CN111488772A - Method and apparatus for smoke detection - Google Patents
Method and apparatus for smoke detection Download PDFInfo
- Publication number
- CN111488772A CN111488772A CN201910087220.5A CN201910087220A CN111488772A CN 111488772 A CN111488772 A CN 111488772A CN 201910087220 A CN201910087220 A CN 201910087220A CN 111488772 A CN111488772 A CN 111488772A
- Authority
- CN
- China
- Prior art keywords
- video frame
- smoke
- probability
- area
- frame image
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B17/00—Fire alarms; Alarms responsive to explosion
- G08B17/10—Actuation by presence of smoke or gases, e.g. automatic alarm devices for analysing flowing fluid materials by the use of optical means
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Software Systems (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Business, Economics & Management (AREA)
- Emergency Management (AREA)
- Computational Linguistics (AREA)
- Image Analysis (AREA)
- Alarm Systems (AREA)
- Closed-Circuit Television Systems (AREA)
Abstract
The disclosure provides a smoke detection method and a smoke detection device, and belongs to the technical field of video monitoring. The method comprises the following steps: during smoke detection, a suspected smoke area in a target video frame image can be determined according to the target video frame image, then a first preset number of video frame images continuously shot at a target position after the target video frame image is shot are obtained, wherein the target position is the position where a camera is located when the target video frame image is shot, then the first preset number of video frame images are detected, and if the camera covers the area and the smoke area exists, a preset smoke prompt signal is controlled to be sent to a remote terminal device. By adopting the method and the device, the accuracy of smoke detection can be improved.
Description
Technical Field
The disclosure relates to the technical field of video monitoring, and in particular relates to a method and a device for smoke detection.
Background
Smoke is an early expression of fire, so people often judge whether a fire exists or not through smoke, and video-based smoke detection is concerned by people because video monitoring is not limited by space distance and scenes, and video monitoring technology is continuously developed and video monitoring points are widely distributed.
In the related art, when smoke is detected, a video frame image is obtained from video monitoring, smoke features are extracted from the video frame image, whether the smoke is real smoke or not is judged based on a preset experience threshold, and whether prompt is required or not is determined.
Since the video frame image is easily affected by factors such as illumination, scene, camera erection mode, etc., it is determined whether the image is real smoke only based on the preset experience threshold, which may result in a high false detection rate.
Disclosure of Invention
In order to solve the problems of the prior art, the embodiments of the present disclosure provide a method and an apparatus for smoke detection. The technical scheme is as follows:
in a first aspect, there is provided a method of smoke detection, the method comprising:
determining a suspected smoke area in a target video frame image according to the target video frame image;
acquiring a first preset number of video frame images continuously shot at the target position after the target video frame image is shot, wherein the target position is the position of a camera when the target video frame image is shot;
and detecting the first preset number of video frame images, and if the camera coverage area has a smoke area, controlling to send a preset smoke prompt signal to a remote terminal device.
Optionally, the camera is a rotary camera;
the acquiring, before a first preset number of video frame images continuously captured at the target position after the target video frame image is captured, further includes:
and controlling the rotating camera to rotate to the target position.
Optionally, the method further includes:
and if the camera coverage area does not have a smoke area, controlling the rotating camera to continue rotating.
Optionally, the detecting the first preset number of video frame images includes:
inputting data of RGB channels of the first preset number of video frame images into a first preset classification model to obtain a first probability that a smog area exists in a camera coverage area and a second probability that the smog area does not exist in the camera coverage area;
determining that a smoke region exists in the camera coverage area if the first probability is greater than or equal to a second probability, and determining that a smoke region does not exist in the camera coverage area if the first probability is less than the second probability.
Optionally, the method further includes:
and controlling to send the position information of the smoke area and the first preset number of video frame images to the remote terminal equipment.
In this way, the worker can be made to determine the smoke region more quickly.
Optionally, the determining, according to the target video frame image, that the suspected smoke area exists in the target video frame image includes:
inputting data of RGB channels of the target video frame image into a second preset classification model to obtain the probability of a suspected smoke area and the probability of the suspected smoke area in the target video frame image;
and determining that the suspected smoke area exists in the target video frame image according to the probability of the suspected smoke area and the probability of the suspected smoke area not existing.
Thus, false alarms can be prevented.
Optionally, the determining that the suspected smoke area exists in the target video frame image according to the probability of the suspected smoke area and the probability of the suspected smoke area not existing includes:
if the probability of the suspected smoke area is larger than or equal to the probability of the suspected smoke area, acquiring a second preset number of video frame images shot at the target position, wherein the second preset number of video frame images refer to video frame images on the left side and video frame images on the right side of the video frame images on the left side which are adjacent to the target video frame image in terms of time;
determining that a suspected smoke region exists in the target video frame image if the features of the suspected smoke region in the second preset number of video frame images are associated with the features of the suspected smoke region in the target video frame image, wherein the features are one or more of position features, shape features and color features.
Thus, false alarms can be prevented.
In a second aspect, there is provided a device for smoke detection, the device comprising:
the determining module is used for determining that a suspected smoke area exists in a target video frame image according to the target video frame image;
the acquisition module is used for acquiring a first preset number of video frame images which are continuously shot at the target position after the target video frame image is shot, wherein the target position is the position of a camera when the target video frame image is shot;
and the control module is used for detecting the first preset number of video frame images, and controlling to send a preset smoke prompt signal to the remote terminal equipment if the camera coverage area has a smoke area.
Optionally, the camera is a rotary camera;
the control module is further configured to:
and controlling the rotary camera to rotate to the target position before a first preset number of video frame images continuously shot at the target position after the target video frame image is shot are obtained.
Optionally, the control module is further configured to:
and if the camera coverage area does not have a smoke area, controlling the rotating camera to continue rotating.
Optionally, the control module is configured to:
inputting data of RGB channels of the first preset number of video frame images into a first preset classification model to obtain a first probability that a smog area exists in a camera coverage area and a second probability that the smog area does not exist in the camera coverage area;
determining that a smoke region exists in the camera coverage area if the first probability is greater than or equal to a second probability, and determining that a smoke region does not exist in the camera coverage area if the first probability is less than the second probability.
Optionally, the control module is further configured to:
and controlling to send the position information of the smoke area and the first preset number of video frame images to the remote terminal equipment.
Optionally, the determining module is configured to:
inputting data of RGB channels of the target video frame image into a second preset classification model to obtain the probability of a suspected smoke area and the probability of the suspected smoke area in the target video frame image;
and determining that the suspected smoke area exists in the target video frame image according to the probability of the suspected smoke area and the probability of the suspected smoke area not existing.
Optionally, the determining module is configured to:
if the probability of the suspected smoke area is larger than or equal to the probability of the suspected smoke area, acquiring a second preset number of video frame images shot at the target position, wherein the second preset number of video frame images refer to video frame images on the left side and video frame images on the right side of the video frame images on the left side which are adjacent to the target video frame image in terms of time;
determining that a suspected smoke region exists in the target video frame image if the features of the suspected smoke region in the second preset number of video frame images are associated with the features of the suspected smoke region in the target video frame image, wherein the features are one or more of position features, shape features and color features.
In a third aspect, there is provided a smoke detection apparatus, the apparatus comprising: a processor and a memory; the memory is used for storing at least one instruction; the processor is configured to execute at least one instruction stored in the memory to implement the method steps of the first aspect.
In a fourth aspect, a computer-readable storage medium is provided, having stored therein at least one instruction, which when executed by a processor, performs the method steps of the first aspect described above.
The beneficial effects brought by the technical scheme provided by the embodiment of the disclosure at least comprise:
in the embodiment of the disclosure, during smoke detection, a suspected smoke area in a target video frame image can be determined according to the target video frame image, then a first preset number of video frame images continuously shot at a target position after the target video frame image is shot are obtained, wherein the target position is a position where a camera is located when the target video frame image is shot, then the first preset number of video frame images are detected, and if the camera covers the area, a preset smoke prompt signal is controlled to be sent to a remote terminal device. In this way, since the determination is made twice after the presence of the suspected smoke region is determined, rather than being based solely on the empirical threshold, the accuracy of smoke detection can be improved.
Drawings
Fig. 1 is a schematic diagram of a camera provided by an embodiment of the present disclosure;
fig. 2 is a flow chart of a method of smoke detection provided by an embodiment of the present disclosure;
fig. 3 is a flow chart of a method of smoke detection provided by an embodiment of the present disclosure;
FIG. 4 is a schematic diagram of a preset classification model provided by an embodiment of the present disclosure;
fig. 5 is a schematic structural diagram of smoke detection provided by the embodiment of the present disclosure;
fig. 6 is a schematic structural diagram of a camera provided in an embodiment of the present disclosure.
Detailed Description
To make the objects, technical solutions and advantages of the present disclosure more apparent, embodiments of the present disclosure will be described in detail with reference to the accompanying drawings.
The embodiment of the disclosure provides a smoke detection method, and an execution subject of the method can be a camera or a detection device. The camera can be any one of cameras which can be used for shooting video frame images, such as cameras with multiple forms of heavy-load pan-tilt cameras, ball machines, gun cameras and the like. The detection device may be a computer, a server, etc.
The camera may be provided with a processor, a memory and a transceiver, the processor may be configured to perform processing of smoke detection and control processing of captured video frame images, the memory may be configured to store data required and generated in a smoke detection process, and the transceiver may be configured to receive and transmit data.
The detection device may be provided with a processor, a memory and a transceiver, the processor may be configured to perform the processing of smoke detection, the memory may be configured to store data required and generated during smoke detection, and the transceiver may be configured to receive and transmit data.
It should be noted that when the execution subject is a detection device, the detection device may establish data connection with the camera, and the camera may send a captured video frame image to the detection device for smoke detection by the detection device.
In this embodiment, a camera is taken as an execution subject to perform detailed description of the scheme, other situations are similar to the above, and the detailed description is omitted here.
Before implementation, an application scenario and related concepts of the embodiments of the present disclosure are first introduced:
the smoke is an early expression of the fire, so that whether the fire exists or not can be judged through the smoke, and the fire can be further avoided. The method can be mainly applied to forest fire prevention monitoring, straw combustion monitoring and pollutant emission monitoring.
Ball machine: as shown in fig. 1, the camera is called a dome camera, which is a representative of the development of modern television monitoring, integrates multiple functions such as a color integrated camera, a pan-tilt, a decoder, a protective cover and the like, is convenient to install, simple to use but powerful, and is widely applied to monitoring in an open area. The ball machine can automatically rotate, and the shooting range is wide.
Gun locking: is one of monitoring CCD (charge coupled device) cameras. As shown in fig. 1, the bolt is generally rectangular in appearance, the front surface of the bolt is provided with a C/CS lens interface, and the bolt does not include a lens. The so-called bolt is mainly distinguished from the exterior and the lens mounting interface. The bolt cannot be automatically rotated after being installed.
Fixing a camera: a camera that cannot be rotated.
Rotating the camera: a camera that can rotate about one axis as opposed to a stationary camera.
As shown in fig. 2, an embodiment of the present disclosure provides a method for smoke detection, and taking a camera as a fixed camera as an example, an execution flow of the method may be as follows:
The target video frame image is any video frame image shot by the camera. The suspected smoke area refers to an area where smoke is likely to be present.
In implementation, a worker may install a camera in an area that needs to be monitored, after the camera is successfully installed, a video frame image may be captured, and each time one video frame image is captured, the video frame image may be stored, and whether a suspected smoke area exists in the target video frame image may be determined according to the one video frame image (which may be referred to as a target video frame image in the following process), and if it is determined that the suspected smoke area exists in the target video frame image, the following processes of step 202 to step 203 are performed.
The captured video frame image may be a captured image in a cycle (for example, 1 second) or may be a video frame image obtained by continuously capturing images and acquiring the captured video frame image from the captured video.
Alternatively, the process of determining the suspected smoke region may be as follows:
and inputting the data of the RGB channels of the target video frame image into a second preset classification model to obtain the probability of the suspected smoke area and the probability of the suspected smoke area in the target video frame image. And determining that the suspected smoke area exists in the target video frame image according to the probability of the suspected smoke area and the probability of the suspected smoke area not existing.
In implementation, a second preset classification model may be obtained through pre-training, where the second preset classification model includes a convolutional layer, a pooling layer, and a fully-connected layer, the convolutional layer may be used to extract image depth features, the pooling layer may be used to perform dimension reduction processing, and the fully-connected layer is used to output the position, category, and confidence of the category of the target frame. The input of the second preset classification model is data of RGB three channels of the target video frame image, image features are extracted through the convolution layer, dimension reduction processing is carried out on the image features through the pooling layer, the position, the confidence coefficient and the category of the target frame are output through the full connection layer, and the category comprises a suspected smoke area and a non-suspected smoke area.
A second preset classification model can be obtained, the data of the RGB channel of the target video frame image is input to the second preset classification model, and the output is the position of the suspected smoke region, the confidence of the suspected smoke region (the sum of the confidence of the suspected smoke region and the confidence of the suspected smoke region is equal to 1), and the confidence is the same as the probability, so that the probability of the suspected smoke region and the probability of the suspected smoke region can be obtained. And then determining that the suspected smoke area exists in the target video frame image according to the probability of the suspected smoke area and the probability of the suspected smoke area not existing.
Optionally, in order to filter false alarms occurring in a short time, it may be determined that a suspected smoke region exists in the target video frame image based on a video frame image temporally adjacent to the target video frame image, and the corresponding processing may be as follows:
and if the probability of the suspected smoke area is larger than or equal to the probability of the suspected smoke area, acquiring a second preset number of video frame images shot at the target position, wherein the second preset number of video frame images refer to video frame images on the left side and video frame images on the right side which are adjacent to the target video frame image in terms of time. And if the characteristics of the suspected smoke areas in the second preset number of video frame images are associated with the characteristics of the suspected smoke areas in the target video frame images, determining that the suspected smoke areas exist in the target video frame images, wherein the characteristics are one or more of position characteristics, shape characteristics, color characteristics and motion characteristics.
Wherein the second preset number may be preset and stored in the camera, e.g. the second preset number may be 4, etc. The position characteristic refers to the position of a suspected smoke area in a video frame image, the shape characteristic refers to the shape of the suspected smoke area, and the color characteristic refers to the color of the suspected smoke area.
In an implementation, the camera may determine the size of the probability of the suspected smoke area and the probability of the suspected smoke area, and if the probability of the suspected smoke area is greater than or equal to the probability of the suspected smoke area, a second preset number of video frame images captured at the target position may be obtained, where the second preset number of video frame images refers to a left video frame image and a right video frame image that are temporally adjacent to the target video frame image, and the number of the left video frame images is generally the same as the number of the right frame images.
The camera may then extract features of the suspected smoke regions in the second preset number of video frame images and extract features of the suspected smoke regions in the target video frame image, the features being one or more of location features, shape features and color features.
And then judging whether the characteristics of the suspected smoke areas in the second preset number of video frame images are associated with the characteristics of the suspected smoke areas in the target video frame images, and if so, determining that the suspected smoke areas exist in the target video frame images.
In addition, if the probability of the suspected smoke area is smaller than the probability of the suspected smoke area, it is indicated that the suspected smoke area does not exist generally, and the subsequent processing may not be performed, that is, the second preset number of video frame images shot at the target position are not acquired.
It should be noted that the left video frame image refers to a video frame image captured before the target video frame image, and the right video frame image refers to a video frame image captured after the target video frame image. In addition, when the judgment features are associated, if the position information of the suspected smoke areas in the target video frame image is the same as the position information of the suspected smoke areas in the second preset number of video frame images (the distance may also be smaller than a first preset value (such as 5 cm), determining that the features of the suspected smoke areas in the second preset number of video frame images are associated with the features of the suspected smoke areas in the target video frame image. If the color information of the suspected smoke area in the target video frame image is the same as the color information of the suspected smoke area in the second preset number of video frame images (or the similarity is higher than a second preset numerical value (such as 90%), determining that the features of the suspected smoke area in the second preset number of video frame images are associated with the features of the suspected smoke area in the target video frame image. If the shape information of the suspected smoke area in the target video frame image is the same as the shape information of the suspected smoke area in the second preset number of video frame images (or the similarity is higher than a third preset value (such as 90%), determining that the features of the suspected smoke area in the second preset number of video frame images are associated with the features of the suspected smoke area in the target video frame image. If the plurality of characteristics are included and all the plurality of characteristics satisfy the association, it may be determined that the characteristics of the suspected smoke areas in the second preset number of video frame images are associated with the characteristics of the suspected smoke areas in the target video frame image.
For example, the features include position information and shape information, and if the position information of the suspected smoke area in the target video frame image is the same as the position information of the suspected smoke area in the second preset number of video frame images and the position information of the suspected smoke area in the target video frame image is the same as the shape information of the suspected smoke area in the second preset number of video frame images, it may be determined that the features of the suspected smoke area in the second preset number of video frame images are associated with the features of the suspected smoke area in the target video frame image.
Therefore, the video frame images adjacent to the target video frame image are used for judging whether the suspected smoke area exists or not, so that the suspected smoke area can be determined more accurately.
In step 202, a first preset number of video frame images continuously shot at a target position after the target video frame image is shot are obtained.
And the target position is the position of the camera when the target video frame image is shot. The first preset number may be preset and stored in the camera, such as 16.
In an implementation, the camera may acquire a first preset number of video frame images consecutively photographed at a target position after photographing a target video frame image.
The remote terminal device refers to a terminal device of a monitored worker.
In implementation, the camera may detect a first preset number of video frame images, determine whether a smoke region exists in a coverage area of the camera, and if it is determined that the smoke region exists, may send a preset smoke notification signal to the remote terminal device. Therefore, after the far-end terminal receives the smoke prompting signal sent by the camera, the smoke prompting signal can be played, so that the staff can see that smoke exists in the monitoring area of the camera.
Optionally, the process of detecting the video frame image may be as follows:
inputting data of RGB channels of a first preset number of video frame images into a first preset classification model to obtain a first probability of a camera coverage area having a smoke area and a second probability of the camera coverage area having no smoke area, if the first probability is greater than or equal to the second probability, determining that the camera coverage area has the smoke area, and if the first probability is less than the second probability, determining that the camera coverage area has no smoke area.
In implementation, a first preset classification model can be trained in advance, the first classification model comprises a convolution layer, a pooling layer and a full-connection layer, the convolution layer can be used for extracting image depth features, the pooling layer can be used for dimension reduction processing, and the full-connection layer can be used for outputting target confidence and categories. The input of the first classification model is data of RGB three channels of a video frame image, image features are extracted through a convolution layer, dimension reduction processing is carried out on the image features through a pooling layer, and the output categories and the confidence degrees of the categories of the full connection layer are output, wherein the categories comprise a smoke existence region and a smoke nonexistence region.
The method comprises the steps of obtaining a first preset classification model, inputting data of RGB channels of a first preset number of video frame images into the first preset classification model, and outputting confidence coefficients of areas with smoke and areas without smoke, wherein the confidence coefficient of the areas with smoke is the same as the probability of the areas with smoke, and the confidence coefficient of the areas without smoke is the same as the probability of the areas without smoke, so that a first probability of the areas with smoke and a second probability of the areas without smoke can be obtained. The first probability and the second probability can then be judged, and if the first probability is greater than or equal to the second probability, the camera coverage area is determined to have a smoke area, and if the first probability is less than the second probability, the camera coverage area is determined to have no smoke area.
Optionally, in order to enable the staff to know what area has smoke, and send location information to the terminal, the corresponding processing may be as follows:
and controlling to send the position information of the smoke area and the first preset number of video frame images to the remote terminal equipment.
In implementation, if it is determined that there is a smoke region, the location information of the smoke region itself may be acquired, the location information may be determined as the location information of the smoke region, and a first preset number of video frame images photographed at a target location may be acquired, and then the location information of the smoke region and the first preset number of video frame images will be transmitted to the remote terminal device. In this way, the worker can know what area has smoke, and can judge whether smoke is really present or not based on the video frame image.
In addition, when the position information of the smoke area is sent to the far-end terminal device, the far-end terminal device can also carry the video image shot when the position information is sent.
In addition, when the position information is not carried, the identification of the camera can be carried, so that the position of the smoke area can be determined based on the identification of the camera.
Another embodiment of the present disclosure provides that the camera is a rotating camera, and as shown in fig. 3, a flow diagram of the method for smoke detection may be as follows:
In implementation, the processing procedure of step 301 is identical to the processing procedure of step 201, and is not described herein again.
Optionally, the process of detecting the video frame image may be as follows:
inputting data of RGB channels of a first preset number of video frame images into a first preset classification model to obtain a first probability of smoke areas and a second probability of smoke areas, if the first probability is larger than or equal to the second probability, determining that smoke areas exist, and if the first probability is smaller than the second probability, determining that smoke areas do not exist.
In implementation, the processing procedure is identical to that in fig. 2, and is not described here again.
Optionally, in order to filter false alarms occurring in a short time, it may be determined that a suspected smoke region exists in the target video frame image based on a video frame image adjacent to the target video frame image, and the corresponding processing may be as follows:
and if the probability of the suspected smoke area is larger than or equal to the probability of the suspected smoke area, acquiring a second preset number of video frame images shot at the target position, wherein the second preset number of video frame images refer to video frame images on the left side and video frame images on the right side which are adjacent to the target video frame image in terms of time. And if the characteristics of the suspected smoke areas in the second preset number of video frame images are associated with the characteristics of the suspected smoke areas in the target video frame images, determining that the suspected smoke areas exist in the target video frame images, wherein the characteristics are one or more of position characteristics, shape characteristics, color characteristics and motion characteristics.
In implementation, the processing procedure is identical to that in fig. 2, and is not described here again.
In the implementation, if the camera is rotated, the rotation may be stopped every time the camera is rotated by a certain angle, and the video frame image is captured, and the captured video frame image is stored in correspondence with the current position information, where the position information may be information such as the angle of rotation. For example, the camera rotates clockwise, the camera has an initial position point, the initial position point is 0 degree each time the camera passes, the initial position point of the camera is 0 degree, the camera rotates 60 degrees, the target video frame image is shot, the position point with the target position of 60 degrees, the 60 degrees and the target video frame image can be stored correspondingly, and the shooting time point can be stored.
In this way, when it is determined that the suspected smoke area exists in the target video frame image, the angle corresponding to the target video frame image can be acquired, and then the camera is controlled to rotate to the target position based on the angle. For example, when the target video frame image is captured at 60 degrees, the camera may be controlled to rotate to 60 degrees with respect to the initial position point.
In the implementation, the processing procedure of step 303 is identical to the processing procedure of step 202, and is not described herein again.
And 304, detecting a first preset number of video frame images, and if a smoke area exists in the camera coverage area, controlling to send a preset smoke prompt signal to the remote terminal equipment.
In the implementation, the processing procedure of step 304 is identical to the processing procedure of step 203, and is not described herein again.
Optionally, the process of detecting the video frame image may be as follows:
inputting data of RGB channels of a first preset number of video frame images into a first preset classification model to obtain a first probability of a camera coverage area having a smoke area and a second probability of the camera coverage area having no smoke area, if the first probability is greater than or equal to the second probability, determining that the camera coverage area has the smoke area, and if the first probability is less than the second probability, determining that the camera coverage area has no smoke area.
In implementation, a first preset classification model can be trained in advance, the first classification model comprises a convolution layer, a pooling layer and a full-connection layer, the convolution layer can be used for extracting image depth features, the pooling layer can be used for dimension reduction processing, and the full-connection layer is used for outputting a target confidence coefficient and a category. The input of the first classification model is data of RGB three channels of a video frame image, image features are extracted through a convolution layer, dimension reduction processing is carried out on the image features through a pooling layer, target confidence coefficient and categories are output through a full connection layer, and the categories comprise a smoke existence region and a smoke nonexistence region.
A first preset classification model can be obtained, the data of the RGB channels of the first preset number of video frame images is input to the first preset classification model, and the output is a first probability of the existence of the smoke region and a second probability of the absence of the smoke region. The first probability and the second probability can then be judged, and if the first probability is greater than or equal to the second probability, the camera coverage area is determined to have a smoke area, and if the first probability is less than the second probability, the camera coverage area is determined to have no smoke area.
In addition, if there is no smoke region, the camera may be controlled to continue rotating, and the corresponding process may be as follows:
and if the camera coverage area does not have a smoke area, controlling the rotary camera to continue rotating.
In an implementation, if the camera coverage area does not have a smoke area, the rotating camera may be controlled to continue to rotate.
Optionally, in order to enable the staff to know what area has smoke, and send location information to the terminal, the corresponding processing may be as follows:
and controlling to send the position information of the smoke area and the first preset number of video frame images to the remote terminal equipment.
In implementation, the processing procedure is identical to that in fig. 2, and is not described here again.
It should be noted that, the above description is made by taking a camera as an execution subject, and if the detection device is taken as the execution subject, the difference is only that the detection device acquires a captured image from the camera, and the detection device controls the camera to rotate.
In addition, it should be noted that the first preset classification model and the second preset classification model are convolutional neural network algorithms obtained through pre-training, as shown in fig. 4, for the first preset classification model, the first preset classification model may include a convolutional layer, a pooling layer, and a full connection layer, the convolutional layer may be used for extracting image depth features, the pooling layer may be used for performing dimensionality reduction, and the full connection layer may be used for outputting confidence and class of the class, during training, an initial classification model may be constructed, then data of three channels RGB of a plurality of video frame images is used as input, parameter values of each parameter to be trained included in the initial classification model are trained, and the parameter values of the parameter to be trained are substituted into the initial classification model to obtain the first preset classification model. In addition, the second predetermined classification model also includes a convolution layer, a pooling layer and a full link layer, and the training mode is the same as that of the first predetermined classification model, which is not described herein again.
In the embodiment of the disclosure, during smoke detection, a suspected smoke area in a target video frame image can be determined according to the target video frame image, then a first preset number of video frame images continuously shot at a target position after the target video frame image is shot are obtained, wherein the target position is a position where a camera is located when the target video frame image is shot, then the first preset number of video frame images are detected, and if the camera covers the area, a preset smoke prompt signal is controlled to be sent to a remote terminal device. In this way, since the determination is made twice after the presence of the suspected smoke region is determined, rather than being based solely on the empirical threshold, the accuracy of smoke detection can be improved.
Based on the same technical concept, the embodiment of the present disclosure also provides a smoke detection device, as shown in fig. 5, the smoke detection device includes:
a determining module 510, configured to determine, according to a target video frame image, that a suspected smoke area exists in the target video frame image;
an obtaining module 520, configured to obtain a first preset number of video frame images continuously captured at the target position after the target video frame image is captured, where the target position is a position where a camera is located when the target video frame image is captured;
a control module 530, configured to detect the first preset number of video frame images, and if a smoke region exists in the camera coverage region, control to send a preset smoke prompt signal to a remote terminal device.
Optionally, the camera is a rotary camera;
the control module 530 is further configured to:
and controlling the rotary camera to rotate to the target position before a first preset number of video frame images continuously shot at the target position after the target video frame image is shot are obtained.
Optionally, the control module 530 is further configured to:
and if the camera coverage area does not have a smoke area, controlling the rotating camera to continue rotating.
Optionally, the control module 530 is configured to:
inputting data of RGB channels of the first preset number of video frame images into a first preset classification model to obtain a first probability that a smog area exists in a camera coverage area and a second probability that the smog area does not exist in the camera coverage area;
determining that a smoke region exists in the camera coverage area if the first probability is greater than or equal to a second probability, and determining that a smoke region does not exist in the camera coverage area if the first probability is less than the second probability.
Optionally, the control module 530 is further configured to:
and controlling to send the position information of the smoke area and the first preset number of video frame images to the remote terminal equipment.
Optionally, the determining module 510 is configured to:
inputting data of RGB channels of the target video frame image into a second preset classification model to obtain the probability of a suspected smoke area and the probability of the suspected smoke area in the target video frame image;
and determining that the suspected smoke area exists in the target video frame image according to the probability of the suspected smoke area and the probability of the suspected smoke area not existing.
Optionally, the determining module 510 is configured to:
if the probability of the suspected smoke area is larger than or equal to the probability of the suspected smoke area, acquiring a second preset number of video frame images shot at the target position, wherein the second preset number of video frame images refer to video frame images on the left side and video frame images on the right side of the video frame images on the left side which are adjacent to the target video frame image in terms of time;
determining that a suspected smoke region exists in the target video frame image if the features of the suspected smoke region in the second preset number of video frame images are associated with the features of the suspected smoke region in the target video frame image, wherein the features are one or more of position features, shape features and color features.
In the embodiment of the disclosure, during smoke detection, a suspected smoke area in a target video frame image can be determined according to the target video frame image, then a first preset number of video frame images continuously shot at a target position after the target video frame image is shot are obtained, wherein the target position is a position where a camera is located when the target video frame image is shot, then the first preset number of video frame images are detected, and if the camera covers the area, a preset smoke prompt signal is controlled to be sent to a remote terminal device. In this way, since the determination is made twice after the presence of the suspected smoke region is determined, rather than being based solely on the empirical threshold, the accuracy of smoke detection can be improved.
It should be noted that: in the device for detecting smoke according to the above embodiment, when detecting smoke, only the division of the functional modules is described as an example, and in practical applications, the functions may be distributed by different functional modules as needed, that is, the internal structure of the device for detecting smoke is divided into different functional modules to complete all or part of the functions described above. In addition, the smoke detection device provided by the above embodiment and the smoke detection method embodiment belong to the same concept, and the specific implementation process thereof is described in the method embodiment and is not described herein again.
Fig. 6 is a schematic structural diagram of a camera 600 according to an embodiment of the present invention, where the camera 600 may generate a relatively large difference due to different configurations or performances, and may include one or more processors (CPUs) 601 and one or more memories 602, where at least one instruction is stored in the memory 602, and the at least one instruction is loaded and executed by the processor 601 to implement the method steps of smoke detection.
The present disclosure also provides an apparatus for smoke detection comprising a processor and a memory; the memory is used for storing at least one instruction; the processor is configured to execute at least one instruction stored in the memory to implement the method steps of smoke detection.
The present disclosure also provides a computer-readable storage medium having stored therein at least one instruction which, when executed by a processor, performs method steps for smoke detection.
It will be understood by those skilled in the art that all or part of the steps for implementing the above embodiments may be implemented by hardware, or may be implemented by a program instructing relevant hardware, where the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, etc.
The above description is only exemplary of the present disclosure and is not intended to limit the present disclosure, so that any modification, equivalent replacement, or improvement made within the spirit and principle of the present disclosure should be included in the scope of the present disclosure.
Claims (16)
1. A method of smoke detection, the method comprising:
determining a suspected smoke area in a target video frame image according to the target video frame image;
acquiring a first preset number of video frame images continuously shot at the target position after the target video frame image is shot, wherein the target position is the position of a camera when the target video frame image is shot;
and detecting the first preset number of video frame images, and if the camera coverage area has a smoke area, controlling to send a preset smoke prompt signal to a remote terminal device.
2. The method of claim 1, wherein the camera is a rotary camera;
the acquiring, before a first preset number of video frame images continuously captured at the target position after the target video frame image is captured, further includes:
and controlling the rotating camera to rotate to the target position.
3. The method of claim 2, further comprising:
and if the camera coverage area does not have a smoke area, controlling the rotating camera to continue rotating.
4. The method according to any one of claims 1 to 3, wherein said detecting the first preset number of video frame images comprises:
inputting data of RGB channels of the first preset number of video frame images into a first preset classification model to obtain a first probability that a smog area exists in a camera coverage area and a second probability that the smog area does not exist in the camera coverage area;
determining that a smoke region is present in the camera coverage area if the first probability is greater than or equal to a second probability, and determining that a smoke region is not present in the camera coverage area if the first probability is less than a second probability.
5. The method of any of claims 1 to 3, further comprising:
and controlling to send the position information of the smoke area and the first preset number of video frame images to the remote terminal equipment.
6. The method according to any one of claims 1 to 3, wherein the determining, from the target video frame image, that the suspected smoke region exists in the target video frame image comprises:
inputting data of RGB channels of the target video frame image into a second preset classification model to obtain the probability of a suspected smoke area and the probability of the suspected smoke area in the target video frame image;
and determining that the suspected smoke area exists in the target video frame image according to the probability of the suspected smoke area and the probability of the suspected smoke area not existing.
7. The method of claim 6, wherein said determining that a suspected smoke region is present in the target video frame image based on the probability of the suspected smoke region and the probability of the suspected smoke region not being present comprises:
if the probability of the suspected smoke area is larger than or equal to the probability of the suspected smoke area, acquiring a second preset number of video frame images shot at the target position, wherein the second preset number of video frame images refer to video frame images on the left side and video frame images on the right side of the video frame images on the left side which are adjacent to the target video frame image in terms of time;
determining that a suspected smoke region exists in the target video frame image if the features of the suspected smoke region in the second preset number of video frame images are associated with the features of the suspected smoke region in the target video frame image, wherein the features are one or more of position features, shape features and color features.
8. A smoke detection apparatus, the apparatus comprising:
the determining module is used for determining that a suspected smoke area exists in a target video frame image according to the target video frame image;
the acquisition module is used for acquiring a first preset number of video frame images which are continuously shot at the target position after the target video frame image is shot, wherein the target position is the position of a camera when the target video frame image is shot;
and the control module is used for detecting the first preset number of video frame images, and controlling to send a preset smoke prompt signal to the remote terminal equipment if the camera coverage area has a smoke area.
9. The apparatus of claim 8, wherein the camera is a rotary camera;
the control module is further configured to:
and controlling the rotary camera to rotate to the target position before a first preset number of video frame images continuously shot at the target position after the target video frame image is shot are obtained.
10. The apparatus of claim 9, wherein the control module is further configured to:
and if the camera coverage area does not have a smoke area, controlling the rotating camera to continue rotating.
11. The apparatus of any one of claims 8 to 10, wherein the control module is configured to:
inputting data of RGB channels of the first preset number of video frame images into a first preset classification model to obtain a first probability that a smog area exists in a camera coverage area and a second probability that the smog area does not exist in the camera coverage area;
determining that a smoke region is present in the camera coverage area if the first probability is greater than or equal to a second probability, and determining that a smoke region is not present in the camera coverage area if the first probability is less than a second probability.
12. The apparatus of any one of claims 8 to 10, wherein the control module is further configured to:
and controlling to send the position information of the smoke area and the first preset number of video frame images to the remote terminal equipment.
13. The apparatus of any one of claims 8 to 10, wherein the determining module is configured to:
inputting data of RGB channels of the target video frame image into a second preset classification model to obtain the probability of a suspected smoke area and the probability of the suspected smoke area in the target video frame image;
and determining that the suspected smoke area exists in the target video frame image according to the probability of the suspected smoke area and the probability of the suspected smoke area not existing.
14. The apparatus of claim 13, wherein the determining module is configured to:
if the probability of the suspected smoke area is larger than or equal to the probability of the suspected smoke area, acquiring a second preset number of video frame images shot at the target position, wherein the second preset number of video frame images refer to video frame images on the left side and video frame images on the right side of the video frame images on the left side which are adjacent to the target video frame image in terms of time;
determining that a suspected smoke region exists in the target video frame image if the features of the suspected smoke region in the second preset number of video frame images are associated with the features of the suspected smoke region in the target video frame image, wherein the features are one or more of position features, shape features and color features.
15. An apparatus for smoke detection comprising a processor and a memory; the memory is used for storing at least one instruction; the processor, configured to execute at least one instruction stored on the memory to implement the method steps of any of claims 1-7.
16. A computer-readable storage medium having stored therein at least one instruction which, when executed by a processor, implements the method steps of any of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910087220.5A CN111488772B (en) | 2019-01-29 | 2019-01-29 | Method and device for detecting smoke |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910087220.5A CN111488772B (en) | 2019-01-29 | 2019-01-29 | Method and device for detecting smoke |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111488772A true CN111488772A (en) | 2020-08-04 |
| CN111488772B CN111488772B (en) | 2023-09-22 |
Family
ID=71812325
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910087220.5A Active CN111488772B (en) | 2019-01-29 | 2019-01-29 | Method and device for detecting smoke |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111488772B (en) |
Citations (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| AU2006202296A1 (en) * | 2006-05-30 | 2007-12-20 | Canon Kabushiki Kaisha | Digital video storage |
| CN101587622A (en) * | 2009-06-18 | 2009-11-25 | 任芳 | Forest rocket detection and recognition methods and equipment based on video image intelligent analysis |
| CN101738394A (en) * | 2009-02-11 | 2010-06-16 | 北京智安邦科技有限公司 | Method and system for detecting indoor smog |
| CN101872526A (en) * | 2010-06-01 | 2010-10-27 | 重庆市海普软件产业有限公司 | Smoke and fire intelligent identification method based on programmable photographing technology |
| CN101916380A (en) * | 2010-09-08 | 2010-12-15 | 大连古野软件有限公司 | Video-based device and method for detecting smog |
| CN102136059A (en) * | 2011-03-03 | 2011-07-27 | 苏州市慧视通讯科技有限公司 | Video- analysis-base smoke detecting method |
| CN102567722A (en) * | 2012-01-17 | 2012-07-11 | 大连民族学院 | Early-stage smoke detection method based on codebook model and multiple features |
| CN103049460A (en) * | 2011-10-17 | 2013-04-17 | 天津市亚安科技股份有限公司 | Video surveillance scene information classifying and storing method and search method |
| CN103116746A (en) * | 2013-03-08 | 2013-05-22 | 中国科学技术大学 | Video flame detecting method based on multi-feature fusion technology |
| US20140211002A1 (en) * | 2013-01-31 | 2014-07-31 | Qnap Systems, Inc. | Video Object Detection System Based on Region Transition, and Related Method |
| CN104794486A (en) * | 2015-04-10 | 2015-07-22 | 电子科技大学 | Video smoke detecting method based on multi-feature fusion |
| KR20160048240A (en) * | 2014-10-23 | 2016-05-04 | 공주대학교 산학협력단 | Smoke detection method using density distribution characteristics of smoke |
| KR101620989B1 (en) * | 2015-02-12 | 2016-05-16 | 계명대학교 산학협력단 | Method of detecting fire automatically from video including night images and system for performing the same |
| JP2016110258A (en) * | 2014-12-03 | 2016-06-20 | 能美防災株式会社 | Smoke detection device and smoke detection method |
| CN105809120A (en) * | 2016-03-03 | 2016-07-27 | 魏晓峰 | Video image analyzing method and system |
| CN106228150A (en) * | 2016-08-05 | 2016-12-14 | 南京工程学院 | Smog detection method based on video image |
| CN106339664A (en) * | 2016-08-10 | 2017-01-18 | 杭州电子科技大学 | Color mixing model and multi-feature combination-based video smoke detection method |
| CN107749067A (en) * | 2017-09-13 | 2018-03-02 | 华侨大学 | Fire hazard smoke detecting method based on kinetic characteristic and convolutional neural networks |
| CN108537726A (en) * | 2017-03-03 | 2018-09-14 | 杭州海康威视数字技术股份有限公司 | A kind of method of track up, equipment and unmanned plane |
| CN108985192A (en) * | 2018-06-29 | 2018-12-11 | 东南大学 | A kind of video smoke recognition methods based on multitask depth convolutional neural networks |
| CN110322659A (en) * | 2019-06-21 | 2019-10-11 | 江西洪都航空工业集团有限责任公司 | A kind of smog detection method |
-
2019
- 2019-01-29 CN CN201910087220.5A patent/CN111488772B/en active Active
Patent Citations (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| AU2006202296A1 (en) * | 2006-05-30 | 2007-12-20 | Canon Kabushiki Kaisha | Digital video storage |
| CN101738394A (en) * | 2009-02-11 | 2010-06-16 | 北京智安邦科技有限公司 | Method and system for detecting indoor smog |
| CN101587622A (en) * | 2009-06-18 | 2009-11-25 | 任芳 | Forest rocket detection and recognition methods and equipment based on video image intelligent analysis |
| CN101872526A (en) * | 2010-06-01 | 2010-10-27 | 重庆市海普软件产业有限公司 | Smoke and fire intelligent identification method based on programmable photographing technology |
| CN101916380A (en) * | 2010-09-08 | 2010-12-15 | 大连古野软件有限公司 | Video-based device and method for detecting smog |
| CN102136059A (en) * | 2011-03-03 | 2011-07-27 | 苏州市慧视通讯科技有限公司 | Video- analysis-base smoke detecting method |
| CN103049460A (en) * | 2011-10-17 | 2013-04-17 | 天津市亚安科技股份有限公司 | Video surveillance scene information classifying and storing method and search method |
| CN102567722A (en) * | 2012-01-17 | 2012-07-11 | 大连民族学院 | Early-stage smoke detection method based on codebook model and multiple features |
| US20140211002A1 (en) * | 2013-01-31 | 2014-07-31 | Qnap Systems, Inc. | Video Object Detection System Based on Region Transition, and Related Method |
| CN103116746A (en) * | 2013-03-08 | 2013-05-22 | 中国科学技术大学 | Video flame detecting method based on multi-feature fusion technology |
| KR20160048240A (en) * | 2014-10-23 | 2016-05-04 | 공주대학교 산학협력단 | Smoke detection method using density distribution characteristics of smoke |
| JP2016110258A (en) * | 2014-12-03 | 2016-06-20 | 能美防災株式会社 | Smoke detection device and smoke detection method |
| KR101620989B1 (en) * | 2015-02-12 | 2016-05-16 | 계명대학교 산학협력단 | Method of detecting fire automatically from video including night images and system for performing the same |
| CN104794486A (en) * | 2015-04-10 | 2015-07-22 | 电子科技大学 | Video smoke detecting method based on multi-feature fusion |
| CN105809120A (en) * | 2016-03-03 | 2016-07-27 | 魏晓峰 | Video image analyzing method and system |
| CN106228150A (en) * | 2016-08-05 | 2016-12-14 | 南京工程学院 | Smog detection method based on video image |
| CN106339664A (en) * | 2016-08-10 | 2017-01-18 | 杭州电子科技大学 | Color mixing model and multi-feature combination-based video smoke detection method |
| CN108537726A (en) * | 2017-03-03 | 2018-09-14 | 杭州海康威视数字技术股份有限公司 | A kind of method of track up, equipment and unmanned plane |
| CN107749067A (en) * | 2017-09-13 | 2018-03-02 | 华侨大学 | Fire hazard smoke detecting method based on kinetic characteristic and convolutional neural networks |
| CN108985192A (en) * | 2018-06-29 | 2018-12-11 | 东南大学 | A kind of video smoke recognition methods based on multitask depth convolutional neural networks |
| CN110322659A (en) * | 2019-06-21 | 2019-10-11 | 江西洪都航空工业集团有限责任公司 | A kind of smog detection method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111488772B (en) | 2023-09-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109040709B (en) | Video monitoring method and device, monitoring server and video monitoring system | |
| WO2020057355A1 (en) | Three-dimensional modeling method and device | |
| CN102833478B (en) | Fault-tolerant background model | |
| KR101423916B1 (en) | Method and apparatus for recognizing the plural number of faces | |
| CN108198199B (en) | Moving object tracking method, moving object tracking device and electronic equipment | |
| CN109981943A (en) | Picture pick-up device, image processing equipment, control method and storage medium | |
| WO2020057353A1 (en) | Object tracking method based on high-speed ball, monitoring server, and video monitoring system | |
| CN109905641B (en) | Target monitoring method, device, equipment and system | |
| KR101442669B1 (en) | Method and apparatus for criminal acts distinction using intelligent object sensing | |
| CN110402456B (en) | Abnormality detection system, abnormality detection method, and storage medium | |
| CN114401354A (en) | Intelligent control method and system for over-the-horizon monitoring of offshore ship | |
| CN110399823B (en) | Subject tracking method and apparatus, electronic device, and computer-readable storage medium | |
| CN111488772B (en) | Method and device for detecting smoke | |
| CN116246200A (en) | Screen display information candid photographing detection method and system based on visual identification | |
| JP2011087214A (en) | Photographing system, method and program | |
| KR100982342B1 (en) | Intelligent security system and operating method thereof | |
| WO2021189429A1 (en) | Image photographing method and device, movable platform, and storage medium | |
| CN114495395A (en) | Human shape detection method, monitoring and early warning method, device and system | |
| TWI448976B (en) | Ultra-wide-angle imaging method and system using the same | |
| CN110874906B (en) | Method and device for starting defense deploying function | |
| KR101311728B1 (en) | System and the method thereof for sensing the face of intruder | |
| AU2009230796A1 (en) | Location-based brightness transfer function | |
| JP7480841B2 (en) | Event management method, event management device, system and program | |
| CN112215045A (en) | Living body detection method and device | |
| WO2024092526A1 (en) | Control device for camera, and camera |
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 |