KR102081577B1 - Intelligence Fire Detecting System Using CCTV - Google Patents

Abstract

The present invention relates to an intelligent fire detection system using CCTV,
A plurality of CCTVs including a camera module to collect the image information, an image storage server for receiving and storing the image information from the plurality of CCTV, receiving the image information from the image storage server to analyze whether or not a fire occurs, The image analysis server generating an event signal when determined to occur, the event analysis server receiving the event signal from the image analysis server to determine a fire type and a fire grade, the image analysis server, and the event analysis server. It relates to an intelligent fire detection system using CCTV, characterized in that it comprises a control client for receiving any one or more of information, whether the fire occurs, the type of fire, the fire rating.

Description

Intelligent Fire Detecting System Using CCTV

The present invention relates to an intelligent fire detection system using CCTV,

A plurality of CCTVs including a camera module to collect the image information, an image storage server for receiving and storing the image information from the plurality of CCTV, receiving the image information from the image storage server to analyze whether or not a fire occurs, The image analysis server generating an event signal, an event analysis server receiving the event signal from the image analysis server to determine a fire type and a fire grade, the image analysis server, and the event analysis server. It relates to an intelligent fire detection system utilizing CCTV, characterized in that it comprises a control client for receiving any one or more of information, whether the fire occurs, the type of fire, the fire rating.

The present invention relates to an intelligent fire detection system utilizing CCTV.

Korean Patent No. 10-1111493, which is a prior art, is equipped with a processor capable of image processing for recognizing a flame from an image, and detects a fire from an intelligent IP camera that monitors fires in each region, and alarms in three stages. An automatic fire recognition alarm system and method using an intelligent IP camera which transmits real-time image information through a wired Internet and automatically transmits wirelessly when a wired communication is impossible is continuously transmitted.

In addition, the prior art Republic of Korea Patent No. 10-1635000 includes the temperature data processing unit for receiving the temperature data for the first area, and generates a first event when the temperature exceeding a predetermined reference is included, the first area The fire detection apparatus including an image analysis unit for generating a second event and a communication unit for transmitting a signal corresponding to the first event and the second event if the flame pattern is included by analyzing the image information of the second area Is provided.

In the case of the prior art as described above, there is a drawback that a false detection may occur because it does not provide a method for discriminating an image similar to a fire occurrence image.

Republic of Korea Patent No.: 10-1111493 Republic of Korea Patent No.: 10-1635000

The present invention aims to solve the problems of the prior art as described above. Specifically, an object of the present invention is to provide an intelligent fire detection system utilizing CCTV.

In addition, by using machine learning to determine whether the first fire, and by analyzing the time-dependent pixel change in the area suspected of a fire by the second judgment to reduce the false detection.

Technical problems to be achieved by the present invention are not limited to the above-mentioned problems, and other technical problems not mentioned above may be clearly understood by those skilled in the art from the following description. There will be.

Intelligent fire detection system using CCTV according to the present invention for achieving this object, a plurality of CCTV including the camera module to collect the image information, image storage server for receiving and storing the image information from the plurality of CCTV, The image analysis server receives the image information from the image storage server and analyzes whether a fire has occurred, and if it is determined that a fire has occurred, an image analysis server generating an event signal, and receiving the event signal from the image analysis server to determine a fire type and a fire rating. And a control client connected to the event analysis server, the image analysis server, and the event analysis server to determine any one or more of the image information, the fire occurrence, the fire type, and the fire grade. .

Receiving a CCTV video stream list from the video storage server, and if the video analysis server requests a video stream address further comprises a video distribution server for transmitting the video stream address to the video analysis server, the video analysis server is the A video stream address is requested from an image distribution server, the address is received from the image distribution server, and the image storage server receives a video stream corresponding to the address.

The fire detection method of the image analysis server comprises the steps of generating a fire determination model by inputting a fire image, a fire-like image similar to a fire into a machine learning algorithm; A first determination step of analyzing the fire determination by inputting the image information to the fire determination model generated in the step; In the first determination step, an area estimated as a fire is extracted from the image determined to be fire, and a pixel change amount is detected according to time of the area. A second determination step of determining that it is a fire if it exceeds; It includes.

As described above, the present invention has the effect of providing an intelligent fire detection system utilizing CCTV.

In addition, it is possible to reduce false detection by primarily determining whether a fire occurs first by using machine learning, and by determining whether the fire is secondary by analyzing the change in pixels of the area suspected to be fired.

The technical effects of the present invention are not limited to the technical effects mentioned above, and other technical effects not mentioned above may be clearly understood by those skilled in the art from the description of the claims. There will be.

1 is a view showing the configuration of an intelligent fire detection system using CCTV according to an embodiment of the present invention.
FIG. 2 is a diagram illustrating an image-based deep learning technology and a concept of context awareness.
3 is a view showing a fire image input as a training image in an embodiment of the present invention.
4 is a view showing a fire-like image input as a training image in an embodiment of the present invention.
FIG. 5 is a diagram illustrating a change degree of a fire image region in an embodiment of the present invention.
FIG. 6 is a diagram illustrating a change degree of a fire-like image region in an embodiment of the present invention. FIG.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. For reference, the size of the components, the thickness of the line, and the like shown in the drawings referred to for describing the present invention may be somewhat exaggerated for ease of understanding. In addition, since terms used in the description of the present invention are defined in consideration of functions in the present invention, they may be changed according to a user, an operator's intention, and a custom. Therefore, the definition of this term should be made based on the contents throughout the present specification.

In addition, although the preferred embodiment of the present invention will be described in detail with reference to the accompanying drawings, the object of the present invention may be specifically realized, which is intended for easier understanding of the present invention, the scope of the present invention is limited thereto It is not. In addition, in describing the embodiments of the present invention, the same name and the same reference numerals are used for the same configuration and additional description thereof will be omitted.

When any part of the specification is to "include" any component, this means that it may further include other components, except to exclude other components unless specifically stated otherwise. In addition, the terms "... unit", "module", etc. described in the specification mean a unit for processing at least one function or operation, which may be implemented in hardware or software or a combination of hardware and software. .

Throughout the specification, when a part is "connected" to another part, it includes not only "directly connected" but also "electrically connected" with another element in between. .

CCTV control system is divided into CMS (Central Monitoring System) and VMS (Video Management System). CMS is a central management system that installs all programs on one PC. It does not use a database, and is suitable for small to medium sized systems and is affected by picture quality in proportion to the number of users. VMS is an open-distributed security management system composed of Server-Client structure. It uses database, maintains the set image quality, and is suitable for medium to large systems. Intelligent fire detection system using CCTV according to an embodiment of the present invention uses a video management system (VMS).

1 is a view showing the configuration of an intelligent fire detection system using CCTV according to an embodiment of the present invention.

1, an intelligent fire detection system using CCTV according to an embodiment of the present invention includes an image storage server 101, an image analysis server 102, an image distribution server 105, an event analysis server 103, It is configured to include a control client 104. In addition, although not shown in Figure 1, including a plurality of CCTV including a camera module, and collects the image information from a plurality of CCTV to analyze the fire in the fire detection system.

The image storage server 101 receives and stores image information from the plurality of CCTVs. The image analysis server 102 receives image information from the image storage server 101 to analyze whether a fire has occurred, and generates an event signal when it is determined that a fire has occurred. In one embodiment of the present invention, the event signal includes image information determined as a fire, date and time when the image was taken, CCTV location, and the like. The event analysis server 103 receives the event signal from the image analysis server to determine a fire type and a fire grade. Fire types are classified into building fires, forest fires, ship fires, and vehicle fires, depending on the fire. The control client 104 is connected to an image analysis server and an event analysis server to receive one or more of video information, fire occurrence, fire type, and fire grade so that the controller can check. The video distribution server 105 receives a CCTV video stream list from the video storage server 101 and transmits the video stream address to the video analysis server 102 when the video analysis server 102 requests the video stream address. .

The image analysis server 102 requests an image stream address from the image distribution server 105, receives an image stream address from the image distribution server 105, and stores the image corresponding to the image stream address in the image storage server 101. Receive the stream.

FIG. 2 is a diagram illustrating an image-based deep learning technology and a concept of context awareness. As shown in FIG. 2, the existing fire-related image is collected and classified into a fire image as shown in FIG. 3 and a fire-like image as shown in FIG. 4 to perform machine learning. Fire-like images include street lamps, car lights, and the sun. The fire determination model generated according to the mechanical heart is connected to the video storage server to determine the fire in the video stream. Even if the machine learning is performed, a fire-like image similar to a fire image may be incorrectly detected as a fire as shown in FIG. 4. In order to prevent this, an embodiment of the present invention detects a pixel change in an image region to determine a second judgment. To perform. It is a figure which shows the fire-like image input.

In one embodiment of the present invention, the machine learning algorithm uses Convolutional Neural Network (CNN). CNN adds a convolution task to the front of the existing neural network to preprocess the data that it enters into the neural network.

FIG. 5 is a diagram illustrating a change degree of a fire image region in an embodiment of the present invention.

FIG. 6 is a diagram illustrating a change degree of a fire-like image region in an embodiment of the present invention. FIG.

In one embodiment of the present invention, the fire detection method identifies whether a fire has occurred by using an intelligent fire determination model, and an image determined to be a fire as shown in FIGS. In the current image (P _T ), the previous image (P _T-1 ), and the next image (P _{T + 1} ), the area estimated to be fired is displayed as a box (Fire ₁ ). In the box, the difference between the pixel values of the current image (P _T ), the previous image (P _T-1 ), the current image (P _T ), and the next image (P _{T + 1} ) is calculated as an absolute value. The sum is divided by the number of pixels to calculate the change amount V _T. If the change amount V _T is equal to or less than the reference value T, it is judged as a false detection and not signaled by the fire.

는 현재 영상(P_T)의 박스영역에서 직전 영상(P_T-1)과 다음 영상(P_T+1)의 화소값의 변화량, Fire₁ 은 현재 영상(P_T)에서 화재로 예측된 영역, l, m은 화재로 예측된 영역의 가로 크기와 세로 크기, T는 화소값의 변화량 기준값을 나타낸다.here,

Is the change amount of the pixel values of the previous image (P _T-1 ) and the next image (P _{T + 1} ) in the box region of the current image (P _T ), Fire ₁ is the area predicted to be fire in the current image (P _T ), l and m represent the horizontal and vertical sizes of the region predicted by the fire, and T represents the reference value of the change amount of the pixel value.

Determining whether the image is a fire image or a fire-like image by exceeding the reference value T, as shown in FIGS. 5 and 6, in the case of a fire-like image, there is almost no change in the area recognized as flame, but in the case of an actual fire Because the change is large and irregular, it is used to prevent false detections.

 Although described with reference to the drawings according to an embodiment of the present invention, those skilled in the art will be able to perform a variety of applications, modifications and adaptations within the scope of the present invention based on the above contents. . Accordingly, the true scope of protection of the invention should be defined only by the appended claims.

101: video storage server
102: video analysis server
103: event analysis server
104: control client
105: video distribution server

Claims (3)

Multiple CCTVs that collect video information, including camera modules
Image storage server for receiving and storing the image information from the plurality of CCTV,
An image analysis server configured to receive the image information from the image storage server and analyze whether a fire has occurred, and generate an event signal when it is determined that a fire has occurred;
An event analysis server which receives the event signal from the image analysis server and determines a fire type and a fire grade;
And a control client connected to the video analysis server and the event analysis server to receive one or more of the video information, whether the fire has occurred, the fire type, and the fire rating.
Receiving a CCTV video stream list from the video storage server, and further comprises a video distribution server for transmitting the video stream address to the video analysis server when the video stream address is requested from the video analysis server,
The video analysis server requests a video stream address from the video distribution server, receives the address from the video distribution server, and receives a video stream corresponding to the address from the video storage server. In a fire detection method of an intelligent fire detection system image analysis server,
Generating a fire judgment model by inputting a fire image and a fire-like image similar to the fire into a machine learning algorithm;
A first judging step of analyzing the fire judging by inputting the image information to the fire judging model generated in the above step;
In the first determination step, a region estimated as a fire is extracted from the image determined to be fire, and the pixel change amount is detected according to time of the region. And a second determination step of determining that it is a fire if exceeded.
The fire detection method of the second determination step determines the fire according to the following formula,

Here, a box (Fire ₁ ) is displayed as a box on the current image (P _T ), the previous image (P _T-1 ), and the next image (P _{T + 1} ), which is estimated to be a fire, and the current image (P _T ) is inside the box. The difference between the pixel values of _T ), the previous image (P _T-1 ), the current image (P _T ), and the next image (P _{T + 1} ) is calculated as an absolute value as shown in the following equation, and the two absolute values are summed and divided by the number of pixels. Calculate the change amount (V _T ). If the change amount V _T is less than or equal to the reference value T, it is judged as a false detection and not signaled by the fire.

Is the change amount of the pixel values of the previous image (P _T-1 ) and the next image (P _{T + 1} ) in the box region of the current image (P _T ), Fire ₁ is the area predicted to be fire in the current image (P _T ), l, m, the horizontal and vertical size of the area predicted by the fire, T represents a reference value change amount of the pixel value
delete delete

Images