CN112700425A - Method for judging quality of X-ray image of power equipment - Google Patents
Method for judging quality of X-ray image of power equipment Download PDFInfo
- Publication number
- CN112700425A CN112700425A CN202110017751.4A CN202110017751A CN112700425A CN 112700425 A CN112700425 A CN 112700425A CN 202110017751 A CN202110017751 A CN 202110017751A CN 112700425 A CN112700425 A CN 112700425A
- Authority
- CN
- China
- Prior art keywords
- ray
- image
- score
- quality
- picture
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 41
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 42
- 238000001514 detection method Methods 0.000 claims abstract description 40
- 238000007781 pre-processing Methods 0.000 claims abstract description 12
- 238000011176 pooling Methods 0.000 claims description 11
- 239000013598 vector Substances 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 3
- 238000003745 diagnosis Methods 0.000 abstract description 5
- 238000012549 training Methods 0.000 description 11
- 230000006870 function Effects 0.000 description 8
- 238000011161 development Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 238000013441 quality evaluation Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 3
- 238000003646 Spearman's rank correlation coefficient Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 2
- 208000003464 asthenopia Diseases 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000003213 activating effect Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a method for judging the quality of an X-ray image of power equipment, which comprises the steps of collecting pictures shot by X-rays to be scored, based on a generation countermeasure network and a target detection network, and according to an international standard, scoring each picture shot by the X-rays to obtain an MOS value; preprocessing a picture shot by an X ray to be scored, inputting the picture shot by the X ray into a convolutional neural network with VGGNET as a framework, and calculating the quality score of the picture shot by the X ray to be scored through the convolutional neural network; and calculating to obtain the final score of the picture shot by the X-ray according to the MOS value and the mass score. The method provided by the invention judges the quality of different X-ray images by utilizing the generation countermeasure network, the target detection network and the convolutional neural network taking VGGNET as a framework, can effectively solve the problems of the existing X-ray image quality judgment method, and improves the diagnosis efficiency and reliability of the detected images.
Description
Technical Field
The invention relates to the field of power equipment, in particular to a method for judging the quality of an X-ray image of the power equipment.
Background
With the development of science and technology, the smart grid construction and the enhancement of power supply reliability are raised as national strategies, the state detection and monitoring of the power equipment are used as new industries developed in recent years, huge growth potential and development space are presented, and the smart grid construction in China enters a new stage of comprehensive and rapid development. The traditional manned duty mode is gradually changed into a centralized monitoring and unmanned duty mode. Through the development and application for many years, X-ray detection becomes an important means for diagnosing the internal quality and faults of power equipment and power transmission equipment in a station, the detection range and the intelligent level are continuously improved, but due to the shortage of professional technicians, the X-ray detectors have no expert guidance when encountering technical problems in the detection process, the encountered technical problems are difficult to solve, the identification of the quality of shot images and the diagnosis of the defects are difficult, the technical assistance means such as telephone and video also have the problems of difficult communication, image distortion, low efficiency and the like, time and labor are wasted, and the professional technicians are difficult to find one by one to solve the encountered problems.
The quality of an image acquired by X-ray digital imaging is directly related to automatic detection and intelligent interpretation of subsequent defects, whether the internal structure of the power equipment can be accurately and visually displayed is related to that a detection result can be correctly judged by a detection person, the detection result seriously depends on the technical level of the detection person, and an X-ray detection robot developed at present cannot autonomously judge whether the quality of a shot image meets the diagnosis requirement, so that the intelligent degree is seriously affected, and the quality judgment technology of the X-ray image at present has the following defects: 1. the service is many, and the expert lacks: because the shooting technologies of all detection personnel are different, the traditional X-ray image quality judgment technology needs a large number of professional technicians to judge the image quality, has multiple services and low efficiency, and is very short of the professional technicians. 2. The intelligent level is low, and the cost of labor is high: the X-ray intelligent detection robot developed at present does not realize remote control over 200 meters, and a shooting system does not have a shooting quality identification function, so that the image quality needs to be judged manually, time and labor are wasted, and a professional technician who requests a far end is required to assist when encountering problems. 3. The current X-ray image quality judging method is time-consuming and labor-consuming mostly, errors are prone to occur when the traffic is large, and meanwhile, the problem of visual fatigue can occur when the image quality is judged manually, so that the judging result is influenced. 4, the technology development of the X-ray detection image in the medical industry is fast at present, but the human body structure and the tissue components are relatively fixed, the structure and the material of the power equipment are complex, the laboratory detection cannot be performed one by one, the field detection is needed, the influence factors of the quality of the X-ray detection image are more, and great difficulty is brought to the artificial intelligent identification of the image quality.
Therefore, how to design a low-cost, high-efficiency and high-reliability determination method for the quality of the X-ray image of the power equipment becomes an urgent problem to be solved.
Disclosure of Invention
The invention provides a method for judging the quality of an X-ray image of electric equipment, which aims to solve the problems of high cost, low efficiency and low reliability of the existing judging method.
The invention provides a method for judging the quality of an X-ray image of a power device, which comprises the following steps:
s1: collecting pictures shot by X-rays to be scored, and scoring each picture shot by the X-rays based on a generation countermeasure network and a target detection network in combination with international standards to obtain an MOS value;
s2: preprocessing a picture shot by an X ray to be scored, inputting the picture shot by the X ray into a convolutional neural network with VGGNET as a framework, and calculating the quality score of the picture shot by the X ray to be scored through the convolutional neural network with VGGNET as the framework;
s3: and calculating to obtain the final score of the X-ray shot picture through the MOS value and the mass score, wherein the final score of the X-ray shot picture is used for judging the quality of the X-ray image.
Optionally, the acquiring the picture taken by the X-ray to be scored, scoring the picture taken by the X-ray to be scored based on the generation countermeasure network and the target detection network in combination with the international standard to obtain the MOS value includes the following steps:
s11: acquiring a picture shot by an X-ray to be scored, and generating X-ray images with different distortion degrees by generating a countermeasure network;
s12: x-ray images with different distortion degrees and original data form a data set;
s13: detecting the double-filament positions of the X-ray images in the data set through a target detection network to obtain images with detection labels;
s14: and (4) scoring the image with the detection label by combining the international standard to obtain an MOS value.
Optionally, the preprocessing the image taken by the X-ray to be scored, inputting the image taken by the X-ray into a convolutional neural network with VGGNET as a framework, and calculating the quality score of the image taken by the X-ray to be scored through the convolutional neural network with VGGNET as a framework includes the following steps:
s21: preprocessing a picture shot by an X-ray to be scored, and intercepting an image block in the picture;
s22: inputting the image block into a convolutional neural network with VGGNET as a framework, and performing feature extraction on the image block by using the convolutional neural network with VGGNET as the framework to obtain a feature vector;
s23: calculating weights and scores of the image blocks based on the feature vectors;
s24: and calculating to obtain the quality score of the image shot by the X-ray to be scored through weighting pooling based on the weight and the score of the image block.
Optionally, after the weighting and pooling, calculating a quality score of the X-ray image to be scored further includes calculating a correction network parameter, and feeding the correction network parameter back to the convolutional neural network with VGGNET as a framework.
Optionally, the generation countermeasure network comprises a generator network and an arbiter network.
Alternatively, the MOS value is calculated by:
the average score of a single image is calculated as follows:
wherein M isiRepresents the average score, S, of the ith imagek,iRepresenting the score of the kth tester for the ith image;
judging whether the average score of the single image is in a 95% confidence interval or not, if not, deleting the data if the average score of the single image is in the 95% confidence interval, otherwise, keeping the data;
after the removal, the calculation is iterated until the obtained data are all within the 95% confidence interval, and the mean value of the data within the 95% confidence interval is taken as the MOS value.
Optionally, the final score of the X-ray captured picture is an average of the MOS value and the mass score.
The invention provides a method for judging the quality of an X-ray image of a power device, which comprises the following steps: collecting pictures shot by X-rays to be scored, and scoring each picture shot by the X-rays based on a generation countermeasure network and a target detection network in combination with international standards to obtain an MOS value; preprocessing a picture shot by an X ray to be scored, inputting the picture shot by the X ray into a convolutional neural network with VGGNET as a framework, and calculating the quality score of the picture shot by the X ray to be scored through the convolutional neural network with VGGNET as the framework; and calculating to obtain the final score of the X-ray shot picture through the MOS value and the mass score, wherein the final score of the X-ray shot picture is used for judging the quality of the X-ray image. The method provided by the invention judges the quality of different X-ray images by utilizing the generation countermeasure network, the target detection network and the convolutional neural network taking VGGNET as a framework, can effectively solve the problems of the existing X-ray image quality judgment method, and improves the diagnosis efficiency and reliability of the detected images.
Drawings
In order to more clearly illustrate the technical solution of the present invention, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious to those skilled in the art that other drawings can be obtained according to the drawings without any inventive exercise.
FIG. 1 is a flow chart of a method for determining the quality of an X-ray image of an electric power device according to the present invention;
fig. 2 is a detailed flowchart of S1 of a determination method for power device X-ray image quality according to the present invention;
fig. 3 is a detailed flowchart of S2 of a determination method for power device X-ray image quality according to the present invention;
FIG. 4 is a partial flow chart of a method for determining the quality of an X-ray image of an electric power device according to the present invention;
FIG. 5 is a flow chart of a method for determining the quality of an X-ray image of an electric power device for generating a countermeasure network according to the present invention;
fig. 6 is a schematic diagram of a convolutional neural network used in the method for determining the quality of an X-ray image of an electrical device according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention are clearly and completely described below with reference to the drawings in the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced in other ways than those specifically described, and it will be appreciated by those skilled in the art that the present invention may be embodied without departing from the spirit and scope of the invention, and therefore the present invention is not limited to the specific embodiments disclosed below.
Referring to fig. 1, the present invention provides a determination method for power equipment X-ray image quality, comprising the steps of:
s1: collecting pictures shot by X-rays to be scored, and scoring each picture shot by the X-rays based on a generation countermeasure network and a target detection network in combination with international standards to obtain an MOS value;
s2: preprocessing a picture shot by an X ray to be scored, inputting the picture shot by the X ray into a convolutional neural network with VGGNET as a framework, and calculating the quality score of the picture shot by the X ray to be scored through the convolutional neural network with VGGNET as the framework;
s3: and calculating to obtain the final score of the X-ray shot picture through the MOS value and the mass score, wherein the final score of the X-ray shot picture is used for judging the quality of the X-ray image.
Referring to fig. 2, optionally, the acquiring a picture taken by an X-ray to be scored, and scoring the picture taken by the X-ray to be scored based on the generation of the countermeasure network and the target detection network in combination with the international standard to obtain the MOS value includes the following steps:
s11: acquiring a picture shot by an X-ray to be scored, and generating X-ray images with different distortion degrees by generating a countermeasure network;
s12: x-ray images with different distortion degrees and original data form a data set;
s13: detecting the double-filament positions of the X-ray images in the data set through a target detection network to obtain images with detection labels;
s14: and (4) scoring the image with the detection label by combining the international standard to obtain an MOS value.
Referring to fig. 3, optionally, the preprocessing the picture taken by the X-ray to be scored, inputting the picture taken by the X-ray into a convolutional neural network with VGGNET as a framework, and calculating the quality score of the picture taken by the X-ray to be scored through the convolutional neural network with VGGNET as a framework includes the following steps:
s21: preprocessing a picture shot by an X-ray to be scored, and intercepting an image block in the picture;
s22: inputting the image block into a convolutional neural network with VGGNET as a framework, and performing feature extraction on the image block by using the convolutional neural network with VGGNET as the framework to obtain a feature vector;
s23: calculating weights and scores of the image blocks based on the feature vectors;
s24: and calculating to obtain the quality score of the image shot by the X-ray to be scored through weighting pooling based on the weight and the score of the image block.
Optionally, after the weighting and pooling, calculating a quality score of the X-ray image to be scored further includes calculating a modified network parameter, and feeding the modified network parameter back to the convolutional neural network.
Optionally, the generation countermeasure network comprises a generator network and an arbiter network.
Alternatively, the MOS value is calculated by:
the average score of a single image is calculated as follows:
wherein M isiRepresents the average score, S, of the ith imagek,iTo represent(ii) the score of the kth tester for the ith image;
judging whether the average score of the single image is in a 95% confidence interval or not, if not, deleting the data if the average score of the single image is in the 95% confidence interval, otherwise, keeping the data;
after the removal, the calculation is iterated until the obtained data are all within the 95% confidence interval, and the mean value of the data within the 95% confidence interval is taken as the MOS value.
Optionally, the final score of the X-ray captured picture is an average of the MOS value and the mass score.
The convolutional neural network is a part of the artificial intelligence research field and is widely applied in many fields, and the convolutional neural network can automatically learn characteristics from data and generalize results to unknown data.
Referring to fig. 4, specific embodiments are described as follows:
the method comprises the steps of collecting a plurality of X-ray shooting images, generating X-ray images with different distortion degrees by using a generation countermeasure network, and forming a data set together with original data, wherein the generation countermeasure network consists of two parts, namely a generator and a discriminator. In the training process, the two are required to be well matched. The generation of the countermeasure network GAN has a powerful capability in picture generation, and includes two parts, a generator network and a discriminator network. The generator inputs random noise for learning true distribution, the discriminator inputs a generated sample and a true sample and judges the truth of the generated image, and the whole generated countermeasure network learns the true sample distribution in continuous countermeasure training to generate vivid data.
After the data is generated, the position of the double wire is detected through the target detection network, and an image with a detection label is generated. By generating the countermeasure network, more than 2000X-ray images with different qualities are finally generated, and the original data and the X-ray images together form a required data set. Then, data of the data set is preprocessed, and the shot images are manually scored according to expert experience (manual operation) to obtain the MOS.
The MOS calculation is divided into two steps, firstly, the average score of a single image is calculated, and the formula is as follows:
wherein M isiRepresents the average score, S, of the ith imagek,iIndicating the score of the kth tester for the ith image.
Some outliers are deleted one by one. First, M is calculatediAnd at the 95% confidence interval, data outside the confidence interval is regarded as abnormal data and is removed, and after removal, the iterative computation is carried out until the obtained data are all within the confidence interval. Finally, the average value of the data is taken as the final result. These results are taken as the grountrituth of image quality.
In the model training stage, image blocks of 32X32 are densely intercepted in the image, and then a convolutional neural network with VGGNET as a framework network is adopted to extract the depth features of the image blocks to obtain depth representation vectors of the image blocks. Once the depth representation vector of the image block is obtained, a fusion sub-network is constructed, and based on the extracted features, on one hand, the weight of the image block is learned, and on the other hand, the quality evaluation score of the image block is estimated. And finally, adding a global pooling layer for obtaining the quality evaluation scores of the whole image, and performing weighted pooling operation on the quality evaluation scores of all the image blocks to obtain final image quality evaluation scores. And (5) taking the average absolute error as a loss function optimization criterion, and feeding back and correcting the network parameters. And finally, selecting a spearman rank correlation coefficient (SROCC) and a Pearson Linear Correlation Coefficient (PLCC) as a training result of the metric scale measurement model. The whole learning process is a supervised regression learning process.
And measuring the difference degree between the image quality score predicted by the convolutional neural network and the ground channel by using a loss function, if the difference is large, performing back propagation to correct parameters of the convolutional neural network, wherein the loss function L is gradually reduced along with the training, and the predicted score is more and more accurate. The loss function here directs the modification of the network parameters. With qcRepresenting the prediction fraction, qtTable representing grountruth, loss functionThe expression is as follows:
L=|qc-qt|
to increase the non-linearity of the algorithm, an activation function is added to the convolutional neural network, which should have the properties: non-linearity; continuously can be micro; the range is preferably unsaturated; monotonicity; approximately linear at the origin. The activating function selects ReLU, and the specific expression is as follows:
f(x)=max(0,x)
x represents the output of the convolution layer, the activation function can well increase the sparsity of the convolution neural network, the larger the sparsity is, the more representative the extracted features are, and the better the generalization capability is.
Generally, distortion in a region attracting the attention of a viewer is more serious than interference in other regions. This results in a local quality score y by dividing the region iiWith corresponding image block weights wiThe idea of combining the visual saliency model with IQM with weighting to the overall image quality Q. Therefore, in the training process of the convolutional neural network, on one hand, the weights of the image blocks are obtained through training, on the other hand, the scores of the image blocks are obtained through training, and the final evaluation score of the whole image is calculated as follows:
wherein, wiRepresenting the weight of the image block i, yiAnd representing the score of the image block i, obtaining the final score of the whole image by calculating a weighted average value, and judging whether the image quality is good or not according to the score. The regions that attract the viewer's attention are weighted more heavily, whereas the weights are less heavily.
The generation countermeasure network designed by the invention, the flow chart refers to fig. 5, and the X-ray image maps with different qualities are obtained through the repeated countermeasure training of the generator and the discriminator.
The network diagram of the convolutional neural network designed by the invention refers to fig. 6, and the convolutional neural network comprises 10 convolutional layers, 5 pooling layers and two full-connection layers. Every two convolutions a pooling operation is performed to extract the main features. The image features are converted to tensors by a number of convolutions. And finally, outputting the weight and the score of the image block through two fully connected layers.
The method comprises the steps of generating an confrontation network expansion data volume, inputting processed data into a neural network to extract characteristics, and obtaining a final weighted score through operations such as characteristic fusion, pooling and the like.
The image quality is judged according to the score output by the neural network, the image quality is good when the score is high, the image quality is relatively poor when the score is low, and the set scoring area is [1,10 ]. The image score predicted by the network can effectively screen out a plurality of images with poor quality and leave images with better quality.
The invention utilizes the convolutional neural network to automatically judge the quality of the X-ray detection image and automatically evaluates the shooting quality of the image according to the prediction score of the convolutional neural network. The algorithm can be suitable for various environments and different types of shot images, and can also keep good stability for images which are not seen in a training set, so that a prediction score which accords with a human visual system is given. The advantages created by the invention are as follows:
1. the method has high efficiency, effectively solves the problems of time and labor waste of more services, reduces the cost of manual identification, and can effectively utilize various data collected before to carry out iterative optimization of an algorithm model.
2. The method has high reliability, effectively solves the problems that the accuracy is reduced due to visual fatigue during manual judgment, and the judgment result seriously depends on the technical level of personnel, and is suitable for various image types and various environments.
3. The generation of the confrontation network is adopted to enlarge the data volume, improve the generalization capability of the network and solve the problems of insufficient data volume and insufficient training samples under specific conditions.
4. The X-ray image definition of equipment with complex structure and material under different thicknesses can be measured.
The invention provides a method for judging the quality of an X-ray image of a power device, which comprises the following steps: collecting pictures shot by X-rays to be scored, and scoring each picture shot by the X-rays based on a generation countermeasure network and a target detection network in combination with international standards to obtain an MOS value; preprocessing a picture shot by an X ray to be scored, inputting the picture shot by the X ray into a convolutional neural network, and calculating the quality score of the picture shot by the X ray to be scored through the convolutional neural network; and calculating to obtain the final score of the X-ray shot picture through the MOS value and the mass score, wherein the final score of the X-ray shot picture is used for judging the quality of the X-ray image. The method provided by the invention judges the quality of different X-ray images by utilizing the generation countermeasure network, the target detection network and the convolutional neural network, can effectively solve the problems of the existing X-ray image quality judgment method, and improves the diagnosis efficiency and reliability of the detected image.
The foregoing is merely a detailed description of the invention, and it should be noted that modifications and adaptations by those skilled in the art may be made without departing from the principles of the invention, and should be considered as within the scope of the invention.
Claims (7)
1. A determination method for power device X-ray image quality, characterized in that the method comprises the steps of:
s1: collecting pictures shot by X-rays to be scored, and scoring each picture shot by the X-rays based on a generation countermeasure network and a target detection network in combination with international standards to obtain an MOS value;
s2: preprocessing a picture shot by an X ray to be scored, inputting the picture shot by the X ray into a convolutional neural network with VGGNET as a framework, and calculating the quality score of the picture shot by the X ray to be scored through the convolutional neural network with VGGNET as the framework;
s3: and calculating to obtain the final score of the X-ray shot picture through the MOS value and the mass score, wherein the final score of the X-ray shot picture is used for judging the quality of the X-ray image.
2. The method for judging the quality of the X-ray image of the power equipment as claimed in claim 1, wherein the step of collecting the picture taken by the X-ray to be scored and scoring the picture taken by the X-ray to be scored based on the generation countermeasure network and the target detection network in combination with international standards to obtain the MOS value comprises the following steps:
s11: acquiring a picture shot by an X-ray to be scored, and generating X-ray images with different distortion degrees by generating a countermeasure network;
s12: x-ray images with different distortion degrees and original data form a data set;
s13: detecting the double-filament positions of the X-ray images in the data set through a target detection network to obtain images with detection labels;
s14: and (4) scoring the image with the detection label by combining the international standard to obtain an MOS value.
3. The method for determining the quality of the X-ray image of the power equipment as claimed in claim 1, wherein the preprocessing is performed on the picture of the X-ray shot to be scored, the picture of the X-ray shot is input into a convolutional neural network with a VGGNET as a framework, and the calculating of the quality score of the picture of the X-ray shot to be scored through the convolutional neural network with the VGGNET as the framework comprises the following steps:
s21: preprocessing a picture shot by an X-ray to be scored, and intercepting an image block in the picture;
s22: inputting the image block into a convolutional neural network with VGGNET as a framework, and performing feature extraction on the image block by using the convolutional neural network with VGGNET as the framework to obtain a feature vector;
s23: calculating weights and scores of the image blocks based on the feature vectors;
s24: and calculating to obtain the quality score of the image shot by the X-ray to be scored through weighting pooling based on the weight and the score of the image block.
4. The method as claimed in claim 3, wherein the step of calculating the quality score of the X-ray image to be scored after the step of weighting and pooling based on the weight and the score of the image block further comprises calculating a modified network parameter, and feeding the modified network parameter back to the convolutional neural network with VGGNET as a skeleton.
5. A determination method for power device X-ray image quality according to claim 2, characterized in that the generation countermeasure network comprises a generator network and a discriminator network.
6. The determination method for the X-ray image quality of an electric power device according to claim 2, wherein the MOS value is calculated by:
the average score of a single image is calculated as follows:
wherein M isiRepresents the average score, S, of the ith imagek,iRepresenting the score of the kth tester for the ith image;
judging whether the average score of the single image is in a 95% confidence interval or not, if not, deleting the data if the average score of the single image is in the 95% confidence interval, otherwise, keeping the data;
after the removal, the calculation is iterated until the obtained data are all within the 95% confidence interval, and the mean value of the data within the 95% confidence interval is taken as the MOS value.
7. A determination method for power device X-ray image quality according to claim 1, characterized in that the final score of the X-ray taken picture is an average of a MOS value and a quality score.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110017751.4A CN112700425B (en) | 2021-01-07 | 2021-01-07 | Determination method for X-ray image quality of power equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110017751.4A CN112700425B (en) | 2021-01-07 | 2021-01-07 | Determination method for X-ray image quality of power equipment |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112700425A true CN112700425A (en) | 2021-04-23 |
CN112700425B CN112700425B (en) | 2024-04-26 |
Family
ID=75514991
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110017751.4A Active CN112700425B (en) | 2021-01-07 | 2021-01-07 | Determination method for X-ray image quality of power equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112700425B (en) |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160358321A1 (en) * | 2015-06-05 | 2016-12-08 | Sony Corporation | Full reference image quality assessment based on convolutional neural network |
CN107743225A (en) * | 2017-10-16 | 2018-02-27 | 杭州电子科技大学 | It is a kind of that the method for carrying out non-reference picture prediction of quality is characterized using multilayer depth |
CN108648188A (en) * | 2018-05-15 | 2018-10-12 | 南京邮电大学 | A kind of non-reference picture quality appraisement method based on generation confrontation network |
CN109215028A (en) * | 2018-11-06 | 2019-01-15 | 福州大学 | A kind of multiple-objection optimization image quality measure method based on convolutional neural networks |
WO2019047949A1 (en) * | 2017-09-08 | 2019-03-14 | 众安信息技术服务有限公司 | Image quality evaluation method and image quality evaluation system |
CN109559310A (en) * | 2018-11-30 | 2019-04-02 | 国网山东省电力公司电力科学研究院 | Power transmission and transformation inspection image quality evaluating method and system based on conspicuousness detection |
CN109754391A (en) * | 2018-12-18 | 2019-05-14 | 北京爱奇艺科技有限公司 | A kind of image quality evaluating method, device and electronic equipment |
CN110570353A (en) * | 2019-08-27 | 2019-12-13 | 天津大学 | Dense connection generation countermeasure network single image super-resolution reconstruction method |
-
2021
- 2021-01-07 CN CN202110017751.4A patent/CN112700425B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160358321A1 (en) * | 2015-06-05 | 2016-12-08 | Sony Corporation | Full reference image quality assessment based on convolutional neural network |
WO2019047949A1 (en) * | 2017-09-08 | 2019-03-14 | 众安信息技术服务有限公司 | Image quality evaluation method and image quality evaluation system |
CN107743225A (en) * | 2017-10-16 | 2018-02-27 | 杭州电子科技大学 | It is a kind of that the method for carrying out non-reference picture prediction of quality is characterized using multilayer depth |
CN108648188A (en) * | 2018-05-15 | 2018-10-12 | 南京邮电大学 | A kind of non-reference picture quality appraisement method based on generation confrontation network |
CN109215028A (en) * | 2018-11-06 | 2019-01-15 | 福州大学 | A kind of multiple-objection optimization image quality measure method based on convolutional neural networks |
CN109559310A (en) * | 2018-11-30 | 2019-04-02 | 国网山东省电力公司电力科学研究院 | Power transmission and transformation inspection image quality evaluating method and system based on conspicuousness detection |
CN109754391A (en) * | 2018-12-18 | 2019-05-14 | 北京爱奇艺科技有限公司 | A kind of image quality evaluating method, device and electronic equipment |
CN110570353A (en) * | 2019-08-27 | 2019-12-13 | 天津大学 | Dense connection generation countermeasure network single image super-resolution reconstruction method |
Non-Patent Citations (1)
Title |
---|
王志强等: "生成式对抗网络的图像超分辨率重建", 西安工业大学学报, vol. 40, no. 1, pages 102 - 108 * |
Also Published As
Publication number | Publication date |
---|---|
CN112700425B (en) | 2024-04-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110598736B (en) | Power equipment infrared image fault positioning, identifying and predicting method | |
CN112990310B (en) | Artificial intelligence system and method for serving electric robot | |
CN113436169A (en) | Industrial equipment surface crack detection method and system based on semi-supervised semantic segmentation | |
CN114973002A (en) | Improved YOLOv 5-based ear detection method | |
CN114548278A (en) | In-service tunnel lining structure defect identification method and system based on deep learning | |
CN111650453A (en) | Power equipment diagnosis method and system based on windowing characteristic Hilbert imaging | |
CN108090515B (en) | Data fusion-based environment grade evaluation method | |
CN109543743B (en) | Multi-sensor fault diagnosis method for refrigerating unit based on reconstructed prediction residual error | |
CN113159334B (en) | Electrical equipment infrared image real-time detection and diagnosis method based on light-weight deep learning | |
CN113252701B (en) | Cloud edge cooperation-based power transmission line insulator self-explosion defect detection system and method | |
CN112801332B (en) | Short-term wind speed prediction method based on gray level co-occurrence matrix | |
CN105163106A (en) | Multi-data-processing video quality evaluation system | |
CN113255590A (en) | Defect detection model training method, defect detection method, device and system | |
CN111476307A (en) | Lithium battery surface defect detection method based on depth field adaptation | |
CN111830070A (en) | Automatic defect identification and judgment system and method based on edge calculation | |
CN114330120A (en) | 24-hour PM prediction based on deep neural network2.5Method of concentration | |
CN115830302B (en) | Multi-scale feature extraction fusion power distribution network equipment positioning identification method | |
CN112700425B (en) | Determination method for X-ray image quality of power equipment | |
CN112668754A (en) | Power equipment defect diagnosis method based on multi-source characteristic information fusion | |
CN110738208A (en) | efficient scale-normalized target detection training method | |
CN115931895A (en) | Method and device for checking defects of power transmission line | |
CN112508946B (en) | Cable tunnel anomaly detection method based on antagonistic neural network | |
CN115147591A (en) | Transformer equipment infrared image voltage heating type defect diagnosis method and system | |
CN110223295B (en) | Significance prediction method and device based on deep neural network color perception | |
CN113326873A (en) | Method for automatically classifying opening and closing states of power equipment based on data enhancement |
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 |