CN112396121A - Smoke image classification method based on neural network - Google Patents

Smoke image classification method based on neural network Download PDF

Info

Publication number
CN112396121A
CN112396121A CN202011377912.2A CN202011377912A CN112396121A CN 112396121 A CN112396121 A CN 112396121A CN 202011377912 A CN202011377912 A CN 202011377912A CN 112396121 A CN112396121 A CN 112396121A
Authority
CN
China
Prior art keywords
image
smoke image
smoke
target
noise
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
Application number
CN202011377912.2A
Other languages
Chinese (zh)
Other versions
CN112396121B (en
Inventor
徐俊生
张俊
陈洋
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Huazheng Tomorrow Information Technology Co ltd
Original Assignee
Beijing Huazheng Tomorrow Information Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing Huazheng Tomorrow Information Technology Co ltd filed Critical Beijing Huazheng Tomorrow Information Technology Co ltd
Priority to CN202011377912.2A priority Critical patent/CN112396121B/en
Publication of CN112396121A publication Critical patent/CN112396121A/en
Application granted granted Critical
Publication of CN112396121B publication Critical patent/CN112396121B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • 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
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
    • 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
    • G06V10/30—Noise filtering
    • G—PHYSICS
    • G08—SIGNALLING
    • G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
    • G08B17/00—Fire alarms; Alarms responsive to explosion
    • G08B17/12—Actuation by presence of radiation or particles, e.g. of infrared radiation or of ions
    • G08B17/125—Actuation by presence of radiation or particles, e.g. of infrared radiation or of ions by using a video camera to detect fire or smoke

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Multimedia (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Biomedical Technology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Business, Economics & Management (AREA)
  • Emergency Management (AREA)
  • Health & Medical Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a smoke image classification method based on a neural network, and belongs to the technical field of image processing. Obtaining a smoke image to be classified; extracting pixel values corresponding to all image points on the smoke image to be classified; processing the smoke image to be classified according to the pixel value to obtain a target smoke image; acquiring characteristic parameters corresponding to each image point on the target smoke image; constructing a characteristic parameter set corresponding to the target smoke image according to the characteristic parameters corresponding to the image points; acquiring reference characteristic parameters of a characteristic parameter set; determining a target feature type corresponding to the reference feature parameter through a preset neural network model; the target characteristic category is used as a classification result of the smoke image to be classified, the smoke image to be classified is processed through the pixel value of the image point, the processed target smoke image is classified, the smoke image is classified more comprehensively, and therefore the accuracy of smoke image classification is improved.

Description

Smoke image classification method based on neural network
Technical Field
The invention relates to the technical field of image processing, in particular to a smoke image classification method based on a neural network.
Background
The task of fire detection is critical to personnel safety. Because the smog diffusion rate is fast and do not have more suitable position to lay the sensor under the open environment, lead to the tradition to detect the mode based on the sensor can't accurately detect conflagration smog, what adopt at present mostly classifies the smog image of conflagration according to textural features to reach the purpose that smog detected. However, the smoke shape, color and texture are very different, and the classification mode adopted by the target is limited, so that the classification accuracy of the smoke image is low.
The above is only for the purpose of assisting understanding of the technical aspects of the present invention, and does not represent an admission that the above is prior art.
Disclosure of Invention
The invention mainly aims to provide a smoke image classification method based on a neural network, and aims to solve the technical problem of low smoke image classification accuracy in the prior art.
In order to achieve the above object, the present invention provides a smoke image classification method based on a neural network, which is characterized in that the smoke image classification method based on the neural network comprises:
acquiring a smoke image to be classified;
extracting pixel values corresponding to all image points on the smoke image to be classified;
processing the smoke image to be classified according to the pixel value to obtain a target smoke image;
acquiring characteristic parameters corresponding to each image point on the target smoke image;
constructing a characteristic parameter set corresponding to the target smoke image according to the characteristic parameters corresponding to the image points;
acquiring reference characteristic parameters of the characteristic parameter set;
determining a target feature type corresponding to the reference feature parameter through a preset neural network model;
and taking the target feature category as a classification result of the smoke image to be classified.
Optionally, the step of acquiring the smoke image to be classified includes:
acquiring an original smoke image and a preset noise sample corresponding to the original smoke image;
inputting the preset noise sample into the original smoke image to obtain a noise smoke image;
processing the noise smoke image to obtain a processed noise smoke image;
and taking the processed noise smoke image as a smoke image to be classified.
Optionally, the step of processing the noise smoke image to obtain a processed noise smoke image includes:
acquiring a gray-scale noise smoke image corresponding to the noise smoke image;
carrying out region division on the gray noise smoke image to obtain a plurality of noise regions corresponding to the gray noise smoke image;
and filtering the noise area according to a preset two-dimensional template to obtain a processed noise smoke image.
Optionally, the step of filtering the noise region according to a preset two-dimensional template to obtain a processed noise smoke image includes:
acquiring a template central point of a preset two-dimensional template and an area central point of the noise area;
moving the preset two-dimensional template on the noise area in sequence;
when the template central point is coincident with the area central point, acquiring a current noise area corresponding to the area central point and a plurality of image points corresponding to the current noise area;
acquiring a gray value corresponding to each image point;
sorting the gray values corresponding to the image points, and adjusting the gray value of the central point of the region according to the sorting result to obtain a target region;
and taking the image formed by the target area as a processed noise smoke image.
Optionally, the step of processing the smoke image to be classified according to the pixel value to obtain a target smoke image includes:
detecting the pixel value according to a first preset pixel threshold value, taking an image point corresponding to the pixel value larger than the first preset pixel threshold value as a target image point, and taking an image point corresponding to the pixel value smaller than or equal to the first preset pixel threshold value as a background image point;
and respectively adjusting the pixel value of the target image point and the pixel value of the background image point to obtain a target smoke image.
Optionally, after the step of detecting the pixel value according to a first preset pixel threshold, taking an image point corresponding to a pixel value greater than the first preset pixel threshold as a target image point, and taking an image point corresponding to a pixel value less than or equal to the first preset pixel threshold as a background image point, the method further includes:
detecting the target image point;
if the target image point has a second preset pixel threshold, taking a target image point corresponding to a pixel value which is greater than the first preset pixel threshold and less than or equal to the second preset pixel threshold as a first target image point, and taking a target image point corresponding to a pixel value which is greater than the second preset pixel threshold as a second target image point, wherein the second preset pixel threshold is greater than the first preset pixel threshold;
correspondingly, the step of respectively adjusting the pixel value of the target image point and the pixel value of the background image point to obtain the target smoke image comprises:
and respectively adjusting the pixel value of the first target image point, the pixel value of the second target image point and the pixel value of the background image point based on a preset function to obtain a target smoke image.
Optionally, the step of obtaining the reference feature parameter of the feature parameter set includes:
acquiring characteristic parameters in each characteristic parameter set, and dividing the same characteristic parameters into a group to obtain a plurality of parameter sets;
and acquiring the quantity of the characteristic parameters in each parameter group, and taking the characteristic parameter with the maximum quantity in each parameter group as the reference characteristic parameter of the characteristic parameter group.
In addition, in order to achieve the above object, the present invention further provides a smoke image classification device based on a neural network, including:
the acquisition module is used for acquiring a smoke image to be classified;
the extraction module is used for extracting pixel values corresponding to all image points on the smoke image to be classified;
the processing module is used for processing the smoke image to be classified according to the pixel value to obtain a target smoke image;
the classification module is used for acquiring characteristic parameters corresponding to each image point on the target smoke image;
the classification module is further configured to construct a feature parameter set corresponding to the target smoke image according to the feature parameters corresponding to the image points;
the classification module is further configured to obtain a reference feature parameter of the feature parameter set;
the classification module is further used for determining a target feature category corresponding to the reference feature parameter through a preset neural network model;
the classification module is further used for taking the target feature category as a classification result of the to-be-classified smoke image.
In addition, in order to achieve the above object, the present invention further provides a smoke image classification device based on a neural network, including: a memory, a processor and a neural network based smoke image classification program stored on the memory and executable on the processor, the neural network based smoke image classification program being configured to implement the steps of the neural network based smoke image classification method as described above.
Furthermore, to achieve the above object, the present invention also proposes a storage medium having stored thereon a neural network-based smoke image classification program, which when executed by a processor, implements the steps of the neural network-based smoke image classification method as described above.
Obtaining a smoke image to be classified; extracting pixel values corresponding to all image points on the smoke image to be classified; processing the smoke image to be classified according to the pixel value to obtain a target smoke image; acquiring characteristic parameters corresponding to each image point on the target smoke image; constructing a characteristic parameter set corresponding to the target smoke image according to the characteristic parameters corresponding to the image points; acquiring reference characteristic parameters of a characteristic parameter set; determining a target feature type corresponding to the reference feature parameter through a preset neural network model; the target characteristic category is used as a classification result of the smoke image to be classified, the smoke image to be classified is processed through the pixel value of the image point, the processed target smoke image is classified, the smoke image is classified more comprehensively, and therefore the accuracy of smoke image classification is improved.
Drawings
Fig. 1 is a schematic structural diagram of a neural network-based smoke image classification device of a hardware operating environment according to an embodiment of the present invention;
FIG. 2 is a schematic flow chart of a smoke image classification method based on a neural network according to a first embodiment of the present invention;
FIG. 3 is a schematic flow chart of a smoke image classification method based on a neural network according to a second embodiment of the present invention;
FIG. 4 is a schematic diagram of noise region filtering in the smoke image classification method based on neural network according to the present invention;
FIG. 5 is a schematic flow chart of a smoke image classification method based on a neural network according to a third embodiment of the present invention;
fig. 6 is a block diagram of a smoke image classification device based on a neural network according to a first embodiment of the present invention.
The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Referring to fig. 1, fig. 1 is a schematic structural diagram of a smoke image classification device based on a neural network in a hardware operating environment according to an embodiment of the present invention.
As shown in fig. 1, the neural network-based smoke image classification apparatus may include: a processor 1001, such as a Central Processing Unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Wherein a communication bus 1002 is used to enable connective communication between these components. The user interface 1003 may include a Display screen (Display), an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a WIreless interface (e.g., a WIreless-FIdelity (WI-FI) interface). The Memory 1005 may be a Random Access Memory (RAM) Memory, or may be a Non-Volatile Memory (NVM), such as a disk Memory. The memory 1005 may alternatively be a storage device separate from the processor 1001.
Those skilled in the art will appreciate that the configuration shown in figure 1 does not constitute a limitation of a neural network-based smoke image classification apparatus and may include more or fewer components than those shown, or some components in combination, or a different arrangement of components.
As shown in fig. 1, a memory 1005, which is a storage medium, may include therein an operating system, a network communication module, a user interface module, and a neural network-based smoke image classification program.
In the neural network-based smoke image classification device shown in fig. 1, the network interface 1004 is mainly used for data communication with a network server; the user interface 1003 is mainly used for data interaction with a user; the processor 1001 and the memory 1005 of the smoke image classification device based on the neural network of the present invention may be disposed in the smoke image classification device based on the neural network, and the smoke image classification device based on the neural network calls the smoke image classification program based on the neural network stored in the memory 1005 through the processor 1001 and executes the smoke image classification method based on the neural network provided by the embodiment of the present invention.
An embodiment of the present invention provides a smoke image classification method based on a neural network, and referring to fig. 2, fig. 2 is a schematic flow diagram of a first embodiment of a smoke image classification method based on a neural network according to the present invention.
In this embodiment, the smoke image classification method based on the neural network includes the following steps:
step S10: and acquiring a smoke image to be classified.
It should be noted that, in this embodiment, the execution subject may be a computer for image acquisition and image processing, or may be other terminal devices, which is not limited in this embodiment.
It should be noted that, because the structure and composition of the smoke image are complex, and the influence of factors such as the image texture definition and the image resolution on the digitized smoke image is large, the original smoke image needs to be screened and processed, so that a clear and characteristic smoke image is obtained.
Step S20: and extracting pixel values corresponding to all image points on the smoke image to be classified.
In this embodiment, the image is composed of a plurality of image points, and after the image of the smoke to be classified is acquired, the pixel values of the image points on the image of the smoke to be classified are extracted.
In a specific implementation, the pixel values may be extracted according to the following manner, for example, a similarity comparison is performed between the smoke image to be classified and an image composed of the red, green and blue channels, and the pixel values of the image points on the smoke image to be classified are determined according to the result of the similarity comparison, for example, the similarity between the image point a on the smoke image to be classified and the image point B on the image composed of the red, green and blue channels is 90%, and the pixel value of the image point B is X, and the pixel value of the image point a may be obtained as 90% X.
Step S30: and processing the smoke image to be classified according to the pixel value to obtain a target smoke image.
It should be noted that, the characteristics displayed on the image by the different pixel values of the image point are different, and not all the image points on the smoke image to be classified contain effective characteristics, and in order to make the classification more accurate, the image point containing the effective characteristics and the image point not containing the effective characteristics need to be distinguished, in this embodiment, the smoke image to be classified is processed according to the pixel values of the image points, so that the characteristics of the image point containing the effective characteristics are more prominent, and the processed smoke image to be classified is the target smoke image.
Step S40: and acquiring characteristic parameters corresponding to each image point on the target smoke image.
In this embodiment, after the target smoke image is obtained, the classification result of the smoke image to be classified can be obtained by classifying the target smoke image.
In a specific implementation, first, a pixel value of each image point on the target smoke image is obtained, after the pixel value of each image point is obtained, one image point is arbitrarily selected as a current image point, other image points are obtained within a preset range of the current image point, the pixel value of the current image point is compared with the pixel values of the other image points, a corresponding binary code is assigned to the current image point according to a comparison result, for example, if the pixel value of the current image point is greater than the pixel values of the other pixel points, the binary code assigned to the current image point is a, if the pixel values of the other image points are greater than the pixel value of the current image point, the binary code assigned to the current image point is B, if the number of the other image points whose pixel values are greater than the pixel value of the current image point is 2, the binary code assigned to the current image point is C, and the like, the specific binary code distribution condition is set according to the actual condition, after the binary code of the current image point is obtained, the characteristic parameter corresponding to the binary code can be searched from the mapping relation table, and the characteristic parameter of each image point can be obtained.
Step S50: and constructing a characteristic parameter set corresponding to the target smoke image according to the characteristic parameters corresponding to the image points.
In a specific implementation, the feature parameters corresponding to each image point are randomly selected, in this embodiment, the number of times of selecting the feature parameters is not less than one, one feature parameter set can be obtained in each selection, and the number of the feature parameters is the same in each selection, and the same feature parameters can be repeatedly selected in one selection, for example, two feature parameters are selected, the feature parameters selected in the first selection are B, C, B, S and a, and the feature parameters selected in the second selection are G, F, U, W and E, so that X, Y two feature parameter sets can be obtained, where X is (B, C, B, S, a), and Y is (G, F, U, W, E).
Step S60: and acquiring the reference characteristic parameters of the characteristic parameter set.
In the present embodiment, the reference feature parameters are feature parameters that can represent the entire feature parameter set, for example, the feature parameter set G is (a, B, C), the feature parameter C is a reference feature parameter of the feature parameter set G, and specifically,the process of acquiring the characteristic parameters and the reference characteristic parameters specifically comprises the following steps: acquiring characteristic parameters in each characteristic parameter set, and dividing the same characteristic parameters into a group to obtain a plurality of parameter sets; and acquiring the quantity of the characteristic parameters in each parameter group, and taking the characteristic parameter with the maximum quantity in each parameter group as the reference characteristic parameter of the characteristic parameter group. E.g. feature parameter set Z ═ (a, D, E), the feature parameter set is divided into T1、T2And T3Three groups, T1＝(A，A，A)、T2(D) and T3When T is expressed as (E), T is known1The number of medium characteristic parameters is 3, T2And T3The number of the medium characteristic parameters is 1, T1The number of medium characteristic parameters is the largest, so T is1The characteristic parameter in (b) is used as a reference characteristic parameter of the characteristic parameter set Z, that is, the characteristic parameter of the characteristic parameter set Z is a.
Step S70: and determining the target characteristic category corresponding to the reference characteristic parameter through a preset neural network model.
In a specific implementation, after obtaining the reference feature parameters of each feature parameter set, the target feature class corresponding to each reference feature parameter may be determined by using a preset neural network model, where the preset neural network model may be a forward neural network model or a generalized regression neural network model, and the present embodiment is not limited thereto. The preset weight values corresponding to the reference characteristic parameters can be obtained through a preset neural network model, the preset weight values can be set according to actual conditions, and are not limited in this embodiment, and the reference characteristic parameters are calculated according to the preset weight values, so that the target characteristic parameters can be obtained, for example, the preset weight value of the reference characteristic parameter O is 0.2, the preset weight value of the reference characteristic parameter P is 0.5, the preset weight value of the reference characteristic parameter Q is 0.3, the target characteristic parameter S is obtained as 0.2O +0.5P +0.3Q, then, the target characteristic category corresponding to the target characteristic parameter S is searched from a preset mapping relation table, the preset mapping relation table can be changed as required, and the limitation in this embodiment is not limited.
Step S80: and taking the target feature category as a classification result of the smoke image to be classified.
It is easy to understand that the essence of classifying the image is to identify the class corresponding to the image feature, and the target feature class is the classification result of the smoke image to be classified.
In the embodiment, a smoke image to be classified is obtained; extracting pixel values corresponding to all image points on the smoke image to be classified; processing the smoke image to be classified according to the pixel value to obtain a target smoke image; acquiring characteristic parameters corresponding to each image point on the target smoke image; constructing a characteristic parameter set corresponding to the target smoke image according to the characteristic parameters corresponding to the image points; acquiring reference characteristic parameters of a characteristic parameter set; determining a target feature type corresponding to the reference feature parameter through a preset neural network model; the target characteristic category is used as a classification result of the smoke image to be classified, the smoke image to be classified is processed through the pixel value of the image point, the processed target smoke image is classified, the smoke image is classified more comprehensively, and therefore the accuracy of smoke image classification is improved.
Referring to fig. 3, fig. 3 is a flowchart illustrating a smoke image classification method based on a neural network according to a second embodiment of the present invention.
Based on the first embodiment, in the smoke image classification method based on a neural network of the present embodiment, the step S10 includes:
step S101: the method comprises the steps of obtaining an original smoke image and a preset noise sample corresponding to the original smoke image.
It should be noted that the original smoke image is an unprocessed smoke image, the image types corresponding to different original smoke images are different, a preset noise sample corresponding to the original pathological smoke image can be determined according to the specific image type corresponding to the original smoke image, the preset noise sample can be gaussian white noise or rayleigh noise, and the embodiment is not limited.
Step S102: and inputting the preset noise sample into the original smoke image to obtain a noise smoke image.
In specific implementation, after a preset noise smoke image corresponding to the original smoke image is obtained, a preset noise sample is input into the original smoke image, that is, the preset noise sample and the original smoke image are fused, so that the noise smoke image containing the preset noise sample is obtained.
Step S103: and processing the noise smoke image to obtain a processed noise smoke image.
It is easy to understand that after obtaining the noise smoke image, the image noise in the noise smoke image needs to be removed, so as to obtain the smoke image not containing the image noise, i.e. the processed noise smoke image.
In a specific implementation, the step of processing the noise smoke image to obtain a processed noise smoke image includes: acquiring a gray-scale noise smoke image corresponding to the noise smoke image; carrying out region division on the gray noise smoke image to obtain a plurality of noise regions corresponding to the gray noise smoke image; and filtering the noise area according to a preset two-dimensional template to obtain a processed noise smoke image.
It should be noted that, processing the noise smoke image requires first obtaining a gray-scale noise smoke image corresponding to the noise smoke image, where the gray-scale noise smoke image is obtained by performing gray-scale processing on the noise smoke image, and then performing region division on the gray-scale noise smoke image to obtain a plurality of noise regions, and the noise smoke image may be subjected to region division in an equal division manner, or may be subjected to region division according to a random area size, and may be set according to an actual situation.
In this embodiment, after obtaining a plurality of noise regions, filtering each noise region in turn according to a preset two-dimensional template, and removing image noise in each noise region through filtering to obtain a processed noise smoke image, the size of the preset two-dimensional template may be the same as the size of the noise region, for example, 4 × 4 or 7 × 7.
In a specific implementation, the step of filtering the noise region according to a preset two-dimensional template to obtain a processed noise smoke image includes: acquiring a template central point of a preset two-dimensional template and an area central point of the noise area; moving the preset two-dimensional template on the noise area in sequence; when the template central point is coincident with the area central point, acquiring a current noise area corresponding to the area central point and a plurality of image points corresponding to the current noise area; acquiring a gray value corresponding to each image point; sorting the gray values corresponding to the image points, and adjusting the gray value of the central point of the region according to the sorting result to obtain a target region; and taking the image formed by the target area as a processed noise smoke image.
In the specific implementation, a template center point of a preset two-dimensional template and a region center point of each noise region are respectively obtained, the preset two-dimensional template is moved on each noise region in sequence, it is easy to understand that after the preset two-dimensional template filters the current noise region, the filtering of the next noise region is performed, during the moving process of the preset two-dimensional template, the position of the template center point is detected in real time, when the coincidence of the template center point and the region center of the noise region is detected, the preset two-dimensional template stops moving, the current noise region corresponding to the region center point at the moment and a plurality of image points corresponding to the current noise region are obtained, the gray values corresponding to the image points are obtained, then the obtained gray values are sorted according to the magnitude sequence, the gray value median value in the middle size can be obtained, and the gray value of the region center point is adjusted to the median value, for convenience of understanding, the noise region after the gray value adjustment is the target region, and for example, fig. 4 is taken as an example to illustrate, as shown in fig. 4, the gray values of the image points in the current noise region are respectively 12, 33, 30, 2, 5, 6, 9, 1 and 22, the gray values are arranged in order of magnitude to obtain a gray value median of 12, the gray value of the center point of the region in the current noise region is adjusted from 5 to 12 to obtain the target region, and the above steps are repeated to obtain a plurality of target regions, wherein an image formed by the plurality of target regions is the processed noise smoke image.
Step S104: and taking the processed noise smoke image as a smoke image to be classified.
In this embodiment, the processed noise smoke image does not include the input preset noise sample and the noise existing in the original smoke image, and the processed noise smoke image is the smoke image to be classified.
In this embodiment, an original smoke image and a preset noise sample corresponding to the original smoke image are obtained; inputting the preset noise sample into the original smoke image to obtain a noise smoke image; processing the noise smoke image to obtain a processed noise smoke image; and taking the processed noise smoke image as a smoke image to be classified, inputting a preset noise sample into the original smoke image to obtain a noise smoke image, and processing the noise smoke image, so that the obtained smoke image to be classified does not contain image noise, and the classification of the smoke image is more accurate.
Referring to fig. 5, fig. 5 is a flowchart illustrating a smoke image classification method based on a neural network according to a third embodiment of the present invention.
Based on the first embodiment or the second embodiment, a third embodiment of the smoke image classification method based on the neural network is provided.
Taking the first embodiment as an example for explanation, the step S30 includes:
step S301: and detecting the pixel value according to a first preset pixel threshold value, taking an image point corresponding to the pixel value larger than the first preset pixel threshold value as a target image point, and taking an image point corresponding to the pixel value smaller than or equal to the first preset pixel threshold value as a background image point.
In this embodiment, the first preset pixel threshold may be set according to the situation, and the first preset pixel threshold is used as the pixel value detection standard of the image point, so as to obtain an image point corresponding to a pixel value greater than the first preset pixel threshold and an image point corresponding to a pixel value less than or equal to the first preset pixel threshold.
It should be noted that detecting the pixel value according to the first preset pixel threshold may be understood as performing a pre-screening on the image point, and in an actual situation, a second preset pixel threshold may exist in a part of the target image points, and a secondary screening on the target image points according to the second preset pixel threshold is required, specifically, after the target image points are obtained, the target image points are continuously detected, if it is detected that the second preset pixel threshold exists in the target image points, the target image points corresponding to the pixel values which are greater than the first preset pixel threshold and less than or equal to the second preset pixel threshold are taken as the first target image points, and the target image points corresponding to the pixel values which are greater than the second preset pixel threshold are taken as the second target image points, where the second preset pixel threshold is greater than the first preset pixel threshold, and the first target image points and the second image points are all points related to the effective features, the number of valid features in the first target image point is smaller than the number of valid features in the second image point.
Step S302: and respectively adjusting the pixel value of the target image point and the pixel value of the background image point to obtain a target smoke image.
In this embodiment, the pixel values of the target image point and the background image point are respectively adjusted based on a preset function, where the target image point includes a first image point and a second image point, and the preset function of this embodiment is as shown in formula 1.
Figure BDA0002808200990000121
Wherein H is a predetermined function corresponding to the pixel value, X is the pixel value of the image point, and T1Is a first predetermined pixel threshold, T2And the second preset pixel threshold value is set, C is the pixel value of the background image point, B is the pixel value of the first target image point, A is the pixel value of the second target image point, and the specific values of A > B > C, A, B and C can be set according to the actual situation, and the pixel value of each image point is adjusted according to the formula 1, so that the target smoke image can be obtained.
In this embodiment, the pixel value is detected according to a first preset pixel threshold, an image point corresponding to a pixel value larger than the first preset pixel threshold is used as a target image point, an image point corresponding to a pixel value smaller than or equal to the first preset pixel threshold is used as a background image point, a first target image point and a second target image point are determined from the target image points according to a second preset pixel threshold, and the pixel value of the first target image point, the pixel value of the second target image point and the pixel value of the background image point are respectively adjusted based on a preset function to obtain a target smoke image, so that the effective feature of the target smoke image is more prominent, and the accuracy of smoke image classification based on a neural network is improved.
Referring to fig. 6, fig. 6 is a block diagram of a smoke image classification device based on a neural network according to a first embodiment of the present invention.
As shown in fig. 6, the smoke image classification apparatus based on the neural network according to the embodiment of the present invention includes:
and the acquisition module 10 is used for acquiring the smoke image to be classified.
It should be noted that, because the structure and composition of the smoke image are complex, and the influence of factors such as the image texture definition and the image resolution on the digitized smoke image is large, the original smoke image needs to be screened and processed, so that a clear and characteristic smoke image is obtained.
And the extracting module 20 is configured to extract pixel values corresponding to each image point on the smoke image to be classified.
In this embodiment, the image is composed of a plurality of image points, and after the image of the smoke to be classified is acquired, the pixel values of the image points on the image of the smoke to be classified are extracted.
In a specific implementation, the pixel values may be extracted according to the following manner, for example, a similarity comparison is performed between the smoke image to be classified and an image composed of the red, green and blue channels, and the pixel values of the image points on the smoke image to be classified are determined according to the result of the similarity comparison, for example, the similarity between the image point a on the smoke image to be classified and the image point B on the image composed of the red, green and blue channels is 90%, and the pixel value of the image point B is X, and the pixel value of the image point a may be obtained as 90% X.
And the processing module 30 is configured to process the smoke image to be classified according to the pixel value to obtain a target smoke image.
It should be noted that, the characteristics displayed on the image by the different pixel values of the image point are different, and not all the image points on the smoke image to be classified contain effective characteristics, and in order to make the classification more accurate, the image point containing the effective characteristics and the image point not containing the effective characteristics need to be distinguished, in this embodiment, the smoke image to be classified is processed according to the pixel values of the image points, so that the characteristics of the image point containing the effective characteristics are more prominent, and the processed smoke image to be classified is the target smoke image.
And the classification module 40 is configured to obtain a characteristic parameter corresponding to each image point on the target smoke image.
In this embodiment, after the target smoke image is obtained, the classification result of the smoke image to be classified can be obtained by classifying the target smoke image.
In a specific implementation, first, a pixel value of each image point on the target smoke image is obtained, after the pixel value of each image point is obtained, one image point is arbitrarily selected as a current image point, other image points are obtained within a preset range of the current image point, the pixel value of the current image point is compared with the pixel values of the other image points, a corresponding binary code is assigned to the current image point according to a comparison result, for example, if the pixel value of the current image point is greater than the pixel values of the other pixel points, the binary code assigned to the current image point is a, if the pixel values of the other image points are greater than the pixel value of the current image point, the binary code assigned to the current image point is B, if the number of the other image points whose pixel values are greater than the pixel value of the current image point is 2, the binary code assigned to the current image point is C, and the like, the specific binary code distribution condition is set according to the actual condition, after the binary code of the current image point is obtained, the characteristic parameter corresponding to the binary code can be searched from the mapping relation table, and the characteristic parameter of each image point can be obtained.
The classification module 40 is further configured to construct a feature parameter set corresponding to the target smoke image according to the feature parameters corresponding to the image points.
In a specific implementation, the feature parameters corresponding to each image point are randomly selected, in this embodiment, the number of times of selecting the feature parameters is not less than one, one feature parameter set can be obtained in each selection, and the number of the feature parameters is the same in each selection, and the same feature parameters can be repeatedly selected in one selection, for example, two feature parameters are selected, the feature parameters selected in the first selection are B, C, B, S and a, and the feature parameters selected in the second selection are G, F, U, W and E, so that X, Y two feature parameter sets can be obtained, where X is (B, C, B, S, a), and Y is (G, F, U, W, E).
The classification module 40 is further configured to obtain a reference feature parameter of the feature parameter set.
In practical applications, a representative feature parameter is used as a feature parameter of the entire feature parameter set, where the reference feature parameter in this embodiment is a feature parameter that can represent the entire feature parameter set, for example, the feature parameter set G is (a, B, C), and the feature parameter C is a reference feature parameter of the feature parameter set G, and specifically, the process of acquiring the feature parameter and the reference feature parameter is specifically: acquiring characteristic parameters in each characteristic parameter set, and dividing the same characteristic parameters into a group to obtain a plurality of parameter sets; and acquiring the quantity of the characteristic parameters in each parameter group, and taking the characteristic parameter with the maximum quantity in each parameter group as the reference characteristic parameter of the characteristic parameter group. E.g. feature parameter set Z ═ (a, D, E), the feature parameter set is divided into T1、T2And T3Three groups, T1＝(A，A，A)、T2(D) and T3When T is expressed as (E), T is known1The number of medium characteristic parameters is 3, T2And T3The number of the medium characteristic parameters is 1, T1The number of medium characteristic parameters is the largest, so T is1The characteristic parameter in (b) is used as a reference characteristic parameter of the characteristic parameter set Z, that is, the characteristic parameter of the characteristic parameter set Z is a.
The classification module 40 is further configured to determine a target feature class corresponding to the reference feature parameter through a preset neural network model.
In a specific implementation, after obtaining the reference feature parameters of each feature parameter set, the target feature class corresponding to each reference feature parameter may be determined by using a preset neural network model, where the preset neural network model may be a forward neural network model or a generalized regression neural network model, and the present embodiment is not limited thereto. The preset weight values corresponding to the reference characteristic parameters can be obtained through a preset neural network model, the preset weight values can be set according to actual conditions, and are not limited in this embodiment, and the reference characteristic parameters are calculated according to the preset weight values, so that the target characteristic parameters can be obtained, for example, the preset weight value of the reference characteristic parameter O is 0.2, the preset weight value of the reference characteristic parameter P is 0.5, the preset weight value of the reference characteristic parameter Q is 0.3, the target characteristic parameter S is obtained as 0.2O +0.5P +0.3Q, then, the target characteristic category corresponding to the target characteristic parameter S is searched from a preset mapping relation table, the preset mapping relation table can be changed as required, and the limitation in this embodiment is not limited.
The classification module 40 is further configured to use the target feature category as a classification result of the smoke image to be classified.
It is easy to understand that the essence of classifying the image is to identify the class corresponding to the image feature, and the target feature class is the classification result of the smoke image to be classified.
In the embodiment, a smoke image to be classified is obtained; extracting pixel values corresponding to all image points on the smoke image to be classified; processing the smoke image to be classified according to the pixel value to obtain a target smoke image; acquiring characteristic parameters corresponding to each image point on the target smoke image; constructing a characteristic parameter set corresponding to the target smoke image according to the characteristic parameters corresponding to the image points; acquiring reference characteristic parameters of a characteristic parameter set; determining a target feature type corresponding to the reference feature parameter through a preset neural network model; the target characteristic category is used as a classification result of the smoke image to be classified, the smoke image to be classified is processed through the pixel value of the image point, the processed target smoke image is classified, the smoke image is classified more comprehensively, and therefore the accuracy of smoke image classification is improved.
In an embodiment, the smoke image classification based on the neural network further includes a denoising module, configured to obtain an original smoke image and a preset noise sample corresponding to the original smoke image; inputting the preset noise sample into the original smoke image to obtain a noise smoke image; processing the noise smoke image to obtain a processed noise smoke image; and taking the processed noise smoke image as a smoke image to be classified.
In an embodiment, the denoising module is further configured to obtain a grayscale noise smoke image corresponding to the noise smoke image; carrying out region division on the gray noise smoke image to obtain a plurality of noise regions corresponding to the gray noise smoke image; and filtering the noise area according to a preset two-dimensional template to obtain a processed noise smoke image.
In an embodiment, the denoising module is further configured to obtain a template central point of a preset two-dimensional template and a region central point of the noise region; moving the preset two-dimensional template on the noise area in sequence; when the template central point is coincident with the area central point, acquiring a current noise area corresponding to the area central point and a plurality of image points corresponding to the current noise area; acquiring a gray value corresponding to each image point; sorting the gray values corresponding to the image points, and adjusting the gray value of the central point of the region according to the sorting result to obtain a target region; and taking the image formed by the target area as a processed noise smoke image.
In an embodiment, the processing module 30 is further configured to detect the pixel value according to a first preset pixel threshold, take an image point corresponding to a pixel value larger than the first preset pixel threshold as a target image point, and take an image point corresponding to a pixel value smaller than or equal to the first preset pixel threshold as a background image point; and respectively adjusting the pixel value of the target image point and the pixel value of the background image point to obtain a target smoke image.
In an embodiment, the processing module 30 is further configured to detect the target image point; if the target image point has a second preset pixel threshold, taking a target image point corresponding to a pixel value which is greater than the first preset pixel threshold and less than or equal to the second preset pixel threshold as a first target image point, and taking a target image point corresponding to a pixel value which is greater than the second preset pixel threshold as a second target image point, wherein the second preset pixel threshold is greater than the first preset pixel threshold; and respectively adjusting the pixel value of the first target image point, the pixel value of the second target image point and the pixel value of the background image point based on a preset function to obtain a target smoke image.
In an embodiment, the classification module 40 is further configured to obtain feature parameters in each feature parameter set, and divide the same feature parameters into a group to obtain a plurality of parameter groups; and acquiring the quantity of the characteristic parameters in each parameter group, and taking the characteristic parameter with the maximum quantity in each parameter group as the reference characteristic parameter of the characteristic parameter group.
It should be understood that the above is only an example, and the technical solution of the present invention is not limited in any way, and in a specific application, a person skilled in the art may set the technical solution as needed, and the present invention is not limited thereto.
It should be noted that the above-described work flows are only exemplary, and do not limit the scope of the present invention, and in practical applications, a person skilled in the art may select some or all of them to achieve the purpose of the solution of the embodiment according to actual needs, and the present invention is not limited herein.
In addition, the technical details that are not described in detail in this embodiment may be referred to a smoke image classification method based on a neural network provided in any embodiment of the present invention, and are not described herein again.
Further, it is to be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solution of the present invention or portions thereof that contribute to the prior art may be embodied in the form of a software product, where the computer software product is stored in a storage medium (e.g. Read Only Memory (ROM)/RAM, magnetic disk, optical disk), and includes several instructions for enabling a terminal device (e.g. a mobile phone, a computer, a server, or a network device) to execute the method according to the embodiments of the present invention.
The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.

Claims (10)

1. A smoke image classification method based on a neural network is characterized by comprising the following steps:
acquiring a smoke image to be classified;
extracting pixel values corresponding to all image points on the smoke image to be classified;
processing the smoke image to be classified according to the pixel value to obtain a target smoke image;
acquiring characteristic parameters corresponding to each image point on the target smoke image;
constructing a characteristic parameter set corresponding to the target smoke image according to the characteristic parameters corresponding to the image points;
acquiring reference characteristic parameters of the characteristic parameter set;
determining a target feature type corresponding to the reference feature parameter through a preset neural network model;
and taking the target feature category as a classification result of the smoke image to be classified.
2. The neural network-based smoke image classification method according to claim 1, wherein the step of obtaining the smoke image to be classified comprises:
acquiring an original smoke image and a preset noise sample corresponding to the original smoke image;
inputting the preset noise sample into the original smoke image to obtain a noise smoke image;
processing the noise smoke image to obtain a processed noise smoke image;
and taking the processed noise smoke image as a smoke image to be classified.
3. The neural network-based smoke image classification method according to claim 2, wherein the step of processing the noise smoke image to obtain a processed noise smoke image comprises:
acquiring a gray-scale noise smoke image corresponding to the noise smoke image;
carrying out region division on the gray noise smoke image to obtain a plurality of noise regions corresponding to the gray noise smoke image;
and filtering the noise area according to a preset two-dimensional template to obtain a processed noise smoke image.
4. The smoke image classification method based on the neural network as claimed in claim 3, wherein the step of filtering the noise region according to a preset two-dimensional template to obtain the processed noise smoke image comprises:
acquiring a template central point of a preset two-dimensional template and an area central point of the noise area;
moving the preset two-dimensional template on the noise area in sequence;
when the template central point is coincident with the area central point, acquiring a current noise area corresponding to the area central point and a plurality of image points corresponding to the current noise area;
acquiring a gray value corresponding to each image point;
sorting the gray values corresponding to the image points, and adjusting the gray value of the central point of the region according to the sorting result to obtain a target region;
and taking the image formed by the target area as a processed noise smoke image.
5. The neural network-based smoke image classification method according to any one of claims 1 to 4, wherein the step of processing the smoke image to be classified according to the pixel values to obtain a target smoke image comprises:
detecting the pixel value according to a first preset pixel threshold value, taking an image point corresponding to the pixel value larger than the first preset pixel threshold value as a target image point, and taking an image point corresponding to the pixel value smaller than or equal to the first preset pixel threshold value as a background image point;
and respectively adjusting the pixel value of the target image point and the pixel value of the background image point to obtain a target smoke image.
6. The neural network-based smoke image classification method according to claim 5, wherein said step of detecting said pixel values according to a first predetermined pixel threshold, wherein the image points corresponding to the pixel values greater than said first predetermined pixel threshold are regarded as target image points, and the image points corresponding to the pixel values less than or equal to said first predetermined pixel threshold are regarded as background image points, further comprises:
detecting the target image point;
if the target image point has a second preset pixel threshold, taking a target image point corresponding to a pixel value which is greater than the first preset pixel threshold and less than or equal to the second preset pixel threshold as a first target image point, and taking a target image point corresponding to a pixel value which is greater than the second preset pixel threshold as a second target image point, wherein the second preset pixel threshold is greater than the first preset pixel threshold;
correspondingly, the step of respectively adjusting the pixel value of the target image point and the pixel value of the background image point to obtain the target smoke image comprises:
and respectively adjusting the pixel value of the first target image point, the pixel value of the second target image point and the pixel value of the background image point based on a preset function to obtain a target smoke image.
7. The neural network-based smoke image classification method according to any one of claims 1 to 4, wherein said step of obtaining reference feature parameters of said set of feature parameters comprises:
acquiring characteristic parameters in each characteristic parameter set, and dividing the same characteristic parameters into a group to obtain a plurality of parameter sets;
and acquiring the quantity of the characteristic parameters in each parameter group, and taking the characteristic parameter with the maximum quantity in each parameter group as the reference characteristic parameter of the characteristic parameter group.
8. A neural network-based smoke image classification device, comprising:
the acquisition module is used for acquiring a smoke image to be classified;
the extraction module is used for extracting pixel values corresponding to all image points on the smoke image to be classified;
the processing module is used for processing the smoke image to be classified according to the pixel value to obtain a target smoke image;
the classification module is used for acquiring characteristic parameters corresponding to each image point on the target smoke image;
the classification module is further configured to construct a feature parameter set corresponding to the target smoke image according to the feature parameters corresponding to the image points;
the classification module is further configured to obtain a reference feature parameter of the feature parameter set;
the classification module is further used for determining a target feature category corresponding to the reference feature parameter through a preset neural network model;
the classification module is further used for taking the target feature category as a classification result of the to-be-classified smoke image.
9. The neural network-based smoke image classification device as claimed in claim 8, wherein the neural network-based smoke image classification device further comprises: a denoising module;
the de-noising module is used for acquiring an original smoke image and a preset noise sample corresponding to the original smoke image;
the denoising module is further configured to input the preset noise sample into the original smoke image to obtain a noise smoke image;
the denoising module is further used for processing the noise smoke image to obtain a processed noise smoke image;
and the denoising module is also used for taking the processed noise smoke image as a smoke image to be classified.
10. A neural network-based smoke image classification device, characterized by comprising: a memory, a processor and a neural network-based smoke image classification program stored on the memory and executable on the processor, the neural network-based smoke image classification program being configured to implement the steps of the neural network-based smoke image classification method as claimed in any one of claims 1 to 7.
CN202011377912.2A 2020-11-30 2020-11-30 Smoke image classification method based on neural network Active CN112396121B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011377912.2A CN112396121B (en) 2020-11-30 2020-11-30 Smoke image classification method based on neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011377912.2A CN112396121B (en) 2020-11-30 2020-11-30 Smoke image classification method based on neural network

Publications (2)

Publication Number Publication Date
CN112396121A true CN112396121A (en) 2021-02-23
CN112396121B CN112396121B (en) 2024-02-02

Family

ID=74605675

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011377912.2A Active CN112396121B (en) 2020-11-30 2020-11-30 Smoke image classification method based on neural network

Country Status (1)

Country Link
CN (1) CN112396121B (en)

Citations (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
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
CN104050478A (en) * 2014-07-09 2014-09-17 湖南大学 Smog detection method and system
CN105225235A (en) * 2015-09-18 2016-01-06 北京航空航天大学 A kind of video flame detecting method based on multispectral characteristic
CN105608443A (en) * 2016-01-22 2016-05-25 合肥工业大学 Multi-feature description and local decision weighting face identification method
JP2016110263A (en) * 2014-12-03 2016-06-20 能美防災株式会社 Smoke detection device and smoke detection method
CN106952438A (en) * 2017-04-19 2017-07-14 天津安平易视智能影像科技有限公司 A kind of fire alarm method based on video image
CN106997461A (en) * 2017-03-28 2017-08-01 浙江大华技术股份有限公司 A kind of firework detecting method and device
CN109165577A (en) * 2018-08-07 2019-01-08 东北大学 A kind of early stage forest fire detection method based on video image
CN109191495A (en) * 2018-07-17 2019-01-11 东南大学 Black smoke vehicle detection method based on self-organizing background subtraction model and multiple features fusion
CN109711345A (en) * 2018-12-27 2019-05-03 南京林业大学 A kind of flame image recognition methods, device and its storage medium
CN109886227A (en) * 2019-02-27 2019-06-14 哈尔滨工业大学 Inside fire video frequency identifying method based on multichannel convolutive neural network
CN110033040A (en) * 2019-04-12 2019-07-19 华南师范大学 A kind of flame identification method, system, medium and equipment

Patent Citations (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
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
CN104050478A (en) * 2014-07-09 2014-09-17 湖南大学 Smog detection method and system
JP2016110263A (en) * 2014-12-03 2016-06-20 能美防災株式会社 Smoke detection device and smoke detection method
CN105225235A (en) * 2015-09-18 2016-01-06 北京航空航天大学 A kind of video flame detecting method based on multispectral characteristic
CN105608443A (en) * 2016-01-22 2016-05-25 合肥工业大学 Multi-feature description and local decision weighting face identification method
CN106997461A (en) * 2017-03-28 2017-08-01 浙江大华技术股份有限公司 A kind of firework detecting method and device
CN106952438A (en) * 2017-04-19 2017-07-14 天津安平易视智能影像科技有限公司 A kind of fire alarm method based on video image
CN109191495A (en) * 2018-07-17 2019-01-11 东南大学 Black smoke vehicle detection method based on self-organizing background subtraction model and multiple features fusion
CN109165577A (en) * 2018-08-07 2019-01-08 东北大学 A kind of early stage forest fire detection method based on video image
CN109711345A (en) * 2018-12-27 2019-05-03 南京林业大学 A kind of flame image recognition methods, device and its storage medium
CN109886227A (en) * 2019-02-27 2019-06-14 哈尔滨工业大学 Inside fire video frequency identifying method based on multichannel convolutive neural network
CN110033040A (en) * 2019-04-12 2019-07-19 华南师范大学 A kind of flame identification method, system, medium and equipment

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
XIAOLIAN LI等: "Forest Fire Smoke Detection Using Back-Propagation Neural Network Based on MODIS Data", 《REMOTE SENSING》, pages 4473 - 4498 *
唐田田: "基于曲波域烟雾特征提取方法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》, no. 02, pages 138 - 1192 *
高云玲: "基于视频的火灾烟雾检测", 《中国优秀硕士学位论文全文数据库 信息科技辑》, no. 04, pages 138 - 847 *

Also Published As

Publication number Publication date
CN112396121B (en) 2024-02-02

Similar Documents

Publication Publication Date Title
US7103215B2 (en) Automated detection of pornographic images
WO2020062360A1 (en) Image fusion classification method and apparatus
CN110826632B (en) Image change detection method, device, equipment and computer readable storage medium
WO2007051963A1 (en) Image processing
CN111582359B (en) Image identification method and device, electronic equipment and medium
CN109903210B (en) Watermark removal method, watermark removal device and server
CN107563427A (en) The method and corresponding use that copyright for oil painting is identified
CN116385863A (en) Deep learning-based method and system for identifying microplastic in water body
CN111784665A (en) OCT image quality assessment method, system and device based on Fourier transform
CN111553241A (en) Method, device and equipment for rejecting mismatching points of palm print and storage medium
CN110874835A (en) Crop leaf disease resistance identification method and system, electronic equipment and storage medium
CN110569716A (en) Goods shelf image copying detection method
CN108171683A (en) A kind of method for cell count using software automatic identification
CN111815594A (en) Reinforcing steel bar detection method and related equipment and device
CN112396121A (en) Smoke image classification method based on neural network
CN111178398A (en) Method, system, storage medium and device for detecting tampering of image information of identity card
CN109543565B (en) Quantity determination method and device
WO2022167483A1 (en) System for clustering data points
CN114266643A (en) Enterprise mining method, device, equipment and storage medium based on fusion algorithm
CN113688810A (en) Target capturing method and system of edge device and related device
CN109670514B (en) Method, device, equipment and storage medium for identifying pore and skeleton of digital core
CN113709563B (en) Video cover selecting method and device, storage medium and electronic equipment
CN116245848B (en) Product defect detection method and related equipment
CN111144244B (en) Furniture wood category identification method and device
CN116433671B (en) Colloidal gold detection method, system and storage medium based on image recognition

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