in the enlarged target body identification area, determines that there is perimeter intrusion.

[0011] According to some embodiments of the present invention, when the monitoring device is a high-altitude monitoring ball camera, the image processing module calls the zoom function of the high-altitude monitoring ball camera to enlarge the area of ​​the sub-target processed image.

[0012] According to some embodiments of the present invention, when the monitoring device is a monocular camera, the image processing module calls the image adjustment function in the computer vision library to enlarge the area of ​​the sub-target processed image; or enlarges the sub-target processed image according to the image super-resolution technology to enlarge the area of ​​the sub-target processed image.

[0013] Another aspect of the present invention provides an electronic device, which includes: one or more processors; and a storage device for storing one or more programs, which, when executed by the one or more processors, enables the one or more processors to implement the perimeter intrusion identification method described above.

[0014] Another aspect of the present invention provides a non-volatile computer-readable storage medium having a computer program stored thereon, which can implement the perimeter intrusion identification method as described above when executed by a processor.

[0015] The present invention can obtain a target processing image by preprocessing the images of the target boundary recognition area and the extended recognition area, and can determine whether there is a target body that invades the perimeter in the target processing image by detecting and recognizing the image of the target processing image; and the present invention can accurately recognize the target body for each sub-target processing image by further segmenting and amplifying the target processing image. The present invention can cover the perimeter area by setting a single monitoring device, and has the characteristics of low cost and high recognition accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. 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.

[0017] Figure 1 A flow chart showing a perimeter intrusion identification method according to an embodiment of the present invention;

[0018] Figure 2 Another flow chart showing the perimeter intrusion identification method according to an embodiment of the present invention;

[0019] Figure 3 A schematic diagram showing a perimeter intrusion identification system according to an embodiment of the present invention is shown.

[0020] Description of reference numerals:

[0021] Perimeter intrusion recognition system 1; image acquisition module 10; area recognition module 20; image processing module 30. DETAILED DESCRIPTION

[0022] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that the present invention will be comprehensive and complete and fully convey the concepts of the example embodiments to those skilled in the art. The same reference numerals in the figures represent the same or similar parts, and thus their repeated description will be omitted.

[0023] The described features, structures or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure may be practiced without one or more of these specific details, or other modes, components, materials, devices, etc. may be adopted. In these cases, known structures, methods, devices, implementations, materials or operations will not be shown or described in detail.

[0024] In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or devices.

[0025] The terms "first", "second" and the like in the specification, claims and the above drawings of the present invention are used to distinguish different objects rather than to describe a specific order.

[0026] The technical solution of the present invention is described clearly and completely 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0027] The full names of the English abbreviations involved in the present invention and their Chinese meanings are:

[0028] SDK: Software Development Kit;

[0029] RTSP: Real Time Streaming Protocol, real-time streaming protocol;

[0030] SSD: Single Shot MultiBox Detector, a target detection algorithm based on deep learning;

[0031] YOLO: You Only Look Once, a real-time target detection algorithm based on deep learning;

[0032] RetinaNet: a single-stage object detection model.

[0033] RT-DETR: Real-Time Detection Transformer, real-time end-to-end object detection model.

[0034] According to one aspect of the present invention, a perimeter intrusion identification method is provided for identifying whether there are illegal persons or vehicles performing perimeter intrusion in the current monitoring image based on image information collected by a monitoring device.

[0035] Figure 1 A flow chart of a perimeter intrusion identification method according to an embodiment of the present invention is shown. Figure 1 As shown, the perimeter intrusion identification method may include steps S100 to S800. Exemplarily, the perimeter intrusion identification method may be executed by a perimeter intrusion identification system with computing capability.

[0036] According to an example embodiment, in step S100 , the perimeter intrusion identification system obtains image information collected by a monitoring device.

[0037] For example, the monitoring device may be a high-altitude monitoring ball camera. The high-altitude monitoring ball camera includes an integrated camera and a pan-tilt mechanism. Based on the integrated camera and the pan-tilt mechanism, the high-altitude monitoring ball camera can achieve 360° horizontal rotation and 90° vertical tilt, can cover a wide monitoring area, and realize all-round monitoring without blind spots.

[0038] Exemplarily, the high-altitude monitoring ball camera can be installed at a preset position according to user needs. Exemplarily, the preset position can be a position with a certain monitoring height (such as a rooftop or an electric tower, etc.).

[0039] The resolution of this high-altitude surveillance ball camera can be above 1080p, and the cruise point can be set through the pan-tilt mechanism according to user needs, and the overhead viewing angle of the integrated camera can be maintained between 15° and 45° at the cruise point.

[0040] The perimeter intrusion identification system can communicate with the high-altitude monitoring ball camera, and by calling the SDK of the high-altitude monitoring ball camera and the preset video stream function interface, it can obtain the RTSP video stream from the high-altitude monitoring ball camera. The perimeter intrusion identification system performs stream decoding processing on the RTSP video and can obtain target image information including the target monitoring area (such as the inspection key area set according to user needs).

[0041] Exemplarily, the target image information may be color image data or grayscale image data.

[0042] In step S200, the perimeter intrusion recognition system sets a target boundary recognition area according to target image information.

[0043] For example, taking color image data as an example, based on color image data, the perimeter intrusion recognition system responds to user instructions and sets the target boundary recognition area according to user needs. The target boundary recognition area can be a user-defined area with monitoring and recognition significance, that is, if there is perimeter intrusion behavior in the target boundary recognition area, an alarm prompt will be issued.

[0044] The perimeter intrusion recognition system can identify the coordinates of a series of points in the target image information, and generate a polygonal area based on the coordinates of these points to form a recognition area.

[0045] Exemplarily, the target boundary recognition area may be a recognition area having boundary features such as a wall area or a fence area.

[0046] In step S300, the perimeter intrusion identification system obtains an extended identification area according to the target boundary identification area.

[0047] For example, the perimeter intrusion identification system performs frame expansion processing on the target boundary identification area based on a preset first frame expansion ratio threshold to obtain an expanded area (ie, an extended identification area) including the target boundary identification area.

[0048] Exemplarily, the first frame expansion ratio threshold may be customized according to user requirements, such as the first frame expansion ratio threshold may be [1.1, 2].

[0049] In step S400, the perimeter intrusion recognition system performs image preprocessing on the target boundary recognition area and the extended recognition area to obtain a target processed image.

[0050] For example, the perimeter intrusion identification system performs regionalized image preprocessing on the inside of the wall area or fence area and the outside of the area after the frame is expanded. For example, the target boundary identification area and the extended identification area are filled with the same pixel value to obtain the filled image (target processed image).

[0051] In this way, the present invention can fill the wall area or fence area and the area outside the expanded area with the same pixel value, so that no person or vehicle can be detected in the wall area or fence area and the area outside the expanded area. After image processing, as long as a person or vehicle is detected in the image, it indicates that a perimeter intrusion behavior is suspected, and subsequent perimeter intrusion identification is performed.

[0052] In step S500, the perimeter intrusion recognition system performs image detection and recognition on the target processing image to determine whether there is a target object in the target processing image.

[0053] For example, the target object may be a subject of behavior that needs to be identified for perimeter intrusion, such as an illegal person or vehicle that performs perimeter intrusion.

[0054] The perimeter intrusion recognition system can detect people or vehicles in the target processing image based on image segmentation recognition technology. Exemplarily, the image segmentation recognition technology includes but is not limited to SSD, YOLO, RetinaNet or RT-DETR.

[0055] If the perimeter intrusion recognition system does not detect a person or vehicle, it proceeds to step S600A (eg Figure 1 When the perimeter intrusion recognition system detects a person or a vehicle, it enters step S600B (as shown in FIG. Figure 2 As shown, the step S600B may specifically include steps S600B1-S600B3).

[0056] In step S600A, when there is no target object in the target processing image, the perimeter intrusion recognition system performs image segmentation on the target processing image to obtain at least two sub-target processing images.

[0057] For example, when no person or vehicle is detected, the perimeter intrusion recognition system performs image segmentation on the filled target processing image based on the image segmentation technology to obtain N sub-target processing images, where N≥2.

[0058] In step S700, the perimeter intrusion recognition system performs image magnification processing on at least two sub-target processed images to obtain at least two magnified sub-target processed images.

[0059] For example, the perimeter intrusion recognition system performs frame expansion area magnification processing, and magnifies each sub-target processing image, so as to obtain multiple magnified sub-target processing images.

[0060] In step S800, the perimeter intrusion recognition system performs image detection and recognition again on at least two enlarged sub-target processed images, and determines that there is perimeter intrusion behavior when a target object exists in at least one enlarged sub-target processed image.

[0061] For example, the perimeter intrusion recognition system detects people or vehicles in each enlarged sub-target processing image based on image segmentation recognition technology. As long as a person or vehicle is detected in an enlarged sub-target processing image, the perimeter intrusion recognition system determines that there is a perimeter intrusion behavior and issues an alarm prompt.

[0062] Through the above embodiments, the present invention can obtain a target processing image by preprocessing the images of the target boundary recognition area and the extended recognition area, and can determine whether there is a target body that invades the perimeter in the target processing image by detecting and recognizing the image of the target processing image; and the present invention can accurately recognize the target body for each sub-target processing image by further segmenting and amplifying the target processing image. The present invention can cover the perimeter area by setting a single monitoring device, and has the characteristics of low cost and high recognition accuracy.

[0063] Figure 2 Another flow chart of the perimeter intrusion identification method according to the embodiment of the present invention is shown.

[0064] Alternatively, if Figure 2 As shown, step S600B may also include steps S600B1-S600B3.

[0065] In step S600B1, when there is a target object in the target processing image, the perimeter intrusion recognition system determines a target object recognition area where the target object is located.

[0066] For example, when the perimeter intrusion recognition system detects a person or a vehicle, the perimeter intrusion recognition system can generate a polygonal area based on the coordinates of a series of points in the target image information to form an identification area. For example, the perimeter intrusion recognition system can generate a person identification area or a vehicle identification area.

[0067] In step S600B2, the perimeter intrusion recognition system performs image magnification processing on the target body recognition area to obtain an enlarged target body recognition area.

[0068] In step S600B3, the perimeter intrusion recognition system performs image detection and recognition on the enlarged target body recognition area again, and when a target body exists in the enlarged target body recognition area, it is determined that there is perimeter intrusion behavior.

[0069] For example, the perimeter intrusion recognition system performs a magnification process on the expanded frame area and magnifies the target body recognition area to obtain an enlarged target body recognition area. The perimeter intrusion recognition system again detects people or vehicles in the enlarged target body recognition area based on image segmentation and recognition technology. When a person or vehicle is detected in the enlarged target body recognition area, the perimeter intrusion recognition system determines that there is a perimeter intrusion behavior and issues an alarm prompt.

[0070] With such arrangement, the present invention can avoid false detection of persons or vehicles detected for the first time by enlarging the expanded frame area, and can improve the recognition accuracy of the present invention by re-enlarging the detection and performing a secondary review.

[0071] Optionally, in step S700, when the monitoring device is an aerial surveillance dome camera, the perimeter intrusion identification system calls a zoom function of the aerial surveillance dome camera to enlarge the area of ​​the sub-target processed image.

[0072] For example, when the monitoring device is a high-altitude monitoring dome camera, the perimeter intrusion identification system calls the SDK of the high-altitude monitoring dome camera and the preset video stream function interface to obtain the 3D positioning function (scaling function) of the high-altitude monitoring dome camera, and by adjusting the 3D positioning function, it can achieve the enlargement of the passed target monitoring area, thereby achieving the enlargement of the sub-target processing image.

[0073] With such arrangement, the present invention can directly call the positioning function of the high-altitude monitoring ball camera to realize the magnification processing of the sub-target processing image. In addition, during the magnification processing, the high-altitude monitoring ball camera can be controlled to automatically zoom, which can ensure the imaging quality of the monitoring picture.

[0074] Optionally, in step S700, when the monitoring device is a monocular camera, the perimeter intrusion recognition system calls an image adjustment function in a computer vision library to magnify the sub-target processed image. Alternatively, the perimeter intrusion recognition system magnifies the sub-target processed image according to an image super-resolution technique to magnify the sub-target processed image.

[0075] For example, the perimeter intrusion recognition system can enlarge the sub-target processing image by calling the resize function (a function used to adjust the image size) in the OpenCV library (computer vision library); for another example, the perimeter intrusion recognition system can also enlarge the sub-target processing image through image super-resolution technology, thereby enlarging the sub-target processing image.

[0076] With such a configuration, the present invention can achieve accurate recognition of small target detection such as sub-target processed images by adjusting the image adjustment function or enlarging the sub-target processed images through image super-resolution technology.

[0077] It can be understood here that in step S600B2, the enlargement processing method for the image enlargement processing of the target body recognition area can be the same as the enlargement processing method for the image enlargement processing of at least two sub-target processing images. Both can adopt enlargement methods such as calling the zoom function of the high-altitude monitoring ball camera, calling the image adjustment function in the computer vision library, or enlarging the image according to the image super-resolution technology. They have been described in detail above and will not be repeated here.

[0078] According to one aspect of the present invention, a perimeter intrusion identification system is provided for identifying whether there are illegal persons or vehicles performing perimeter intrusion in the current monitoring image based on image information collected by monitoring equipment.

[0079] Figure 3 FIG. 1 is a schematic diagram showing a perimeter intrusion identification system according to an embodiment of the present invention. Figure 3 As shown, the perimeter intrusion identification system 1 may include an image acquisition module 10 , a region identification module 20 and an image processing module 30 .

[0080] According to an exemplary embodiment, the image acquisition module 10 acquires image information collected by a monitoring device.

[0081] For example, the monitoring device may be a high-altitude monitoring ball camera. The high-altitude monitoring ball camera includes an integrated camera and a pan-tilt mechanism. Based on the integrated camera and the pan-tilt mechanism, the high-altitude monitoring ball camera can achieve 360° horizontal rotation and 90° vertical tilt, can cover a wide monitoring area, and realize all-round monitoring without blind spots.

[0082] Exemplarily, the high-altitude monitoring ball camera can be installed at a preset position according to user needs. Exemplarily, the preset position can be a position with a certain monitoring height (such as a rooftop or an electric tower, etc.).

[0083] The resolution of this high-altitude surveillance ball camera can be above 1080p, and the cruise point can be set through the pan-tilt mechanism according to user needs, and the overhead viewing angle of the integrated camera can be maintained between 15° and 45° at the cruise point.

[0084] The image acquisition module 10 can be connected to the aerial surveillance camera, and can obtain the RTSP video stream from the aerial surveillance camera by calling the SDK of the aerial surveillance camera and the preset video stream function interface. The perimeter intrusion identification system performs stream decoding processing on the RTSP video to obtain target image information including the target monitoring area (such as the inspection key area set according to user needs).

[0085] Exemplarily, the target image information may be color image data or grayscale image data.

[0086] According to an example embodiment, the region recognition module 20 sets a target boundary recognition region according to target image information.

[0087] For example, taking color image data as an example, based on the color image data, the perimeter intrusion recognition system responds to user instructions and sets the target boundary recognition area according to user needs.

[0088] The target boundary identification area can be a user-defined area with monitoring and identification significance, that is, an alarm prompt will be issued if there is perimeter intrusion in the target boundary identification area.

[0089] The region recognition module 20 can recognize the coordinates of a series of points in the target image information, and generate a polygonal region according to the coordinates of these points to form a recognition region.

[0090] Exemplarily, the target boundary recognition area may be a recognition area having boundary features such as a wall area or a fence area.

[0091] According to an example embodiment, the region recognition module 20 obtains an extended recognition region according to the target boundary recognition region.

[0092] For example, the region identification module 20 performs frame expansion processing on the target boundary identification region based on a preset first frame expansion ratio threshold to obtain a frame-expanded region (ie, an extended identification region) including the target boundary identification region.

[0093] Exemplarily, the first frame expansion ratio threshold may be customized according to user requirements, such as the first frame expansion ratio threshold may be [1.1, 2].

[0094] According to an exemplary embodiment, the image processing module 30 performs image preprocessing on the target boundary recognition area and the extended recognition area to obtain a target processed image.

[0095] For example, the image processing module 30 performs regionalized image preprocessing on the inside of the wall area or the fence area and the outside of the area after the frame expansion, such as using the same pixel value to fill the target boundary recognition area and the extended recognition area to obtain the filled image (target processed image).

[0096] In this way, the present invention can fill the wall area or fence area and the area outside the expanded area with the same pixel value, so that no person or vehicle can be detected in the wall area or fence area and the area outside the expanded area. After image processing, as long as a person or vehicle is detected in the image, it indicates that a perimeter intrusion behavior is suspected, and subsequent perimeter intrusion identification is performed.

[0097] According to an exemplary embodiment, the image processing module 30 performs image detection and recognition on the target processing image to determine whether a target object exists in the target processing image.

[0098] For example, the target object may be a subject of behavior that needs to be identified for perimeter intrusion, such as an illegal person or vehicle that performs perimeter intrusion.

[0099] The image processing module 30 can detect people or vehicles in the target processing image based on image segmentation and recognition technology. For example, the image segmentation and recognition technology includes but is not limited to SSD, YOLO, RetinaNet or RT-DETR.

[0100] According to an exemplary embodiment, when there is no target body in the target processing image, the image processing module 30 performs image segmentation on the target processing image to obtain at least two sub-target processing images.

[0101] For example, when no person or vehicle is detected, the image processing module 30 performs image segmentation on the filled target processing image based on the image segmentation technology to obtain N sub-target processing images, where N≥2.

[0102] According to an exemplary embodiment, the image processing module 30 performs image enlargement processing on at least two sub-target processed images to obtain at least two enlarged sub-target processed images.

[0103] For example, the image processing module 30 performs frame expansion area magnification processing and magnifies each sub-target processed image, thereby obtaining a plurality of magnified sub-target processed images.

[0104] The image processing module 30 performs image detection and recognition again on the at least two enlarged sub-target processed images, and determines that there is perimeter intrusion behavior when a target object exists in at least one enlarged sub-target processed image.

[0105] For example, the image processing module 30 detects people or vehicles in each enlarged sub-target processed image based on the image segmentation recognition technology. As long as a person or vehicle is detected in an enlarged sub-target processed image, the perimeter intrusion recognition system determines that there is perimeter intrusion behavior and issues an alarm prompt.

[0106] Through the above embodiments, the present invention can obtain a target processing image by preprocessing the images of the target boundary recognition area and the extended recognition area, and can determine whether there is a target body that invades the perimeter in the target processing image by detecting and recognizing the image of the target processing image; and the present invention can accurately recognize the target body for each sub-target processing image by further segmenting and amplifying the target processing image. The present invention can cover the perimeter area by setting a single monitoring device, and has the characteristics of low cost and high recognition accuracy.

[0107] Optionally, when there is a target object in the target processing image, the image processing module 30 determines a target object recognition area where the target object is located.

[0108] For example, when the image processing module 30 detects a person or a vehicle, the perimeter intrusion recognition system can generate a polygonal area according to the coordinates of a series of points in the target image information to form an identification area. For example, the perimeter intrusion recognition system can generate a person identification area or a vehicle identification area.

[0109] The image processing module 30 performs image magnification processing on the target body recognition area to obtain an enlarged target body recognition area.

[0110] The image processing module 30 performs image detection and recognition on the enlarged target body recognition area again, and determines that there is a perimeter intrusion behavior when there is a target body in the enlarged target body recognition area.

[0111] For example, the image processing module 30 performs a magnification process on the expanded frame area and magnifies the target body recognition area to obtain an enlarged target body recognition area. The perimeter intrusion recognition system again detects people or vehicles in the enlarged target body recognition area based on the image segmentation and recognition technology. When a person or vehicle is detected in the enlarged target body recognition area, the perimeter intrusion recognition system determines that there is a perimeter intrusion behavior and issues an alarm prompt.

[0112] With such arrangement, the present invention can avoid false detection of persons or vehicles detected for the first time by enlarging the expanded frame area, and can improve the recognition accuracy of the present invention by re-enlarging the detection and performing a secondary review.

[0113] Optionally, when the monitoring device is a high-altitude monitoring ball camera, the image processing module 30 calls the zoom function of the high-altitude monitoring ball camera to enlarge the area of ​​the sub-target processed image.

[0114] For example, when the monitoring device is a high-altitude monitoring dome camera, the image processing module 30 calls the SDK of the high-altitude monitoring dome camera and the preset video stream function interface to obtain the 3D positioning function (scaling function) of the high-altitude monitoring dome camera, and by adjusting the 3D positioning function, it is possible to enlarge the target monitoring area passed in, thereby achieving the enlargement of the sub-target processing image.

[0115] With such arrangement, the present invention can directly call the positioning function of the high-altitude monitoring ball camera to realize the magnification processing of the sub-target processing image. In addition, during the magnification processing, the high-altitude monitoring ball camera can be controlled to automatically zoom, which can ensure the imaging quality of the monitoring picture.

[0116] Optionally, when the monitoring device is a monocular camera, the image processing module 30 calls the image adjustment function in the computer vision library to enlarge the sub-target processed image. Alternatively, the image processing module 30 enlarges the sub-target processed image according to the image super-resolution technology to enlarge the sub-target processed image.

[0117] For example, the image processing module 30 can enlarge the sub-target processed image by calling the resize function (a function used to adjust the image size) in the OpenCV library (computer vision library); for another example, the image processing module 30 can also enlarge the sub-target processed image through image super-resolution technology, thereby enlarging the sub-target processed image.

[0118] With such a configuration, the present invention can achieve accurate recognition of small target detection such as sub-target processed images by adjusting the image adjustment function or enlarging the sub-target processed images through image super-resolution technology.

[0119] It can be understood here that the enlargement processing method for performing image enlargement processing on the target body recognition area in the present invention can be the same as the enlargement processing method for performing image enlargement processing on at least two sub-target processing images. Both can adopt enlargement methods such as calling the zoom function of the high-altitude monitoring ball camera, calling the image adjustment function in the computer vision library, or enlarging the image according to the image super-resolution technology. They have been described in detail above and will not be repeated here.

[0120] According to another aspect of the present invention, an electronic device is provided, which includes: one or more processors; and a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors can implement the perimeter intrusion identification method described above.

[0121] According to another aspect of the present invention, a non-volatile computer-readable storage medium is provided, wherein a computer program is stored on the storage medium, and when the computer program is executed by a processor, the perimeter intrusion identification method described above can be implemented.

[0122] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention is described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions of the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A perimeter intrusion identification method, characterized in that: include: Obtain target image information collected by monitoring equipment; Setting a target boundary recognition area according to the target image information; Obtaining an extended identification area according to the target boundary identification area; Performing image preprocessing on the target boundary recognition area and the extended recognition area to obtain a target processed image; Performing image detection and recognition on the target processing image to determine whether a target object exists in the target processing image; In the case that the target object does not exist in the target processing image, segmenting the target processing image to obtain at least two sub-target processing images; Performing image magnification processing on the at least two sub-target processed images to obtain at least two magnified sub-target processed images; Image detection and identification are performed again on the at least two enlarged sub-target processed images, and when the target object exists in at least one of the enlarged sub-target processed images, it is determined that there is perimeter intrusion behavior.

2. The perimeter intrusion identification method according to claim 1, characterized in that: The perimeter intrusion identification method further includes: In the case where the target object exists in the target processing image, determining a target object recognition area where the target object is located; Performing image magnification processing on the target body recognition area to obtain an enlarged target body recognition area; Image detection and recognition are performed again on the enlarged target body recognition area, and when the target body exists in the enlarged target body recognition area, it is determined that there is a perimeter intrusion behavior.

3. The perimeter intrusion identification method according to claim 1, characterized in that: In the case where the monitoring device is a high-altitude monitoring ball camera, the image enlargement processing of the at least two sub-target processed images includes: The zoom function of the high-altitude monitoring ball camera is called to enlarge the area of ​​the sub-target processed image.

4. The perimeter intrusion identification method according to claim 1, characterized in that: In the case where the monitoring device is a monocular camera, the image magnification processing of the at least two sub-target processed images includes: Calling an image adjustment function in a computer vision library to magnify the sub-target processed image; or The sub-target processed image is enlarged according to the image super-resolution technology so that the sub-target processed image is enlarged in area.

5. A perimeter intrusion identification system, characterized in that: include: An image acquisition module acquires target image information collected by the monitoring device; A region recognition module, which sets a target boundary recognition region according to the target image information, and obtains an extended recognition region according to the target boundary recognition region; The image processing module performs image preprocessing on the target boundary identification area and the extended identification area to obtain a target processing image, performs image detection and identification on the target processing image to determine whether a target body exists in the target processing image, and if the target body does not exist in the target processing image, performs image segmentation on the target processing image to obtain at least two sub-target processing images, performs image enlargement processing on the at least two sub-target processing images to obtain at least two enlarged sub-target processing images, and performs image detection and identification again on the at least two enlarged sub-target processing images, and if the target body exists in at least one of the enlarged sub-target processing images, determines that there is perimeter intrusion behavior.

6. The perimeter intrusion identification system according to claim 5, characterized in that: The image processing module determines a target object recognition area where the target object is located when the target object exists in the target processing image; The image processing module performs image magnification processing on the target body recognition area to obtain an enlarged target body recognition area; The image processing module performs image detection and recognition on the enlarged target body recognition area again, and determines that there is a perimeter intrusion behavior when the target body exists in the enlarged target body recognition area.

7. The perimeter intrusion identification system according to claim 5, characterized in that: In the case where the monitoring device is a high-altitude monitoring ball camera, the image processing module calls the zoom function of the high-altitude monitoring ball camera to enlarge the area of ​​the sub-target processed image.

8. The perimeter intrusion identification system according to claim 5, characterized in that: In the case where the monitoring device is a monocular camera, the image processing module calls an image adjustment function in a computer vision library to magnify the sub-target processed image; or The sub-target processed image is enlarged according to the image super-resolution technology so that the sub-target processed image is enlarged in area.

9. An electronic device, characterized in that: include: one or more processors; A storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the perimeter intrusion identification method as described in any one of claims 1-4.

10. A non-volatile computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the perimeter intrusion identification method as described in any one of claims 1 to 4 is implemented.

Images