CN115631158B - Coal detection method for carbon check - Google Patents

Coal detection method for carbon check Download PDF

Info

Publication number
CN115631158B
CN115631158B CN202211273060.1A CN202211273060A CN115631158B CN 115631158 B CN115631158 B CN 115631158B CN 202211273060 A CN202211273060 A CN 202211273060A CN 115631158 B CN115631158 B CN 115631158B
Authority
CN
China
Prior art keywords
image
coal
representing
hidden layer
neural network
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.)
Active
Application number
CN202211273060.1A
Other languages
Chinese (zh)
Other versions
CN115631158A (en
Inventor
王延敦
秦云松
吴鹏
王书诤
宋博
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.)
Central Carbon And Beijing Technology Co ltd
Original Assignee
Central Carbon And Beijing 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 Central Carbon And Beijing Technology Co ltd filed Critical Central Carbon And Beijing Technology Co ltd
Priority to CN202211273060.1A priority Critical patent/CN115631158B/en
Publication of CN115631158A publication Critical patent/CN115631158A/en
Application granted granted Critical
Publication of CN115631158B publication Critical patent/CN115631158B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • 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
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/20—Image enhancement or restoration using local operators
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/70—Denoising; Smoothing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/12—Edge-based segmentation
    • 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
    • G06T7/80—Analysis of captured images to determine intrinsic or extrinsic camera parameters, i.e. camera calibration
    • 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
    • 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
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/80—Management or planning
    • Y02P90/84—Greenhouse gas [GHG] management systems

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Multimedia (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Databases & Information Systems (AREA)
  • Medical Informatics (AREA)
  • Biomedical Technology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Quality & Reliability (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Molecular Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Investigating Or Analyzing Materials Using Thermal Means (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a coal detection method for carbon check, which relates to the technical field of machine vision and image processing and comprises the following steps: collecting images of coal to be detected; calibrating the sampling range of the image; setting a noise template; filtering the sampled image; expanding the filtered image according to a Taylor formula to obtain an edge characteristic image; thresholding the edge feature image; obtaining a gray scale compressed image according to the filtered image and the binary edge characteristic image; calculating a coal characteristic sample; establishing a prediction neural network model; training a prediction neural network model to obtain a consumption function; iteratively calculating a predicted neural network model by using the consumption function to obtain a trained predicted neural network model; and inputting the coal characteristic sample into the trained prediction neural network model to obtain an output value. According to the invention, the image data of the coal fuel is shot by the camera, the coal in the image is detected and reserved, and the heating value of the coal fuel can be automatically analyzed and predicted.

Description

Coal detection method for carbon check
Technical Field
The invention relates to the technical fields of machine vision and image processing, in particular to a coal detection method for carbon check.
Background
The emission of coal fuel is a main production emission source of chemical enterprises, and carbon emission data can be effectively predicted by measuring the calorific value of the coal fuel. From the chemical industry perspective, accurately accounting and reporting of greenhouse gas emissions and related data is a trend and requirement of industry management development.
At present, the problems of unclear accounting boundary, inaccurate accounting method, incomplete accounting data and the like commonly exist in the carbon checking process of chemical enterprises, and the problems are the energy conservation and emission reduction of popularization enterprises, the improvement of environmental protection production technology of the enterprises is encouraged, and the environmental protection is hindered. The traditional coal fuel calorific value determination method comprises an oxygen bomb method and an element analysis method, and is complex in operation steps, long in detection period and low in safety. With the development of machine vision technology, the detection of the calorific value of the coal can be realized based on the spectral image processing analysis method, and the method has the advantages of high analysis efficiency, simple equipment deployment, small influence on environment and manpower saving, is suitable for being used by chemical enterprises of different scales, and has wide application prospect.
Disclosure of Invention
The invention provides a coal detection method for carbon check, which is used for overcoming at least one technical problem in the prior art.
The embodiment of the invention provides a coal detection method for carbon check, which comprises the following steps:
acquiring images of the coal to be detected according to a preset frame rate by using a camera;
calibrating the sampling range of the image to obtain a sampled image
Figure 442135DEST_PATH_IMAGE001
；
Setting a noise template as according to the shape characteristics of the coal to be detected
Figure 465324DEST_PATH_IMAGE002
Wherein->
Figure 127249DEST_PATH_IMAGE003
Template for representing elliptic curve
Figure 344735DEST_PATH_IMAGE004
，
Figure 459322DEST_PATH_IMAGE005
Representing coordinates of pixels in the template with the center of the template as a reference;
filtering the sampled image by using the noise template to obtain a filtered image
Figure 363166DEST_PATH_IMAGE006
Wherein->
Figure 82861DEST_PATH_IMAGE007
Representing convolution operation,/->
Figure 159663DEST_PATH_IMAGE008
Representing pixel coordinates in the image;
expanding the filtered image according to a Taylor formula to obtain an edge characteristic image as
Figure 105754DEST_PATH_IMAGE009
Figure 893581DEST_PATH_IMAGE010
Wherein->
Figure 779366DEST_PATH_IMAGE011
For the derivative image,
Figure 853502DEST_PATH_IMAGE012
is a second derivative image;
thresholding the edge feature image to obtain a binary edge feature image
Figure 958992DEST_PATH_IMAGE013
Figure 550510DEST_PATH_IMAGE014
Wherein->
Figure 41535DEST_PATH_IMAGE015
Representing a threshold value;
obtaining a gray scale compressed image according to the filtered image and the binary edge feature image
Figure 538768DEST_PATH_IMAGE016
Wherein->
Figure 318506DEST_PATH_IMAGE017
Expressed in terms of
Figure 120239DEST_PATH_IMAGE018
The offset coordinates in the square neighborhood with the point as the center, W represents the side length of the neighborhood;
calculating a coal characteristic sample according to the gray scale compressed image and the binary edge characteristic image;
taking the coal characteristic sample as input, defining output as
Figure 934612DEST_PATH_IMAGE019
Figure 334238DEST_PATH_IMAGE020
Establishing a prediction neural network model; wherein (1)>
Figure 194747DEST_PATH_IMAGE021
Representing a third hidden layer->
Figure 128067DEST_PATH_IMAGE022
A linear weight representing the output of the element of the third hidden layer and the neural network model, ++>
Figure 875575DEST_PATH_IMAGE023
Representing a linear deviation +.>
Figure 931255DEST_PATH_IMAGE024
For activating a function, defined as +.>
Figure 265678DEST_PATH_IMAGE025
，
Figure 799428DEST_PATH_IMAGE026
Representing an activation function convergence parameter; the prediction neural network model comprises a first hidden layer, a second hidden layer and a third hidden layer, wherein the first hidden layer is
Figure 401442DEST_PATH_IMAGE027
Figure 628024DEST_PATH_IMAGE028
，
Figure 915654DEST_PATH_IMAGE029
Representation ofCoordinates of the first hidden layer->
Figure 784253DEST_PATH_IMAGE030
、
Figure 975194DEST_PATH_IMAGE031
Respectively indicate->
Figure 779202DEST_PATH_IMAGE032
、
Figure 836020DEST_PATH_IMAGE033
Linear weights of local neighborhoods of +.>
Figure 217630DEST_PATH_IMAGE034
Representing the linear deviation; the second hidden layer is->
Figure 512345DEST_PATH_IMAGE035
，
Figure 628199DEST_PATH_IMAGE036
Representing the coordinates of the second hidden layer, +.>
Figure 641155DEST_PATH_IMAGE037
Linear weights representing pixels of the first hidden layer and pixels of the second hidden layer, +.>
Figure 569665DEST_PATH_IMAGE038
Representing the linear deviation; the third hidden layer is
Figure 718887DEST_PATH_IMAGE039
Y represents the dimension of the third hidden layer, < >>
Figure 5643DEST_PATH_IMAGE040
Representing a linear weight +.>
Figure 912419DEST_PATH_IMAGE041
Representing the linear deviation;
by making said predictionsTraining a neural network model to obtain a consumption function of the predicted neural network model
Figure 192091DEST_PATH_IMAGE042
，
Figure 133502DEST_PATH_IMAGE043
Representing the thermal value of the training sample, +.>
Figure 30307DEST_PATH_IMAGE044
Representing the output value +.>
Figure 549013DEST_PATH_IMAGE045
Representing the control coefficient;
iteratively calculating the prediction neural network model by using the consumption function to obtain a trained prediction neural network model;
and inputting the coal characteristic sample into the trained predictive neural network model to obtain an output value.
Optionally, the calibrating the sampling range of the image specifically includes:
calibrating the sampling range of the image according to the movement linear speed of the conveyor belt, the field angle of the camera, the distance between the camera and the conveyor belt and the shooting period of the camera to obtain calibration parameters
Figure 242163DEST_PATH_IMAGE046
Wherein->
Figure 913447DEST_PATH_IMAGE047
Represents the linear speed of movement of the conveyor belt, +.>
Figure 666639DEST_PATH_IMAGE048
Representing the angle of view of the camera, +.>
Figure 672641DEST_PATH_IMAGE049
Indicating the distance of the camera from the conveyor belt, +.>
Figure 903902DEST_PATH_IMAGE050
Representing the shooting period of the camera.
Optionally, calibrating the parameter
Figure 193807DEST_PATH_IMAGE051
The value of (2) is 0.7.
Optionally, before the filtering image is expanded according to the taylor formula to obtain the edge feature image, the method further includes:
deriving the filtered image to obtain a derivative image;
and deriving the derivative image to obtain a second derivative image.
Optionally, the coal characteristic sample comprises a first coal characteristic sample and a second coal characteristic sample;
the calculated coal characteristic sample specifically comprises the following components:
calculating a first coal characteristic sample according to the gray scale compressed image
Figure 258846DEST_PATH_IMAGE052
；
Calculating a second coal characteristic sample according to the binary edge characteristic image
Figure 893090DEST_PATH_IMAGE053
。
The innovation points of the embodiment of the invention include:
1. in the embodiment, the image data of the coal fuel of the enterprises are shot through the camera, the coal in the image is detected and reserved, the heating value of the coal fuel is automatically analyzed and predicted, and the coal fuel is used as the objective data of carbon emission accounting, so that the accounting of the enterprises and the reporting of the greenhouse gas emission and related data are facilitated, and the method is one of the innovation points of the embodiment of the invention.
2. In this embodiment, compared with directly preserving the filtered image, the method can effectively reduce the dimension of the input image data by preserving the gray-scale compressed image and the binary edge feature image, greatly reduce the sample reserving storage amount of the data, and save the sample reserving storage space, which is one of the innovation points of the embodiment of the invention.
3. In this embodiment, compared with the hyperspectral image data source commonly used in industry, the adoption of the visible light image data source can greatly save the data acquisition cost, and in addition, the problems of low fitting degree and large detection error of the modeling of the characteristics of the coal sample by the traditional image regression analysis method are also overcome, so that the method is one of the innovation points of the embodiment of the invention.
4. In this embodiment, the provided coal heat prediction neural network model is dedicated to coal detection, and the model used for coal detection in the past is not specifically optimized. For the image recognition models in other fields, because of different detection objects, the network structure difference is huge, and the image recognition models cannot be directly applied to coal detection, so that the image recognition model is one of the innovation points of the embodiment of the invention.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method for detecting coal according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a coal sampling according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of another embodiment of the invention;
FIG. 4 is a schematic diagram of a reprocessing process according to an embodiment of the present invention;
FIG. 5 is a schematic diagram of a position structure of a camera and a coal material according to an embodiment of the present invention;
FIG. 6 is a schematic diagram of a calibration sampling range provided by an embodiment of the present invention;
fig. 7 is another flowchart of a coal detection method according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without any inventive effort, are intended to be within the scope of the invention.
It should be noted that the terms "comprising" and "having" and any variations thereof in the embodiments of the present invention and the accompanying drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those listed steps or elements but may include other steps or elements not listed or inherent to such process, method, article, or apparatus.
The embodiment of the invention discloses a coal detection method for carbon check. The following will describe in detail.
Fig. 1 is a flowchart of a coal detection method according to an embodiment of the present invention, fig. 2 is a schematic coal sampling diagram according to an embodiment of the present invention, fig. 3 is another schematic coal sampling diagram according to an embodiment of the present invention, fig. 4 is a post-reprocessed schematic diagram according to an embodiment of the present invention, fig. 5 is a schematic position structure diagram of a camera and a coal according to an embodiment of the present invention, fig. 6 is a schematic diagram of a calibrated sampling range according to an embodiment of the present invention, please refer to fig. 1 to 6, and the coal detection method for carbon check according to the embodiment of the present invention includes:
step 1: acquiring images of the coal to be detected according to a preset frame rate by using a camera;
step 2: calibrating the sampling range of the image to obtain a sampled image
Figure 52676DEST_PATH_IMAGE054
；
Step 3: setting a noise template according to the shape characteristics of the coal to be detected
Figure 557606DEST_PATH_IMAGE055
，
Figure 29432DEST_PATH_IMAGE056
Representing coordinates of pixels in the template with the center of the template as a reference;
step 4: filtering the sampled image by using a noise template to obtain a filtered image
Figure 213289DEST_PATH_IMAGE057
Wherein->
Figure 51932DEST_PATH_IMAGE058
Representing convolution operation,/->
Figure 755577DEST_PATH_IMAGE059
Representing pixel coordinates in the image;
step 5: expanding the filtered image according to a Taylor formula to obtain an edge characteristic image
Figure 146107DEST_PATH_IMAGE060
Wherein, the method comprises the steps of, wherein,
Figure 332107DEST_PATH_IMAGE061
for derivative image +.>
Figure 492217DEST_PATH_IMAGE062
Is a second derivative image;
step 6: thresholding the edge feature image to obtain a binary edge feature image
Figure 34057DEST_PATH_IMAGE063
Wherein->
Figure 142958DEST_PATH_IMAGE064
Representing a threshold value;
step 7: obtaining a gray scale compressed image according to the filtered image and the binary edge characteristic image
Figure 814591DEST_PATH_IMAGE065
Wherein, the method comprises the steps of, wherein,
Figure 667140DEST_PATH_IMAGE066
expressed as +.>
Figure 375071DEST_PATH_IMAGE067
The offset coordinates in the square neighborhood with the point as the center, W represents the side length of the neighborhood;
step 8: calculating a coal characteristic sample according to the gray scale compressed image and the binary edge characteristic image;
step 9: taking a coal characteristic sample as input, defining output as
Figure 576245DEST_PATH_IMAGE068
Establishing a prediction neural network model; wherein,,
Figure 769460DEST_PATH_IMAGE069
representing a third hidden layer->
Figure 815914DEST_PATH_IMAGE022
A linear weight representing the output of the element of the third hidden layer and the neural network model, ++>
Figure 115701DEST_PATH_IMAGE070
Representing a linear deviation +.>
Figure 894302DEST_PATH_IMAGE071
For activating a function, it is defined as
Figure 637130DEST_PATH_IMAGE072
，
Figure 487274DEST_PATH_IMAGE073
Representing an activation function convergence parameter; the predictive neural network model comprises a first hidden layer, a second hidden layer and a third hidden layer, wherein the first hidden layer is
Figure 795896DEST_PATH_IMAGE074
Figure 119298DEST_PATH_IMAGE075
，
Figure 146160DEST_PATH_IMAGE076
Representing the coordinates of the first hidden layer, +.>
Figure 65575DEST_PATH_IMAGE077
、
Figure 104069DEST_PATH_IMAGE078
Respectively indicate->
Figure 224472DEST_PATH_IMAGE079
、
Figure 863263DEST_PATH_IMAGE080
Linear weights of local neighborhoods of +.>
Figure 307408DEST_PATH_IMAGE081
Representing the linear deviation; the second hidden layer is
Figure 325042DEST_PATH_IMAGE082
，
Figure 6559DEST_PATH_IMAGE083
Representing the coordinates of the second hidden layer, +.>
Figure 86642DEST_PATH_IMAGE084
Linear weights representing pixels of the first hidden layer and pixels of the second hidden layer, +.>
Figure 754384DEST_PATH_IMAGE085
Representing the linear deviation; the third hidden layer is->
Figure 16738DEST_PATH_IMAGE086
Y represents the dimension of the third hidden layer, < >>
Figure 744522DEST_PATH_IMAGE040
Representing a linear weight +.>
Figure 76015DEST_PATH_IMAGE087
Representing the linear deviation;
step 10: obtaining the consumption function of the predictive neural network model by training the predictive neural network model
Figure 875344DEST_PATH_IMAGE088
，
Figure 867571DEST_PATH_IMAGE089
Representing the thermal value of the training sample, +.>
Figure 376044DEST_PATH_IMAGE090
Representing the output value +.>
Figure 86511DEST_PATH_IMAGE091
Representing the control coefficient;
step 11: iteratively calculating a predicted neural network model by using the consumption function to obtain a trained predicted neural network model;
step 12: and inputting the coal characteristic sample into the trained prediction neural network model to obtain an output value.
Specifically, please refer to fig. 1, 2, 3 and 5, in the method for detecting a coal material for carbon check provided in the present embodiment, firstly, in step 1, an image of the coal material to be detected is captured by using a camera 101, and when the image is collected, the image needs to be continuously collected according to a predetermined frame rate, therefore, in this embodiment, a conveyor belt 102 is arranged on a workshop site, the coal material to be detected is uniformly placed on the conveyor belt, in order to clearly collect the image of the coal material, a visible light camera 101 is disposed right above the conveyor belt 102, and the camera 101 faces downwards to the object plane of the conveyor belt 102, so that each captured image 104 includes a plurality of coal materials, and according to the characteristics of the coal materials in the image 104, the information of the coal material effective for predicting the heat productivity can be extracted. Note that, the black dots in the region 103 in fig. 5 represent the coal uniformly placed on the conveyor belt.
Because the quality of different areas of the image is different, after the image of the coal to be detected is acquired, the sampling range of the image 104 is calibrated in step 2, and referring to fig. 5 and 6, the sampling range of the image is related to the parameters such as the moving linear speed of the conveyor belt, the field angle of the camera, the distance between the camera and the conveyor belt, the shooting period of the camera, and the calibration parameters can be calculated according to the parameters
Figure 220689DEST_PATH_IMAGE092
According to the calibration parameters->
Figure 240991DEST_PATH_IMAGE093
The sampling range in each shot image 104 can be defined as a duty ratio in the image length direction +.>
Figure 310578DEST_PATH_IMAGE094
Is not limited in terms of the range of (a). The region 105 as marked in fig. 5 is a sampled image obtained by a calibrated sampling range. The sampling range is calibrated, so that the sampled images obtained each time are not overlapped, and the heating value passing through the recognition each time can be accumulated linearly.
The image also typically contains noise, and thus noise filtering of the image is required. Firstly in step 3, according to the shape characteristics of the coal to be detected, a noise template is set
Figure 118128DEST_PATH_IMAGE095
In this embodiment, the noise template is set as
Figure 259259DEST_PATH_IMAGE096
Wherein->
Figure 226078DEST_PATH_IMAGE097
The template of the elliptic curve is represented,
Figure 574889DEST_PATH_IMAGE098
representing a Gaussian filter template, ">
Figure 259948DEST_PATH_IMAGE099
Representing coordinates of pixels in the template with respect to the center of the template,/->
Figure 1508DEST_PATH_IMAGE100
Representing the variance of the Gaussian template, +.>
Figure 822834DEST_PATH_IMAGE101
Representing a natural exponential function. The noise template is set to be a combination of the Gaussian filtering template and the elliptic curve template, so that noise can be filtered more efficiently, and the texture characteristics of coal particles can be maintained.
Then in step 4, the sampled image is filtered by using the noise template obtained in step 3 to filter noise. During filtering, the sampled image and the noise template are subjected to convolution operation to obtain a filtered image
Figure 47273DEST_PATH_IMAGE102
Wherein->
Figure 78683DEST_PATH_IMAGE103
Representing convolution operation,/->
Figure 233721DEST_PATH_IMAGE104
Representing pixel coordinates in the image.
After the filtering processing, in order to save the storage space, the invention reprocesss the filtered image. Referring to fig. 1 and 4, the reprocessing of the filtered image includes steps 5 to 7, firstly, the filtered image is expanded according to taylor formula through step 5 to obtain an edge feature image of the filtered image
Figure 540244DEST_PATH_IMAGE105
Wherein, the method comprises the steps of, wherein,
Figure 122535DEST_PATH_IMAGE106
for derivative image +.>
Figure 641241DEST_PATH_IMAGE107
Is a second derivative image.
It should be noted that, since the derivative image and the second derivative image are required when the edge feature image is acquired, before performing step 5, the method further includes: step 41, deriving the filtered image to obtain a derivative image; step 42, deriving the derivative image to obtain a second derivative image. Referring to fig. 7, fig. 7 is another flowchart of a coal detection method according to an embodiment of the present invention, in which the filtered image is processed in step 41
Figure 334391DEST_PATH_IMAGE108
Deriving, obtaining a derivative image as +.>
Figure 208937DEST_PATH_IMAGE109
. In step 42, the obtained derivative image is further derived to obtain a second derivative image +.>
Figure 86763DEST_PATH_IMAGE110
。
In step 6, thresholding the edge feature image to obtain a binary edge feature image
Figure 702552DEST_PATH_IMAGE111
Wherein->
Figure 573294DEST_PATH_IMAGE064
Representing a threshold. The binary edge feature image reflects the change degree of each pixel in the image, and the probability that the value of the binary edge feature image at the edge of the coal particles is 1 is far greater than the probability that the value of the binary edge feature image at the edge of the coal particles is 0 because the change degree of the pixels at the edge of the coal particles is high, so that the characteristics of the coal images can be reflected.
In step 7, a gray-scale compressed image is obtained according to the filtered image and the binary edge feature image
Figure 489297DEST_PATH_IMAGE112
Wherein, the method comprises the steps of, wherein,
Figure 538025DEST_PATH_IMAGE113
expressed as +.>
Figure 906689DEST_PATH_IMAGE114
Offset coordinates in a square neighborhood with a point as the center, W represents the side length of the neighborhood, double-diagonal +.>
Figure 613745DEST_PATH_IMAGE115
Representing the integer part after the division operation. The gray scale number of the pixels of the gray scale compressed image G is 3, and the corresponding binary edge feature image +.>
Figure 384255DEST_PATH_IMAGE116
The pixel gray scale number is 1, so the total pixel gray scale number of the two is 4; filtered image +.>
Figure 59343DEST_PATH_IMAGE117
The pixel gray scale number is 8. Thus, each time the sample is left, the gray-scale compressed image G and the binary edge feature image are left +.>
Figure 774359DEST_PATH_IMAGE118
Compared to directly preserving the filtered image +.>
Figure 613002DEST_PATH_IMAGE119
Saving half of the storage space. Furthermore, according to the gray-scale compressed image G and the binary edge feature image +.>
Figure 378963DEST_PATH_IMAGE120
The filtered image may be approximately restored so that the coal may be viewed.
The gray-scale compressed image G and the binary edge characteristic image are obtained through the reprocessing steps
Figure 644860DEST_PATH_IMAGE118
Then, in step 8, a coal characteristic sample is established according to the gray-scale compressed image and the binary edge characteristic image, and the obtained coal characteristic sample contains gray scales from the gray-scale compressed imageThe characteristics of the coal material are characterized by comprising local texture characteristics from the binary edge characteristic image, so that the heating value of the coal material can be reflected more accurately.
In step 9, a coal heat prediction neural network model taking a coal characteristic sample as input and a heat value as output is established. The predictive neural network model includes, in addition to input and output, a plurality of hidden layers, each hidden layer including a linear model expressed as a linear weighted sum of the outputs of the previous layer and a nonlinear model called an activation function defined as
Figure 847171DEST_PATH_IMAGE121
，
Figure 223926DEST_PATH_IMAGE122
Represents the convergence parameter of the activation function for controlling the convergence speed of the activation function. As a preferred value, +.>
Figure 280612DEST_PATH_IMAGE123
The activation function of each hidden layer is the same. />
In the invention, the hidden layer comprises a first hidden layer, which is defined as
Figure 842044DEST_PATH_IMAGE124
Figure 407017DEST_PATH_IMAGE125
，
Figure 462829DEST_PATH_IMAGE126
Representing the coordinates of the first hidden layer, +.>
Figure 328017DEST_PATH_IMAGE127
Representation of
Figure 60349DEST_PATH_IMAGE128
Offset coordinates in square neighborhood with point as center, +.>
Figure 847040DEST_PATH_IMAGE129
Representing the corresponding coordinates in the input image, +.>
Figure 411270DEST_PATH_IMAGE130
、
Figure 990019DEST_PATH_IMAGE131
Respectively indicate->
Figure 768619DEST_PATH_IMAGE132
、
Figure 917971DEST_PATH_IMAGE133
Linear weights of local neighborhoods of +.>
Figure 440220DEST_PATH_IMAGE134
Representing the linear deviation. Features locally related to the image may be extracted by the first hidden layer.
The hidden layer further comprises a second hidden layer defined as
Figure 139054DEST_PATH_IMAGE135
Wherein->
Figure 88556DEST_PATH_IMAGE136
Representing the coordinates of the first hidden layer, +.>
Figure 958160DEST_PATH_IMAGE137
Representing the coordinates of the second hidden layer, each pixel of the first layer (++>
Figure 346416DEST_PATH_IMAGE138
) With any one of the pixels of the second layer (+)>
Figure 650490DEST_PATH_IMAGE139
) All have a linear weight, noted +.>
Figure 770893DEST_PATH_IMAGE140
，
Figure 144105DEST_PATH_IMAGE141
Representing the linear deviation. Features related to the global distribution of the image may be extracted by the second hidden layer.
The hidden layer further comprises a third hidden layer defined as
Figure 8156DEST_PATH_IMAGE142
，
Figure 933780DEST_PATH_IMAGE143
Representing the coordinates of the second hidden layer, the third hidden layer +.>
Figure 225084DEST_PATH_IMAGE144
Is a vector with a fixed dimension of 64, y represents the dimension of the vector of the third hidden layer, each pixel of the second hidden layer (">
Figure 288855DEST_PATH_IMAGE139
) A linear weight with any dimension of the third hidden layer vector is recorded as
Figure 753335DEST_PATH_IMAGE145
，
Figure 766421DEST_PATH_IMAGE146
Representing the linear deviation. Defining output as according to the third hidden layer
Figure 556522DEST_PATH_IMAGE147
，
Figure 779693DEST_PATH_IMAGE148
A linear weight representing an element of the third hidden layer and the neural network model output (i.e. the calorific value,)>
Figure 625027DEST_PATH_IMAGE149
Representing the linear deviation.
In step 10, a prediction is obtained by training a prediction neural network modelConsumption function of neural network model
Figure 882833DEST_PATH_IMAGE150
，
Figure 640574DEST_PATH_IMAGE151
Representing the thermal value of the training sample, +.>
Figure 351041DEST_PATH_IMAGE152
Representing the output value +.>
Figure 235951DEST_PATH_IMAGE153
Representing the control coefficient, in the present invention is set +.>
Figure 82685DEST_PATH_IMAGE154
Which contributes to an improved robustness of the method to data noise, is taken as a preferred value +.>
Figure 480168DEST_PATH_IMAGE155
. Then in step 11, the predictive neural network model is iteratively calculated using the consumption function such that +.>
Figure 209090DEST_PATH_IMAGE156
And (5) converging, namely finishing training to obtain parameters of each hidden layer, and thus obtaining a trained predictive neural network model. In the iterative calculation, a BP (Back Propagation) algorithm may be adopted, and the BP algorithm may refer to the existing data, which is not described herein.
After the trained prediction neural network model is obtained, in step 12, the coal characteristic sample is input into the trained prediction neural network model, so that an output value can be obtained, and the heat value of the coal is obtained. The coal heat prediction neural network model obtained through training is special for coal detection, and the model used for coal detection in the past is not specially optimized. For the image recognition models in other fields, because of different detection objects, the network structure difference is huge, and the image recognition models cannot be directly applied to coal detection.
According to the coal detection method for carbon check, the image data of the enterprise coal fuel is shot through the camera, the coal in the image is detected and reserved, the heating value of the coal fuel is automatically analyzed and predicted, the coal is used as objective data for carbon emission accounting, and the accounting of chemical enterprises and the reporting of the greenhouse gas emission and related data are facilitated. The method solves the problems of low fitting degree and large detection error of the modeling of the coal sample characteristics by the traditional image regression analysis method; compared with the hyperspectral image data source commonly adopted in the industry, the visible light image data source can greatly save the data acquisition cost. In addition, the method can effectively reduce the dimension of the input image data, greatly reduce the data sample reserving storage space and save the sample reserving storage space.
Optionally, referring to fig. 1, in step 2, the sampling range of the image is calibrated, specifically: calibrating the sampling range of the image according to the movement linear speed of the conveyor belt, the field angle of the camera, the distance between the camera and the conveyor belt and the shooting period of the camera to obtain calibration parameters
Figure 614137DEST_PATH_IMAGE157
Wherein->
Figure 440010DEST_PATH_IMAGE158
Represents the linear speed of movement of the conveyor belt, +.>
Figure 946078DEST_PATH_IMAGE159
Representing the angle of view of the camera, +.>
Figure 240924DEST_PATH_IMAGE160
Indicating the distance of the camera from the conveyor belt, +.>
Figure 592271DEST_PATH_IMAGE161
Representing the shooting period of the camera.
Specifically, referring to fig. 1 and 5, in order to facilitate the photographing of the coal, a conveyor belt 102 is disposed on site in a workshop, the coal to be detected is uniformly disposed on the conveyor belt 102, and the conveyor belt 1 is disposedA visible light camera 101 is arranged right above the belt conveyor 102, and the camera 101 is arranged downwards to face the object plane of the belt conveyor 102. Therefore, when the sampling range of the image is calibrated in step 2, the calibration parameters are related to parameters such as the movement linear velocity of the conveyor belt 102, the angle of view of the camera 101, the distance between the camera 101 and the conveyor belt 102, the photographing period of the camera 101, and the like, and the calibration parameters are obtained based on the above parameters
Figure 538231DEST_PATH_IMAGE162
Wherein->
Figure 526784DEST_PATH_IMAGE163
Represents the linear speed of movement of the conveyor belt, +.>
Figure 433560DEST_PATH_IMAGE164
Representing the angle of view of the camera, +.>
Figure 713232DEST_PATH_IMAGE165
Indicating the distance of the camera from the conveyor belt, +.>
Figure 389064DEST_PATH_IMAGE166
Representing the shooting period of the camera.
The sampling range is calibrated, so that the sampled images obtained each time are not overlapped, and the heating value passing through the recognition each time can be accumulated linearly. From the viewpoint of image quality, the shooting quality of the area close to the center of the image is higher, so that the calibration parameters can be controlled
Figure 112300DEST_PATH_IMAGE167
Improving the image quality, experimental preference +.>
Figure 506373DEST_PATH_IMAGE168
. Of course, is->
Figure 324156DEST_PATH_IMAGE169
The value of 0.7 is only one preferred embodiment of the present invention in this example, and is not intended to limit the present invention, in other examples，
Figure 120073DEST_PATH_IMAGE167
The value of (2) may also be other as long as +.>
Figure 984518DEST_PATH_IMAGE170
And (3) obtaining the product.
Optionally, the coal characteristic samples comprise a first coal characteristic sample and a second coal characteristic sample; the calculation coal characteristic sample specifically comprises: calculating a first coal characteristic sample according to the gray scale compressed image
Figure 865886DEST_PATH_IMAGE171
The method comprises the steps of carrying out a first treatment on the surface of the Calculating a second coal characteristic sample from the binary edge characteristic image>
Figure 487360DEST_PATH_IMAGE172
。
Specifically, the coal characteristic sample comprises a first coal characteristic sample calculated according to the gray scale compressed image G
Figure 137784DEST_PATH_IMAGE173
And from binary edge feature images
Figure 937244DEST_PATH_IMAGE120
The calculated second coal characteristic sample
Figure 368226DEST_PATH_IMAGE174
. The first coal characteristic sample and the second coal characteristic sample are combined to form the coal characteristic sample, so that the coal characteristic sample simultaneously contains gray characteristics from gray-scale compressed images and local texture characteristics from binary edge characteristic images, and the calorific value of the coal can be reflected more accurately.
Based on the method, the inventor detects the coal sample in the image and reserves the coal material by shooting the image data of the coal fuel of the enterprise, and automatically analyzes and predicts the heating value of the coal fuel. According to the method, the inventors carried out tests on coal materials with different granularities and different colors, and obtained test results shown in table 1. Referring to the field test data given in table 1, the results show that the method has lower error compared with the true value in identifying the heating value of the coal materials with different granularities, and can meet the application requirements.
TABLE 1
Figure 403178DEST_PATH_IMAGE175
Those of ordinary skill in the art will appreciate that: the drawing is a schematic diagram of one embodiment and the modules or flows in the drawing are not necessarily required to practice the invention.
Those of ordinary skill in the art will appreciate that: the modules in the apparatus of the embodiments may be distributed in the apparatus of the embodiments according to the description of the embodiments, or may be located in one or more apparatuses different from the present embodiments with corresponding changes. The modules of the above embodiments may be combined into one module, or may be further split into a plurality of sub-modules.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims (5)

1. A coal detection method for carbon verification, comprising:
acquiring images of the coal to be detected according to a preset frame rate by using a camera;
calibrating the sampling range of the image to obtain a sampled image
Figure 267010DEST_PATH_IMAGE001
；
Setting a noise template according to the shape characteristics of the coal to be detected
Figure 325096DEST_PATH_IMAGE002
，
Figure 416417DEST_PATH_IMAGE003
Representing coordinates of pixels in the template with the center of the template as a reference;
filtering the sampled image by using the noise template to obtain a filtered image
Figure 77206DEST_PATH_IMAGE004
Wherein->
Figure 219474DEST_PATH_IMAGE005
Representing convolution operation,/->
Figure 323828DEST_PATH_IMAGE006
Representing pixel coordinates in the image;
expanding the filtered image according to a Taylor formula to obtain an edge characteristic image as
Figure 122019DEST_PATH_IMAGE007
Figure 852078DEST_PATH_IMAGE008
Wherein->
Figure 786536DEST_PATH_IMAGE009
Is a derivative image, ++>
Figure 359993DEST_PATH_IMAGE010
The second derivative image;
thresholding the edge feature image to obtain a binary edge feature image
Figure 379902DEST_PATH_IMAGE011
Wherein->
Figure 929963DEST_PATH_IMAGE012
Representing a threshold value;
obtaining a gray scale compressed image according to the filtered image and the binary edge feature image
Figure 515665DEST_PATH_IMAGE013
Wherein->
Figure 522673DEST_PATH_IMAGE014
Expressed in terms of
Figure 29878DEST_PATH_IMAGE015
The offset coordinates in the square neighborhood with the point as the center, W represents the side length of the neighborhood;
calculating a coal characteristic sample according to the gray scale compressed image and the binary edge characteristic image;
taking the coal characteristic sample as input, defining output as
Figure 836160DEST_PATH_IMAGE016
Figure 10789DEST_PATH_IMAGE017
Establishing a prediction neural network model; wherein (1)>
Figure 627846DEST_PATH_IMAGE018
Representing a third hidden layer->
Figure 684664DEST_PATH_IMAGE019
A linear weight representing the output of the element of the third hidden layer and the neural network model, ++>
Figure 294637DEST_PATH_IMAGE020
Representing a linear deviation +.>
Figure 310391DEST_PATH_IMAGE021
For activating a function, it is defined as
Figure 613196DEST_PATH_IMAGE022
，
Figure 891731DEST_PATH_IMAGE023
Representing an activation function convergence parameter; the prediction neural network model comprises a first hidden layer, a second hidden layer and a third hidden layer, wherein the first hidden layer is
Figure 56127DEST_PATH_IMAGE024
Figure 674190DEST_PATH_IMAGE025
，
Figure 147897DEST_PATH_IMAGE026
Representing the coordinates of the first hidden layer, +.>
Figure 116990DEST_PATH_IMAGE027
、
Figure 114771DEST_PATH_IMAGE028
Respectively indicate->
Figure 118499DEST_PATH_IMAGE029
、
Figure 497527DEST_PATH_IMAGE030
Linear weights of local neighborhoods of +.>
Figure 235807DEST_PATH_IMAGE031
Representing the linear deviation; the second hidden layer is
Figure 788012DEST_PATH_IMAGE032
，
Figure 367285DEST_PATH_IMAGE033
Representing the coordinates of the second hidden layer,
Figure 510690DEST_PATH_IMAGE034
linear weights representing pixels of the first hidden layer and pixels of the second hidden layer, +.>
Figure 205108DEST_PATH_IMAGE035
Representing the linear deviation; the third hidden layer is->
Figure 764265DEST_PATH_IMAGE036
Y represents the dimension of the third hidden layer, < >>
Figure 211427DEST_PATH_IMAGE037
Representing a linear weight +.>
Figure 932258DEST_PATH_IMAGE038
Representing the linear deviation;
obtaining a consumption function of the predictive neural network model by training the predictive neural network model
Figure 409245DEST_PATH_IMAGE039
，
Figure 240935DEST_PATH_IMAGE040
Representing the thermal value of the training sample, +.>
Figure 73762DEST_PATH_IMAGE041
Representing the output value +.>
Figure 512964DEST_PATH_IMAGE042
Representing the control coefficient;
iteratively calculating the prediction neural network model by using the consumption function to obtain a trained prediction neural network model;
and inputting the coal characteristic sample into the trained predictive neural network model to obtain an output value.
2. The method for detecting coal for carbon check according to claim 1, wherein the calibrating the sampling range of the image specifically comprises:
calibrating the sampling range of the image according to the movement linear speed of the conveyor belt, the field angle of the camera, the distance between the camera and the conveyor belt and the shooting period of the camera to obtain calibration parameters
Figure 431242DEST_PATH_IMAGE043
Wherein->
Figure 66622DEST_PATH_IMAGE044
Represents the linear speed of movement of the conveyor belt, +.>
Figure 525593DEST_PATH_IMAGE045
Representing the angle of view of the camera, +.>
Figure 853806DEST_PATH_IMAGE046
Indicating the distance of the camera from the conveyor belt, +.>
Figure 197063DEST_PATH_IMAGE047
Representing the shooting period of the camera.
3. The method for detecting coal for carbon check according to claim 2, wherein the parameter is calibrated
Figure 449183DEST_PATH_IMAGE048
The value of (2) is 0.7.
4. The method for detecting coal for carbon check according to claim 1, wherein before the filtering image is expanded according to taylor formula to obtain an edge feature image, further comprising:
deriving the filtered image to obtain a derivative image;
and deriving the derivative image to obtain a second derivative image.
5. The method for carbon check as claimed in claim 1, wherein the coal characteristic samples include a first coal characteristic sample and a second coal characteristic sample;
the calculated coal characteristic sample specifically comprises the following components:
calculating a first coal characteristic sample according to the gray scale compressed image
Figure 53340DEST_PATH_IMAGE049
；
Calculating a second coal characteristic sample according to the binary edge characteristic image
Figure 801722DEST_PATH_IMAGE050
。/>
CN202211273060.1A 2022-10-18 2022-10-18 Coal detection method for carbon check Active CN115631158B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211273060.1A CN115631158B (en) 2022-10-18 2022-10-18 Coal detection method for carbon check

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211273060.1A CN115631158B (en) 2022-10-18 2022-10-18 Coal detection method for carbon check

Publications (2)

Publication Number Publication Date
CN115631158A CN115631158A (en) 2023-01-20
CN115631158B true CN115631158B (en) 2023-05-12

Family

ID=84907344

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211273060.1A Active CN115631158B (en) 2022-10-18 2022-10-18 Coal detection method for carbon check

Country Status (1)

Country Link
CN (1) CN115631158B (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2011148924A (en) * 2010-01-22 2011-08-04 Nippon Steel Corp Method for estimating coke oven gas yield and method for producing coke
CN112836902A (en) * 2021-03-11 2021-05-25 西北大学 Coal combustion calorific capacity prediction method based on improved BP neural network

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105026902A (en) * 2013-03-07 2015-11-04 西门子公司 Systems and methods for boosting coal quality measurement statement of related cases

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2011148924A (en) * 2010-01-22 2011-08-04 Nippon Steel Corp Method for estimating coke oven gas yield and method for producing coke
CN112836902A (en) * 2021-03-11 2021-05-25 西北大学 Coal combustion calorific capacity prediction method based on improved BP neural network

Non-Patent Citations (6)

* Cited by examiner, † Cited by third party
Title
A method for in-situ measurement of calorific value of coal_ a numerical study;Linglong Wang,Xue cheng Wu,et.al;Thermochimica Acta;全文 *
Determination of Calorific Value of Mixed Coals by Analysis of Major Elements Using Data Pre-Processing in Laser-Induced Breakdown Spectroscopy;Jong Hyun Park,Choong Mo Ryu,et.al;Applied Sciences;全文 *
Quantitative Analysis of Calorific Value of Coal Based on Spectral Preprocessing by Laser-Induced Breakdown Spectroscopy(LIBS);Wenbing Li,Jidong Lu,et.al;Energy Fuels;全文 *
一种燃煤发热量的综合预测方法;王晓红，吴德会;煤炭科学技术;第34卷(第6期);全文 *
基于高光谱图像和卷积神经网络的燃煤热值估计算法;杨明花，张克涵;中国电力;第52卷(第9期);全文 *
采用KPCA特征提取的近红外煤炭发热量预测模型;雷萌，李明;化工学报;第63卷(第12期);全文 *

Also Published As

Publication number Publication date
CN115631158A (en) 2023-01-20

Similar Documents

Publication Publication Date Title
CN111094956B (en) Processing thermal imaging images with neural networks to identify subsurface erosion on insulation (CUI)
CN111325713B (en) Neural network-based wood defect detection method, system and storage medium
KR102166458B1 (en) Defect inspection method and apparatus using image segmentation based on artificial neural network
CN108960245B (en) Tire mold character detection and recognition method, device, equipment and storage medium
CN108918536B (en) Tire mold surface character defect detection method, device, equipment and storage medium
CN111126136A (en) Smoke concentration quantification method based on image recognition
CN111630356A (en) Method for characterizing a sample using a neural network
KR102141302B1 (en) Object detection method based 0n deep learning regression model and image processing apparatus
Laga et al. Image-based plant stornata phenotyping
CN117541483B (en) Structural performance evaluation method and system for diffusion plate
CN117152119A (en) Profile flaw visual detection method based on image processing
CN117668518B (en) Discrete intelligent manufacturing method and system
CN111627018B (en) Steel plate surface defect classification method based on double-flow neural network model
CN113155839A (en) Steel plate outer surface defect online detection method based on machine vision
US20240087105A1 (en) Systems and Methods for Paint Defect Detection Using Machine Learning
CN115631158B (en) Coal detection method for carbon check
Zhang et al. Weld joint penetration state sequential identification algorithm based on representation learning of weld images
CN113112482A (en) PCB defect detection method based on attention mechanism network
CN117011759A (en) Method and system for analyzing multi-element geological information of surrounding rock of tunnel face by drilling and blasting method
CN111105417A (en) Image noise positioning method and system
CN113607068B (en) Method for establishing and extracting recognition model of photoacoustic measurement signal characteristics
CN113077002B (en) Machine olfaction visual sensing data analysis method based on space heterodyne Raman spectrum
CN114548250A (en) Mobile phone appearance detection method and device based on data analysis
CN112767365A (en) Flaw detection method
Huang et al. ASD-YOLO: An aircraft surface defects detection method using deformable convolution and attention mechanism

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