KR102077632B1 - Hybrid intellgent monitoring system suing local image analysis and cloud service - Google Patents

Abstract

The present invention relates to a hybrid intelligent intrusion monitoring system using a cloud service and local image analysis which allows a local unit and a cloud service unit to cooperate with each other to analyze an image for monitoring an intruder or the like to increase quickness and accuracy of image analysis and reduce a cost burden of system construction. According to the present invention, the hybrid intelligent intrusion monitoring system using a cloud service and local image analysis comprises: a local unit including a camera to photograph an image of a site, a local image analysis module to analyze an image of the camera, and a transceiving module to transmit an image of the camera; and a cloud service unit including a communication module to communicate with the transceiving module, a database to store and manage an image of the camera received via the communication module and a cloud image analysis module to analyze an image stored in the database. The local unit requests analysis for a corresponding image from the cloud service unit when a change of the image is recognized by the local image analysis module and takes measures by the cloud image analysis module depending on whether the change of the image is recognized in response to the request.

Description

Hybrid intelligent intrusion monitoring system using local image analysis and cloud service {Hybrid intellgent monitoring system suing local image analysis and cloud service}

The present invention is a hybrid intelligence that utilizes a local video analysis and cloud service that increases the speed and accuracy of video analysis and reduces the cost of system construction by collaborating the analysis of video for surveillance of intruders in the local unit and the cloud service unit. It relates to an intrusion monitoring system.

Surveillance cameras (CCTVs) that monitor and record the scene have evolved from a service that monitors the image of the scene in real time or record the image to check it later, so that objects appear in the field in recent years (i.e., an event occurs in the image) To reduce the storage burden of image capacity by intensively recording the images in the time zone before and after the object emerges, and to strengthen the monitoring function by notifying the person concerned immediately when the emergence of the object is recognized, taking one step further while tracking the movement of the object And it is developing into an intelligent service such as photographing by expanding (focusing) the part that can specify the object.

In order to provide intelligent service of surveillance cameras, the events of the current majority of surveillance cameras (CCTV) are stored (recorded) in DVR (Digital Video Recoder) or NVR (Network Video Recoder) based storage devices, and the video is analyzed. A server is built to recognize (object appearance) and have the camera track or focus the object.

For such intelligent service of surveillance cameras, a local video analysis system that builds a DVR or NVR-based storage device in the field where the surveillance camera is installed, and a server that analyzes the stored images and tracks the objects, has a high cost, and There is a problem that it is difficult to objectively verify the results of the analysis.

 For reference, prior art related to the local image analysis system has been disclosed, such as "Digital image monitoring and analysis system" registered patent No. 10-1016243, "Multipurpose integrated operation CCTV system" registered patent No. 10-1786923.

Recently, a cloud image analysis system with high processing power and high reliability has been introduced, so the video captured by the field surveillance camera (cloud cam) is stored in a cloud storage device built on the Internet and analyzed by a cloud server. have.

The cloud image analysis system has a high cost of storing the image in the cloud, has a high dependency on the network state according to the real-time transmission of the video from the surveillance camera, and separates it from the local unit to receive and process the real-time image analysis result of the cloud service. Software is required.

Therefore, the cloud service is suitable to be applied to a small amount of surveillance cameras, but it is difficult to apply in a large-scale surveillance camera system due to the large burden of cloud usage.

For reference, prior art related to the cloud image analysis system is registered patent No. 10-1738334 "CCTV security monitoring device and method using cloud computing environment", registered patent No. 10-1095951 "Image information stored in digital video recorder A cloud computing based video surveillance system based on using "has been disclosed.

The present invention is an invention devised to solve the problems of the local image analysis system and the cloud image analysis system according to the prior art, and reduces the cost burden through collaboration between the local unit and the cloud unit, and results in fast and reliable image analysis. The objective is to provide a hybrid intelligent intrusion monitoring system that utilizes a cloud service and local video analysis to acquire and monitor functions.

Hybrid intelligent intrusion monitoring system using local image analysis and cloud service according to the present invention to achieve the above object

A camera that shoots footage from the scene,

Local image analysis module for analyzing the image of the camera,

A local unit including a transmitting and receiving module for transmitting the image of the camera;

A communication module communicating with the transmission / reception module,

A database for storing and managing the image of the camera received through the communication module;

Includes a cloud service unit including a cloud image analysis module for analyzing the image stored in the database;

The local unit

When a change in the image is recognized by the local image analysis module, the cloud service unit requests analysis of the corresponding image,

In response to the request, the cloud image analysis module is characterized by taking action according to whether or not the change in the image is recognized.

In addition, the local unit is characterized in that the image capture ratio of the image for analysis is adjusted in the local image analysis module according to whether the change in the image is recognized in the cloud image analysis module in response to the request,

The local image analysis module is characterized by having a background extraction algorithm for recognizing scene changes and an object extraction algorithm for recognizing human objects.

The hybrid intelligent intrusion monitoring system using the local image analysis and cloud service according to the present invention configured as above is a software for briefly analyzing the image at a pre-processing level without additional hardware (server) in the local unit, and the local image analysis module By adding, it is possible to reduce the construction cost of the local unit, and the cloud unit analyzes the video at the time requested by the local unit as the main target, thereby minimizing the burden on the usage fee of the cloud, and primary video from the local unit. It is a highly useful invention for industrial development as it is a highly reliable invention for image analysis results by analyzing images in a cloud unit second when a change in image (eg, intruder occurrence) is recognized by analyzing.

1 is a block diagram of a hybrid intelligent intrusion monitoring system utilizing local image analysis and cloud services according to the present invention.
2 is a view showing intuitively the image analysis process in a hybrid intelligent intrusion monitoring system using a local image analysis and cloud service according to the present invention.
3 is a diagram showing a frame capture ratio during normal and event events.
Figure 4 is a process diagram showing the processing during the intrusion event.
5 is a process diagram showing a processing process during a roaming event.

Hereinafter, a hybrid intelligent intrusion monitoring system using local image analysis and a cloud service according to the present invention will be described in detail with reference to the drawings.

Before explaining in more detail about the hybrid intelligent intrusion monitoring system using the local video analysis and cloud service according to the present invention,

The present invention can be applied to a variety of changes and can have a variety of forms, the implementation (態 樣, aspect) (or embodiments) will be described in detail in the text. However, it is not intended to limit the present invention to a specific disclosure form, it should be understood to include all modifications, equivalents, or substitutes included in the spirit and scope of the present invention.

The same reference numbers in each drawing, especially the number of tens and ones, or the same number of tens, ones and alphabets indicate the members having the same or similar functions, and each of the drawings unless otherwise specified. The member indicated by the reference numeral can be understood as a member conforming to these standards.

In addition, in each drawing, the elements are exaggeratedly large (or thick) or smallly (or thinly) or simplified to express the size or thickness in consideration of convenience, etc., thereby limiting the scope of the present invention. It should not be.

The terminology used herein is only used to describe a specific embodiment (sun, 態 樣, aspect) (or embodiment), and is not intended to limit the present invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this application, terms such as ~ include ~ or ~ consist of ~ are intended to designate the existence of features, numbers, steps, operations, components, parts, or combinations thereof described in the specification, and one or more other features. It should be understood that the existence or addition possibilities of fields or numbers, steps, actions, components, parts or combinations thereof are not excluded in advance.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by a person skilled in the art to which the present invention pertains. Terms such as those defined in a commonly used dictionary should be interpreted as having meanings consistent with meanings in the context of related technologies, and should not be interpreted as ideal or excessively formal meanings unless explicitly defined in the present application. Does not.

As shown in FIG. 1, a hybrid intelligent intrusion monitoring system utilizing a local video analysis and cloud service according to the present invention is built in the field, and a local unit 10 managed and operated by a user and a cloud company when built on the Internet. It is classified as a component of the cloud unit 20 managed and operated.

The local unit 10 includes a camera 11, a memory 13, a local image analysis module 15, a transmission / reception module 19 and a controller 17.

The camera 11 is installed in the field and photographs the surveillance area, which may be desirable to have a shooting direction adjustment function and a zoom-in / zoom-out focusing function so that objects appearing in the surveillance area can be tracked and photographed.

The memory 13 stores an image captured by the camera 11. The image storage of the memory 13 may be stored in a DVR or NVR manner.

 The memory 13 captures and stores image frames constituting the image taken by the camera 11, and as shown in FIG. 3, the capture ratio of the image frame is variable, thereby reducing the burden of storage capacity. Therefore, it is desirable to enhance the monitoring function of the intrusion object by reducing the capture rate of the image frame during normal times when there is no intrusion, and increasing the capture rate of the image frame during the event section where the intrusion occurs.

As described above, if the capture ratio of the image frame is different in the normal and event period, the hardware burden of the local unit 10 can be reduced, and the storage capacity burden and cost of the cloud unit 20 that stores and manages it can be reduced.

The local image analysis module 15 analyzes the image stored in the memory 13 to recognize whether there is a change in the image taken by the camera 11, that is, whether an intruder or a rover appears in the surveillance area.

 The local image analysis module 15 analyzes an image captured by the camera 11 (more specifically, an image frame captured and stored in the memory 13) to determine whether there is a change in the image.

Two methods of scene change detection and object recognition may be used for the analysis of the local image analysis module.

The scene change detection may compare the current image with the previous image, and determine that there is an appearance or movement of an object in the image when a change over a certain value occurs.

The background extraction algorithm 15a may be used to determine scene change detection. The background extraction algorithm 15a calculates a foreground mask by binarizing an image obtained through subtraction of a background model scene for extracting a current frame scene and an object. The background extraction algorithm 15a determines that a scene change is detected when a change in the surveillance area is detected.

The scene change detection has a disadvantage in that it does not recognize objects that are in the surveillance area but do not move. In order to compensate for this, an object extraction algorithm 15b is provided.

The object extraction algorithm 15b performs recognition of a person (ie, an object) using a detection expression.

Haar feature and Hog Descriptor exist in the calculation formula to detect a representative person, but the Haar feature has relatively low accuracy, but the processing speed is faster and the learning is done with a smaller template than the Hog Descriptor trained with a 64 * 128 pedestrian template. Since it has the advantage of being able to recognize a human object of a smaller size, it may be desirable to proceed by selecting a Haar feature.

Background extraction algorithm (15a) and object extraction algorithm (15b) of the local image analysis module 15 is a program that analyzes the pre-processed image using a relatively simple technique, which has the advantage of being small in size and capable of rapid analysis, while being accurate. It has a relatively low disadvantage. These disadvantages are complemented through collaboration with the cloud unit (20). This will be described later.

The transmission / reception module 19 communicates with the communication module of the cloud unit 20 through the Internet network, and transmits various data such as an image captured by the camera 11, an analysis request of the image, and the like, and the cloud unit 20, The cloud unit 20 receives data on an analysis result transmitted from the cloud unit 20.

The controller 17 controls the local unit 10 as a whole.

For example, according to the normal and during an event, the capture rate of a frame for storing the image taken by the camera 11 in the memory 13 is adjusted, and there is a change in the image in the local image analysis module 15. When recognized, the cloud unit 20 requests analysis of the corresponding image, and when an event is received as an analysis result from the cloud unit 20, the frame capture rate is increased, and the shooting direction of the camera 11 is adjusted or focused. To control the local unit 10 by tracking the object or zooming in.

The cloud unit 20 stores and manages the image of the field transmitted by the local unit 10, analyzes the stored image, determines whether an object appears in the field's surveillance area, and determines the determination result as the local unit 10 ) And related parties.

The cloud unit 20 is a communication module 29 that communicates with the local unit 10 through an Internet network, a database 23 that stores and manages images transmitted by the local unit 10, and the database 23 ) Includes a cloud image analysis module (25) that analyzes the stored image, a controller (27) that controls the cloud unit (20) as a whole.

The cloud image analysis module 25 analyzes images with high accuracy and reliability.

The cloud image analysis module 25 may analyze the image closely to distinguish the types of objects appearing in the image, and may predict the direction of movement of the object.

The present invention performs pre-processing for analyzing the image of the camera 11 in the local image analysis module 15 of the local unit 10, the cloud image analysis module 25 of the cloud unit 20 is a local image analysis module In (15), the pre-processed image is analyzed to reduce the analysis load.

In addition, the local image analysis module 15 quickly analyzes the image through the background extraction algorithm 15a and the object extraction algorithm 15b, in which case the camera 11 is shaken by the wind to cause a change in the image. Even if an animal such as a bird or a cat appears in the surveillance area, even if there is no intruder or rover, the error of recognizing the intrusion or roaming is complemented by accurate analysis of the cloud image analysis module 25.

FIG. 2 is a view intuitively showing the process of analyzing and processing the image of the camera 11 in the hybrid intelligent intrusion monitoring system using the local image analysis and cloud service according to the present invention.

Referring to FIG. 2, an image captured by the camera 11 of the local unit 10 is captured and stored in the memory 13 (S11).

At this time, the frame of the captured image is adjusted according to whether an event occurs or not. That is, the capture ratio is usually low, and the capture ratio is high during the event (S11).

  The local image analysis module 15 analyzes whether a person has appeared in the image using the object extraction algorithm 15b (S12), and analyzes whether there is a change in the screen of the image through the background extraction algorithm 15a (S13). When it is analyzed that a person appears in a video or changes in the video screen (S14), the local unit 10 requests an accurate analysis of the video from the cloud unit 20 (S15).

The cloud unit 20 transmits the analysis result according to the request to the local unit 10, and the local unit 10 changes the image or the unmonitored object (eg; In case of non-event (False) that does not require monitoring, such as the appearance of birds, cats), it monitors in normal mode, and when an analysis result shows a monitoring object, it monitors in event mode (True). The capture rate of the image frame is increased, and information about the event is recorded (S16, S17, S18).

Events that need to be monitored because an object is recognized in an image can be divided into cases in which a person enters the surveillance area, and a person enters the surveillance area and a person who enters the surveillance area does not leave the surveillance area for more than a certain period of time. In the former case, an intrusion event is generated, and in the latter case, a roving event is generated.

4 and 5 show flowcharts for an intrusion event and flowcharts for a roving event, respectively.

First, as a process of determining an intrusion event and a roaming event, the local image analysis module 15 of the local unit 10 analyzes the captured image (screen) of the image to determine whether there is a human object, and the captured current image Compare and analyze the image captured immediately before and check if there is a change in the screen (S21 ~ S23).

 If any one of the check results in an abnormality, the cloud unit 20 requests accurate analysis of the corresponding image (S24 to S25).

Referring to FIG. 4, when a cloud analysis module recognizes that a person enters a surveillance area as a result of an analysis in response to a request (S31), an intrusion event occurs and the frame rate is increased by using an adaptive frame capture rate function ( S32) At the same time as accurate detection, the event occurrence time is recorded, and intrusion detection personnel are updated and images are stored in preparation for additional personnel to enter the surveillance area (S33, S34). At this time, the event occurrence time is determined as the time when the last person enters.

After that, when all the people are out of the surveillance area and a certain time (10 seconds) has elapsed, the intrusion event ends and the frame rate is lowered again and another intrusion situation is prepared (S35 ~ S39).

Referring to FIG. 5, when the cloud analysis module responds to the request and the analysis result shows that a person enters the surveillance area (S31 to S33) and 10 seconds elapse (S41), a roaming event occurs. At this time, the frame rate is also increased to increase accuracy and record and process events in the same way as intrusion events (S41 to S54).

In the above description of the present invention, a hybrid intelligent intrusion monitoring system using a local image analysis and cloud service having a specific shape and structure has been described with reference to the accompanying drawings, but the present invention can be modified and modified in various ways by those skilled in the art. And, such modifications and variations should be construed as belonging to the protection scope of the present invention.

10: local unit 11: camera
13: memory 15: local image analysis module
17: controller 19: send and receive module
20: cloud unit 23: database
25: cloud image analysis module 27: controller
29: communication module

Claims (3)

A camera that shoots footage from the scene,
Local image analysis module for analyzing the image of the camera,
A local unit including a transmitting and receiving module for transmitting the image of the camera;

A communication module communicating with the transmission / reception module,
A database for storing and managing the image of the camera received through the communication module;
Includes a cloud service unit including a cloud image analysis module for analyzing the image stored in the database;

The local unit
When a change in the image is recognized by the local image analysis module, the cloud service unit requests analysis of the corresponding image,
In response to the request, the cloud image analysis module takes action according to whether the image is recognized or not,

The local unit responds to requests for changes in video in the cloud video analysis module.
In case of non-event that does not require monitoring, the capture rate of the image frame is lowered to monitor in normal mode.
A hybrid intelligent intrusion monitoring system that utilizes a local video analysis and cloud service, which is characterized in that it monitors in an event mode by increasing the capture rate of an image frame when an event is needed. delete According to claim 1,
The local image analysis module is a hybrid intelligent intrusion monitoring system that utilizes a local image analysis and cloud service, which is equipped with a background extraction algorithm that recognizes scene changes and an object extraction algorithm that recognizes human objects.

Images