CN113780359A - Method and device for identifying insulator in infrared image and readable storage medium - Google Patents

Method and device for identifying insulator in infrared image and readable storage medium Download PDF

Info

Publication number
CN113780359A
CN113780359A CN202110936295.3A CN202110936295A CN113780359A CN 113780359 A CN113780359 A CN 113780359A CN 202110936295 A CN202110936295 A CN 202110936295A CN 113780359 A CN113780359 A CN 113780359A
Authority
CN
China
Prior art keywords
insulator
classifier
sample
image
infrared
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.)
Pending
Application number
CN202110936295.3A
Other languages
Chinese (zh)
Inventor
伍伟权
文安
刘国特
周锦辉
周妙娴
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.)
Foshan University
Original Assignee
Foshan University
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 Foshan University filed Critical Foshan University
Priority to CN202110936295.3A priority Critical patent/CN113780359A/en
Publication of CN113780359A publication Critical patent/CN113780359A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G06F18/2148—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the process organisation or structure, e.g. boosting cascade
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Investigating Or Analysing Materials By Optical Means (AREA)

Abstract

The invention relates to the technical field of image recognition processing, in particular to a method and a device for recognizing insulators in infrared images and a readable storage medium, wherein the method comprises the following steps: acquiring a sample image data set of an insulator, wherein the sample image data set of the insulator comprises a plurality of sample images; the sample image comprises a positive sample image and a negative sample image, and the sample image is obtained according to infrared image preprocessing of the insulator; constructing a weak classifier, and training the constructed weak classifier according to the sample image data set to obtain an improved cascade classifier model; receiving an infrared image of an insulator to be detected, and predicting an insulator region in the infrared image of the insulator to be detected through an improved cascade classifier model; the invention also correspondingly provides a device for identifying the insulator in the infrared image and a readable storage medium, and the device and the method can improve the accuracy and the efficiency of identifying the insulator in the infrared image.

Description

Method and device for identifying insulator in infrared image and readable storage medium
Technical Field
The invention relates to the technical field of image recognition processing, in particular to a method and a device for recognizing insulators in infrared images and a readable storage medium.
Background
Insulators are important component equipment with huge application quantity in overhead transmission lines, the traditional manual detection is difficult to accurately recognize multiple targets of infrared images of a large number of insulators, and the recognition speed and accuracy of the existing intelligent recognition algorithm are difficult to meet the field application requirements.
Therefore, how to improve the existing intelligent recognition algorithm and improve the accuracy and recognition efficiency of insulator recognition in the infrared image becomes a problem to be solved urgently.
Disclosure of Invention
The present invention is directed to a method and an apparatus for identifying insulators in an infrared image, and a readable storage medium, to solve one or more technical problems in the prior art, and to provide at least one useful choice or creation condition.
In order to achieve the purpose, the invention provides the following technical scheme:
a method for identifying an insulator in an infrared image comprises the following steps:
s100, obtaining a sample image data set of the insulator, wherein the sample image data set of the insulator comprises a plurality of sample images; the sample image comprises a positive sample image and a negative sample image, and the sample image is obtained according to infrared image preprocessing of the insulator;
s200, constructing a weak classifier, and training the constructed weak classifier according to the sample image data set to obtain an improved cascade classifier model;
step S300, receiving an infrared image of the insulator to be detected, and predicting an insulator region in the infrared image of the insulator to be detected through an improved cascade classifier model.
Further, the step S100 includes:
step S110, acquiring an infrared image of the transformer substation, wherein the infrared image comprises an insulator;
step S120, unifying the pixels of the infrared image, and carrying out gray processing on the infrared image with unified pixels to convert the infrared image into a gray histogram;
step S130, carrying out equalization processing on the gray level histogram to obtain a final sample image;
step S140, determining the characteristic values of the sample images by using an integral graph method, and constructing a sample image data set according to the characteristic values of the sample images.
Further, the step S200 includes:
step S210, determining a weak classifier; the weak classifier adopts a cascade Gentle Adaboost classifier;
step S220, determining the minimum classification error rate of each level of classifier in the weak classifier, and searching the minimum threshold factor of the weak classifier according to the classification error rate of each level of classifier;
step S230, finding the minimum classification error rate from the minimum classification error rates of all weak classifiers, updating the weight of the sample image input into the weak classifier for training according to the minimum classification error, and outputting a strong classifier;
step S240, setting the minimum hit rate and the maximum false alarm rate of each stage of the strong classifier, training the cascaded strong classifiers, and obtaining an improved cascaded classifier model when the hit rate of each stage of the strong classifier reaches the minimum hit rate and the false alarm rate is lower than the maximum false alarm rate.
Further, the weak classifier is defined by:
Figure BDA0003212986820000021
in the formula: f (x)i) For the region x to be detected in the sample imageiA feature calculation function of (a); theta is a minimization threshold factor set by the weak classifier; class value alpha1And alpha2Is in the absolute value of [0,1 ]]Between, the classification value alpha1And alpha2Is proportional to the confidence of the weak classifier.
Further, the step S220 includes:
step S221, let j equal to 1,2,3.., p, where p is the total number of features;
step S222, calculating the weight sum T of all insulator samples in the sample image data set+And the sum of the weights T of all non-insulator samples-；
Step S223, calculating the sum of the weights of all positive sample images before the sample i
Figure BDA0003212986820000023
And sum of weights of all negative sample images
Figure BDA0003212986820000024
Wherein i is a sample serial number corresponding to the positive sample image; j is a characteristic serial number corresponding to the negative sample image;
step S224, calculating to obtain a minimum classification error rate according to the following formula, and simultaneously recording a minimum threshold factor theta corresponding to the weak classifier;
Figure BDA0003212986820000022
further, the step S230 includes:
step S231, marking N sample images as (x)1,y1)，…，(xn,yn) Wherein y isi∈{－1,1}，yi1, denoted as insulator sample; y isiA non-insulator sample is denoted by-1.
Step S232, initializing sample weight by using the following formula;
wi＝1/N,i＝1,...,N；
step S233, let J equal 1,2,3., J, where J represents the total number of iterations;
calling a weak classifier training algorithm to find out a weak classifier h with the minimum classification error from all weak classifiersj(xi) The corresponding minimization threshold factor θ, which is expressed as the weak classifier h by the following formulaj(xi) The classification error of (2);
Figure BDA0003212986820000031
in the formula ofjIs the minimum classification error for the jth iteration; k is a constant number, 0<k<1；
Order to
Figure BDA0003212986820000032
Calculating the average weight of the sample images in the round by using the following formula;
Figure BDA0003212986820000033
calculating the standard deviation of the weight by using the following formula;
Figure BDA0003212986820000034
the weights of the sample images are updated using the following formula:
Figure BDA0003212986820000035
in the formula: alpha is alphajIs an intermediate variable; h isj(xi) The classifier corresponding to the jth iteration is represented; ziIs hj(xi) A normalization factor of the classifier; m is the training average weight;
normalization factor ZiCan be expressed as:
Figure BDA0003212986820000036
an output strong classifier G (x), wherein:
Figure BDA0003212986820000037
a computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, implements the steps of the method for identifying insulators in infrared images according to any one of the above.
A system for identifying insulators in infrared images, the system comprising:
at least one processor;
at least one memory for storing at least one program;
when the at least one program is executed by the at least one processor, the at least one processor is enabled to implement any one of the above methods for identifying insulators in infrared images.
The invention has the beneficial effects that: the invention discloses a method, a device and a readable storage medium for identifying insulators in infrared images. According to the invention, through improving the classifier training algorithm, the classifier identification accuracy is improved, and the classifier training and identification efficiency is effectively improved by adopting a new weight value updating rule.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
Fig. 1 is a schematic flow chart of a method for identifying insulators in an infrared image according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of an embodiment of equalizing a gray histogram;
FIG. 3 is a schematic illustration of a sample image in an embodiment of the invention;
FIG. 4 is a diagram illustrating integral definition in an embodiment of the present invention;
FIG. 5 is a diagram illustrating a correspondence between the number of cascade layers and the hit rate in an embodiment of the present invention;
fig. 6 is a schematic diagram illustrating a correspondence relationship between the number of cascade layers and the hit rate false alarm rate in the embodiment of the present invention.
Detailed Description
The conception, specific structure and technical effects of the present application will be described clearly and completely with reference to the following embodiments and the accompanying drawings, so that the purpose, scheme and effects of the present application can be fully understood. It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.
Referring to fig. 1, as shown in fig. 1, a method for identifying an insulator in an infrared image according to an embodiment of the present application includes the following steps:
s100, obtaining a sample image data set of the insulator, wherein the sample image data set of the insulator comprises a plurality of sample images; the sample image comprises a positive sample image and a negative sample image, and the sample image is obtained according to infrared image preprocessing of the insulator;
s200, constructing a weak classifier, and training the constructed weak classifier according to the sample image data set to obtain an improved cascade classifier model;
step S300, receiving an infrared image of the insulator to be detected, and predicting an insulator region in the infrared image of the insulator to be detected through an improved cascade classifier model.
In the embodiment provided by the invention, firstly, a large number of infrared images acquired on site are used for constructing an insulator infrared data set, and a plurality of weak classifiers are constructed; the classifier is improved through training, so that the recognition accuracy of the classifier is improved, and the training and recognition efficiency of the classifier is effectively improved.
As a further improvement of the above embodiment, the step S100 includes:
step S110, acquiring an infrared image of the transformer substation, wherein the infrared image comprises an insulator;
step S120, unifying the pixels of the infrared image, and carrying out gray processing on the infrared image with unified pixels to convert the infrared image into a gray histogram;
step S130, carrying out equalization processing on the gray level histogram to obtain a final sample image;
in some embodiments, an infrared image data set of the insulator is constructed by using a large number of infrared images of the transformer substation, pixels of the infrared images are unified into 30 × 100 in order to improve training and recognition speed of a subsequent classifier, and the infrared images are subjected to graying processing and converted into grayscale histograms. Considering that the gray value distribution of each gray level in the gray histogram is not uniform, then the gray histogram is equalized (as shown in fig. 2) to make the image uniform in brightness and detail features clear, so as to obtain the final sample image. The processing results are shown in fig. 3.
Step S140, determining the characteristic values of the sample images by using an integral graph method, and constructing a sample image data set according to the characteristic values of the sample images.
As shown in fig. 4, an integral definition map is used, and haar-like feature values of a sample image can be rapidly calculated by using an integral map method. In fig. 4, the D region is indicated as a dark portion in the feature rectangle, and the remaining 3-portion region indicates a white portion in the feature rectangle. The sum of the pixel values of the regions is:
Figure BDA0003212986820000051
Figure BDA0003212986820000052
Figure BDA0003212986820000053
Figure BDA0003212986820000054
characteristic value lambda of D regionDCan be expressed as:
λD＝IA(x,y)+IB(x,y)+IC(x,y)-ID(x,y)(5)；
in the formula: i (x, y) is the sum of pixel values of each region; i (x ', y') is the gray value of the point (x ', y').
The integral image of the sample image can be obtained by only scanning the whole sample image once.
As a further refinement of the above embodiment, the weak classifier is defined by:
Figure BDA0003212986820000055
in the formula: f (x)i) For the region x to be detected in the sample imageiA feature calculation function of (a); theta is a minimization threshold factor set by the weak classifier; class value alpha1And alpha2Is in the absolute value of [0,1 ]]Between, the classification value alpha1And alpha2Is proportional to the confidence of the weak classifier.
As a further improvement of the above embodiment, the step S200 includes:
step S210, determining a weak classifier; the weak classifier adopts a cascade Gentle Adaboost classifier;
in the weak classifier training process, there are mainly 2 error sources: firstly, identifying an insulator sub-target as a non-insulator target; and identifying the non-insulator target as an insulator target. The sum of the weights of the 2 error samples can be calculated by equations (7) and (8), respectively:
Figure BDA0003212986820000061
Figure BDA0003212986820000062
in the formula: w is ai()The sample weight of the ith area under the condition of (1); n is the total number of samples; w is ai()The sample weight of the ith area under the condition of the second condition is taken as the sample weight of the ith area under the condition of the second condition; epsilonpn、εnpRespectively, the sum of the weights in the corresponding case.
In the training process of the current classifier, the 2 error conditions are treated in the same way. However, in the insulator target recognition image, the proportion of the insulator sub-targets in the image is small. Thus, the erroneous recognition of the insulator sub-target as a non-insulator region has a greater impact than the erroneous recognition of the non-insulator target as the insulator sub-target. In the cascade process of the classifiers, if a certain classifier identifies the insulator sub-target as a non-insulator target, the region is permanently excluded; and the non-insulator target is judged as an insulator target area which is most possibly eliminated in a down classifier, so that the influence on the training process is small.
In order to balance the influence of the 2 error conditions on the classifier and improve the accuracy of classifier identification, as a further improvement of the above embodiment, the step S200 further includes:
step S220, determining the minimum classification error rate of each level of classifier in the weak classifier, and searching the minimum threshold factor of the weak classifier according to the classification error rate of each level of classifier;
step S230, finding the minimum classification error rate from the minimum classification error rates of all weak classifiers, updating the weight of the sample image input into the weak classifier for training according to the minimum classification error, and outputting a strong classifier;
step S240, setting the minimum hit rate and the maximum false alarm rate of each stage of the strong classifier, training the cascaded strong classifiers, and obtaining an improved cascaded classifier model when the hit rate of each stage of the strong classifier reaches the minimum hit rate and the false alarm rate is lower than the maximum false alarm rate.
As a further improvement of the above embodiment, the step S220 includes:
step S221, let j equal to 1,2,3.., p, where p is the total number of features;
step S222, calculating all of the sample image data setsSum of weights T of insulator samples+And the sum of the weights T of all non-insulator samples-；
Step S223, calculating the sum of the weights of all positive sample images before the sample i
Figure BDA0003212986820000063
And sum of weights of all negative sample images
Figure BDA0003212986820000064
Wherein i is a sample serial number corresponding to the positive sample image; j is a characteristic serial number corresponding to the negative sample image;
step S224, calculating according to the formula (9) to obtain a minimum classification error rate, and recording a minimum threshold factor theta corresponding to the weak classifier;
Figure BDA0003212986820000071
in the current cascade classifier training algorithm, great attention is paid to the positive sample target. If some insulators have noise or shielding phenomena in the training process, the algorithm focuses on samples with difficult classification, so that the weight of the samples is increased exponentially. As the number of iterations increases, a phenomenon of degradation of the strong classifier may result.
In the embodiment, in the weight updating algorithm, an improved weak classifier training algorithm is called first, and a minimum threshold factor theta is found through continuous iteration of the algorithm according to a redefined minimum classification error rate; and then, limiting the rising condition of the weight index of the difficult sample through a new weight updating rule. By adopting a new weight updating rule, the training and identifying efficiency of the classifier is effectively improved.
In one embodiment, with the weight update rule of the improved cascade Gentle Adaboost classifier, by calculating the average weight m and the weight standard deviation, the weight of the sample will not increase when the sample weight is greater than 3.
As a further improvement of the above embodiment, the step S230 includes:
step S231, marking N sample images as (x)1,y1)，…，(xn,yn) Wherein y isi∈{－1,1}，yi1, denoted as insulator sample; y isiA non-insulator sample is denoted by-1.
Step S232, initializing sample weight by using an equation (10);
wi＝1/N,i＝1,...,N (10)；
step S233, let J equal 1,2,3., J, where J represents the total number of iterations;
calling a weak classifier training algorithm to find out a weak classifier h with the minimum classification error from all weak classifiersj(xi) Corresponding to the minimization threshold factor θ, the weak classifier h is represented by equation (11)j(xi) The classification error of (2);
Figure BDA0003212986820000072
in the formula ofjIs the minimum classification error for the jth iteration; k is a constant number, 0<k<1；
Order to
Figure BDA0003212986820000073
Calculating the average weight of the round sample image by using the formula (13);
Figure BDA0003212986820000074
calculating a weight standard deviation by using the formula (14);
Figure BDA0003212986820000075
the weights of the sample image are updated using equation (15):
Figure BDA0003212986820000081
in the formula: alpha is alphajIs an intermediate variable; h isj(xi) The classifier corresponding to the jth iteration is represented; ziIs hj(xi) A normalization factor of the classifier; m is the training average weight;
normalization factor ZiCan be expressed as:
Figure BDA0003212986820000082
and (g) an output strong classifier g (x), as shown in formula (17):
Figure BDA0003212986820000083
the following are experiments performed according to the above-described embodiments of the present invention:
1100 positive sample images and 3100 negative sample images are selected as a current test data set, and 1000 positive sample images and 3000 negative sample images in the data set are used as a training set. In the experiment, the minimum hit rate of the strong classifier of each stage is set to be 99.5%, and the maximum false alarm rate is set to be 50%.
The relationship between the cascade layer number and the hit rate, and the relationship between the cascade layer number and the hit rate false alarm rate are shown in fig. 5 and fig. 6. In the training process, the hit rate of each stage of cascade classifier is not lower than 99%; and the false alarm rate is rapidly converged after the training layer number reaches 10, and when the training layer number is 20, the hit rate of the whole cascade classifier is 99.8%, and the false alarm rate is 7%.
In the embodiment, an improved cascade classifier model is trained, and 400 insulator images under different backgrounds are selected as a test set, wherein each image comprises a plurality of insulator targets. The test samples are all taken from the post insulators with different voltage grades on the spot, the statistics of the results are shown in table 1, in 400 test sets, the number of correct identification post insulators is 1118, and the correct identification rate reaches 91.2%.
Table 1 statistical table of test results
Figure BDA0003212986820000084
The method provided by this example is compared with several popular algorithms currently available, as shown in table 2. On the recognition accuracy under different backgrounds, the algorithm obtains better recognition results; while at recognition time, the recognition time of the algorithm herein is slightly slower than Y OLOv3 and slightly faster than FasterR-CNN.
TABLE 2 comparison of different algorithms
Figure BDA0003212986820000085
Figure BDA0003212986820000091
Corresponding to the method in fig. 1, an embodiment of the present invention further provides a computer-readable storage medium, where a computer program is stored, and when the computer program is executed by a processor, the computer program implements the steps of any one of the above methods for identifying an insulator in an infrared image.
Corresponding to the method in fig. 1, an embodiment of the present invention further provides a system for identifying an insulator in an infrared image, where the system includes:
at least one processor;
at least one memory for storing at least one program;
when the at least one program is executed by the at least one processor, the at least one processor is enabled to implement the method for identifying insulators in infrared images according to any one of the above embodiments.
The contents in the above method embodiments are all applicable to the present system embodiment, the functions specifically implemented by the present system embodiment are the same as those in the above method embodiment, and the beneficial effects achieved by the present system embodiment are also the same as those achieved by the above method embodiment.
The Processor may be a Central-Processing Unit (CPU), other general-purpose Processor, a Digital Signal Processor (DSP), an Application-Specific-Integrated-Circuit (ASIC), a Field-Programmable Gate array (FPGA) or other Programmable logic device, a discrete Gate or transistor logic device, a discrete hardware component, or the like. The general purpose processor may be a microprocessor or the processor may be any conventional processor or the like, said processor being the control centre for the insulator identification system in said infrared image, the various parts of the device being operable by the insulator identification system in the entire infrared image being connected by means of various interfaces and lines.
The memory may be configured to store the computer program and/or the module, and the processor may implement various functions of the identification system of the insulator in the infrared image by operating or executing the computer program and/or the module stored in the memory and calling data stored in the memory. The memory may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required by at least one function (such as a sound playing function, an image playing function, etc.), and the like; the storage data area may store data (such as audio data, a phonebook, etc.) created according to the use of the cellular phone, and the like. In addition, the memory may include high speed random access memory, and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a Smart-Media-Card (SMC), a Secure-Digital (SD) Card, a Flash-memory Card (Flash-Card), at least one magnetic disk storage device, a Flash memory device, or other volatile solid state storage device.
While the description of the present application has been made in considerable detail and with particular reference to a few illustrated embodiments, it is not intended to be limited to any such details or embodiments or any particular embodiments, but it is to be construed that the present application effectively covers the intended scope of the application by reference to the appended claims, which are interpreted in view of the broad potential of the prior art. Further, the foregoing describes the present application in terms of embodiments foreseen by the inventor for which an enabling description was available, notwithstanding that insubstantial changes from the present application, not presently foreseen, may nonetheless represent equivalents thereto.

Claims (8)

1. A method for identifying an insulator in an infrared image is characterized by comprising the following steps:
s100, obtaining a sample image data set of the insulator, wherein the sample image data set of the insulator comprises a plurality of sample images; the sample image comprises a positive sample image and a negative sample image, and the sample image is obtained according to infrared image preprocessing of the insulator;
s200, constructing a weak classifier, and training the constructed weak classifier according to the sample image data set to obtain an improved cascade classifier model;
step S300, receiving an infrared image of the insulator to be detected, and predicting an insulator region in the infrared image of the insulator to be detected through an improved cascade classifier model.
2. The method for identifying the insulator in the infrared image according to claim 1, wherein the step S100 comprises:
step S110, acquiring an infrared image of the transformer substation, wherein the infrared image comprises an insulator;
step S120, unifying the pixels of the infrared image, and carrying out gray processing on the infrared image with unified pixels to convert the infrared image into a gray histogram;
step S130, carrying out equalization processing on the gray level histogram to obtain a final sample image;
step S140, determining the characteristic values of the sample images by using an integral graph method, and constructing a sample image data set according to the characteristic values of the sample images.
3. The method for identifying the insulator in the infrared image according to claim 2, wherein the step S200 comprises:
step S210, determining a weak classifier; the weak classifier adopts a cascade Gentle Adaboost classifier;
step S220, determining the minimum classification error rate of each level of classifier in the weak classifier, and searching the minimum threshold factor of the weak classifier according to the classification error rate of each level of classifier;
step S230, finding the minimum classification error rate from the minimum classification error rates of all weak classifiers, updating the weight of the sample image input into the weak classifier for training according to the minimum classification error, and outputting a strong classifier;
step S240, setting the minimum hit rate and the maximum false alarm rate of each stage of the strong classifier, training the cascaded strong classifiers, and obtaining an improved cascaded classifier model when the hit rate of each stage of the strong classifier reaches the minimum hit rate and the false alarm rate is lower than the maximum false alarm rate.
4. The method according to claim 3, wherein the weak classifier is defined by the following formula:
Figure FDA0003212986810000011
in the formula: f (x)i) For the region x to be detected in the sample imageiA feature calculation function of (a); theta is a minimization threshold factor set by the weak classifier; class value alpha1And alpha2Is in the absolute value of [0,1 ]]Between, the classification value alpha1And alpha2Is proportional to the confidence of the weak classifier.
5. The method for identifying insulators in infrared images as claimed in claim 4, wherein the step S220 includes:
step S221, let j equal to 1,2,3.., p, where p is the total number of features;
step S222, calculating the number of sample imagesSum of weights T of all insulator samples in data set+And the sum of the weights T of all non-insulator samples-；
Step S223, calculating the sum of the weights of all positive sample images before the sample i
Figure FDA0003212986810000021
And sum of weights of all negative sample images
Figure FDA0003212986810000026
Wherein i is a sample serial number corresponding to the positive sample image; j is a characteristic serial number corresponding to the negative sample image;
step S224, calculating to obtain a minimum classification error rate according to the following formula, and simultaneously recording a minimum threshold factor theta corresponding to the weak classifier;
Figure FDA0003212986810000022
6. the method for identifying insulators in infrared images as claimed in claim 5, wherein the step S230 includes:
step S231, marking N sample images as (x)1,y1)，…，(xn,yn) Wherein y isi∈{－1,1}，yi1, denoted as insulator sample; y isiA non-insulator sample is denoted by-1.
Step S232, initializing sample weight by using the following formula;
wi＝1/N,i＝1,...,N；
step S233, let J equal 1,2,3., J, where J represents the total number of iterations;
calling a weak classifier training algorithm to find out a weak classifier h with the minimum classification error from all weak classifiersj(xi) The corresponding minimization threshold factor θ, which is expressed as the weak classifier h by the following formulaj(xi) The classification error of (2);
Figure FDA0003212986810000023
in the formula ofjIs the minimum classification error for the jth iteration; k is a constant number, 0<k<1；
Figure FDA0003212986810000024
Calculating the average weight of the sample images in the round by using the following formula;
Figure FDA0003212986810000025
calculating the standard deviation of the weight by using the following formula;
Figure FDA0003212986810000031
the weights of the sample images are updated using the following formula:
Figure FDA0003212986810000032
in the formula: alpha is alphajIs an intermediate variable; h isj(xi) The classifier corresponding to the jth iteration is represented; ziIs hj(xi) A normalization factor of the classifier; m is the training average weight;
normalization factor ZiCan be expressed as:
Figure FDA0003212986810000033
an output strong classifier G (x), wherein:
Figure FDA0003212986810000034
7. the utility model provides an identification device of insulator in infrared image which characterized in that includes:
at least one processor;
at least one memory for storing at least one program;
when the at least one program is executed by the at least one processor, the at least one program causes the at least one processor to implement the method for identifying insulators in infrared images according to any one of claims 1 to 6.
8. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a computer program which, when being executed by a processor, carries out the steps of the method for identifying insulators in infrared images as claimed in any one of claims 1 to 6.
CN202110936295.3A 2021-08-16 2021-08-16 Method and device for identifying insulator in infrared image and readable storage medium Pending CN113780359A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110936295.3A CN113780359A (en) 2021-08-16 2021-08-16 Method and device for identifying insulator in infrared image and readable storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110936295.3A CN113780359A (en) 2021-08-16 2021-08-16 Method and device for identifying insulator in infrared image and readable storage medium

Publications (1)

Publication Number Publication Date
CN113780359A true CN113780359A (en) 2021-12-10

Family

ID=78837906

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110936295.3A Pending CN113780359A (en) 2021-08-16 2021-08-16 Method and device for identifying insulator in infrared image and readable storage medium

Country Status (1)

Country Link
CN (1) CN113780359A (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103679677A (en) * 2013-12-12 2014-03-26 杭州电子科技大学 Dual-model image decision fusion tracking method based on mutual updating of models
CN104021394A (en) * 2014-06-05 2014-09-03 华北电力大学（保定） Insulator image recognition method based on Adaboost algorithm
CN108830311A (en) * 2018-06-01 2018-11-16 南京邮电大学 A kind of infrared image photovoltaic array recognition methods based on Adaboost
US20190094124A1 (en) * 2017-09-22 2019-03-28 Saudi Arabian Oil Company Thermography image processing with neural networks to identify corrosion under insulation (cui)

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103679677A (en) * 2013-12-12 2014-03-26 杭州电子科技大学 Dual-model image decision fusion tracking method based on mutual updating of models
CN104021394A (en) * 2014-06-05 2014-09-03 华北电力大学（保定） Insulator image recognition method based on Adaboost algorithm
US20190094124A1 (en) * 2017-09-22 2019-03-28 Saudi Arabian Oil Company Thermography image processing with neural networks to identify corrosion under insulation (cui)
CN108830311A (en) * 2018-06-01 2018-11-16 南京邮电大学 A kind of infrared image photovoltaic array recognition methods based on Adaboost

Non-Patent Citations (8)

* Cited by examiner, † Cited by third party
Title
JIAMING HAN等: "A Method of Insulator Faults Detection in Aerial Images for High-Voltage Transmission Lines Inspection", 《APPLIED SCIENCES》, vol. 9, no. 10, pages 1 - 22 *
LIAO SHAOWEN等: "A kind of improved AdaBoost algorithm", 《2014 7TH INTERNATIONAL CONFERENCE ON INTELLIGENT COMPUTATION TECHNOLOGY AND AUTOMATION》, pages 3 - 4 *
YINGJIE LUO等: "Application of Ear Detection in Non-frontal Face Recognition", 《JOURNAL OF PHYSICS: CONFERENCE SERIES》, vol. 1067, no. 1, pages 2 - 3 *
刘国特等: "基于改进级联Gentle Adaboost分类器的支柱绝缘子红外图像AI识别", 《高电压技术》, pages 1 - 9 *
周明辉: "运动车辆车牌识别算法的研究与实现", 《中国优秀硕士学位论文全文数据库：信息科技辑》, no. 2014, pages 138 - 794 *
朱喜燕等: "Gentle-Adaboost在红外视频驾驶员疲劳检测中的应用研究", 《计算机测量与控制》, vol. 23, no. 4, pages 1414 - 1417 *
胡德文等: "生物特征识别技术与方法", 31 August 2013, 国防工业出版社, pages: 26 - 27 *
赵旋: "基于红外图像的劣化绝缘子检测方法研究", 《中国优秀硕士学位论文全文数据库：信息科技辑》, no. 2017, pages 138 - 3896 *

Similar Documents

Publication Publication Date Title
CN113378632B (en) Pseudo-label optimization-based unsupervised domain adaptive pedestrian re-identification method
KR20210110823A (en) Image recognition method, training method of recognition model, and related devices and devices
Xiao et al. A fast method for particle picking in cryo-electron micrographs based on fast R-CNN
CN107038416B (en) Pedestrian detection method based on binary image improved HOG characteristics
CN108021908B (en) Face age group identification method and device, computer device and readable storage medium
CN116030237A (en) Industrial defect detection method and device, electronic equipment and storage medium
CN107358172B (en) Human face feature point initialization method based on human face orientation classification
CN110852327A (en) Image processing method, image processing device, electronic equipment and storage medium
CN113902944A (en) Model training and scene recognition method, device, equipment and medium
CN111177447B (en) Pedestrian image identification method based on depth network model
CN112926592A (en) Trademark retrieval method and device based on improved Fast algorithm
CN111597875A (en) Traffic sign identification method, device, equipment and storage medium
CN115033721A (en) Image retrieval method based on big data
CN114037886A (en) Image recognition method and device, electronic equipment and readable storage medium
CN112784494A (en) Training method of false positive recognition model, target recognition method and device
CN105844299B (en) A kind of image classification method based on bag of words
CN112560856A (en) License plate detection and identification method, device, equipment and storage medium
CN117115197A (en) Intelligent processing method and system for design data of LED lamp bead circuit board
CN115984671A (en) Model online updating method and device, electronic equipment and readable storage medium
CN111444816A (en) Multi-scale dense pedestrian detection method based on fast RCNN
CN113780359A (en) Method and device for identifying insulator in infrared image and readable storage medium
CN112288045B (en) Seal authenticity distinguishing method
CN112634869B (en) Command word recognition method, device and computer storage medium
Jeong et al. Homogeneity patch search method for voting-based efficient vehicle color classification using front-of-vehicle image
CN112329798A (en) Image scene classification method based on optimized visual bag-of-words model

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