CN110827297A - Insulator segmentation method for generating countermeasure network based on improved conditions - Google Patents
Insulator segmentation method for generating countermeasure network based on improved conditions Download PDFInfo
- Publication number
- CN110827297A CN110827297A CN201911066870.8A CN201911066870A CN110827297A CN 110827297 A CN110827297 A CN 110827297A CN 201911066870 A CN201911066870 A CN 201911066870A CN 110827297 A CN110827297 A CN 110827297A
- Authority
- CN
- China
- Prior art keywords
- insulator
- segmentation
- image
- sample
- countermeasure 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.)
- Pending
Links
- 239000012212 insulator Substances 0.000 title claims abstract description 200
- 230000011218 segmentation Effects 0.000 title claims abstract description 190
- 238000000034 method Methods 0.000 title claims abstract description 64
- 238000012549 training Methods 0.000 claims abstract description 49
- 238000011176 pooling Methods 0.000 claims abstract description 11
- 238000005070 sampling Methods 0.000 claims abstract description 9
- 238000012360 testing method Methods 0.000 claims description 21
- 238000011156 evaluation Methods 0.000 claims description 16
- 238000010606 normalization Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 claims description 7
- 241000023320 Luma <angiosperm> Species 0.000 claims description 2
- OSWPMRLSEDHDFF-UHFFFAOYSA-N methyl salicylate Chemical compound COC(=O)C1=CC=CC=C1O OSWPMRLSEDHDFF-UHFFFAOYSA-N 0.000 claims description 2
- 238000007781 pre-processing Methods 0.000 claims description 2
- 238000003709 image segmentation Methods 0.000 abstract description 9
- 238000007689 inspection Methods 0.000 abstract description 6
- 230000003042 antagnostic effect Effects 0.000 abstract description 4
- 230000005540 biological transmission Effects 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 26
- 238000013135 deep learning Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 238000001514 detection method Methods 0.000 description 4
- 238000011161 development Methods 0.000 description 3
- 230000018109 developmental process Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 230000036544 posture Effects 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The invention belongs to the field of image segmentation and high-voltage transmission line inspection, and particularly relates to an insulator segmentation method, system and device for generating an antagonistic network based on improved conditions, aiming at solving the problems of poor segmentation precision and low efficiency of the insulator segmentation method for generating the antagonistic network based on the conditions. The method of the system comprises the steps of obtaining an image containing an insulator as an input image; acquiring an insulator segmentation image through an insulator segmentation model based on the input image; the insulator segmentation model is constructed by a generator for generating a countermeasure network based on conditions; the generator is constructed based on a self-encoder, and comprises an encoder and a decoder; the encoder comprises an asymmetric convolution layer and a maximum pooling layer; the decoder comprises an asymmetric convolution layer and an upper sampling layer; the training sample of the insulator segmentation model comprises an input image sample and a real segmentation image of the insulator contained in the input image sample. According to the invention, the countermeasure network is generated through improved conditions, and the segmentation precision and efficiency of the insulator are improved.
Description
Technical Field
The invention belongs to the field of image segmentation and high-voltage transmission line inspection, and particularly relates to an insulator segmentation method, system and device for generating a countermeasure network based on improved conditions.
Background
Insulators are widely used in power systems, and once damaged, a power grid is broken down, so that serious power economic loss is caused, and therefore, the insulator detection becomes an indispensable task in a power inspection process. Along with the rapid development of robots and unmanned aerial vehicles and the improvement of image detection technology, dangerous and complex artificial power inspection is gradually replaced by machines. In recent years, with the development of an artificial intelligent neural network, power routing inspection based on deep learning becomes a focus of attention of researchers in recent years, and how to accurately divide and identify insulators by using deep learning becomes an important research direction of routing inspection at present.
Image segmentation is an important research topic in computer vision. The method mainly researches a task of allocating a label to each pixel in an image and carries out pixel-level identification on a target. Image segmentation methods can be mainly classified into three major categories. First, it is based on traditional methods such as "Normalized cut", "Grab cut". The method mainly uses pixel-level bottom information during segmentation, and has low complexity of the whole algorithm, no need of training and high segmentation efficiency. But in the face of an image with a slightly complex background, auxiliary information needs to be added to assist the segmentation, otherwise the effect is not ideal.
Another very important way is a deep learning approach. At present, the image segmentation methods for deep learning mainly include a full convolution network, a self-encoder network, and a generative countermeasure network (GAN). The full convolution network FCN up-samples the feature map output by the last convolutional layer using the deconvolution layer to obtain an image with the same input size, and completes pixel-level segmentation on this image. The FCN adopts a full convolution network and becomes a classic algorithm in an image segmentation algorithm. The self-encoder method mainly uses an encoder to extract image features to obtain a feature map, and a decoder is used to divide the feature map at a pixel level. This approach is more complex and time consuming than a full convolutional network. The image segmentation completed by the generative impedance network is generally mapped into a segmentation model of an object by a generator from extracted low-dimensional features, and a discriminator network is generally symmetrical to the generator network and is used for distinguishing whether a predicted segmentation model is true or not. And continuously training to obtain a high-quality generator model, and finishing image segmentation by using the generator model obtained by training. The method has higher segmentation precision and can better improve the efficiency. However, when the method faces the insulator images with complex image backgrounds and various types and postures, the segmentation precision and efficiency are still poor.
Therefore, the present patent proposes an insulator segmentation method for generating a countermeasure network based on improved conditions, which has great advantages over the existing methods in terms of image segmentation accuracy and efficiency.
Disclosure of Invention
In order to solve the above problems in the prior art, that is, to solve the problems of poor segmentation accuracy and low efficiency of the existing insulator segmentation method for generating a countermeasure network based on conditions, a first aspect of the present invention provides an insulator segmentation method for generating a countermeasure network based on improved conditions, the method including:
step S100, acquiring an image containing an insulator as an input image;
step S200, based on the input image, obtaining an insulator segmentation image through an insulator segmentation model;
the insulator segmentation model is constructed by a generator for generating a countermeasure network cGAN based on conditions; the generator is constructed based on a self-encoder, which comprises an encoder and a decoder; the encoder comprises an asymmetric convolutional layer and a maximum pooling layer; the decoder comprises an asymmetric convolutional layer and an upper sampling layer; the training sample of the insulator segmentation model comprises an input image sample and a real segmentation image of the insulator contained in the input image sample.
In some preferred embodiments, the asymmetric convolution layer of the encoder is composed of a convolution function, a batch normalization function, a linear rectification function; the asymmetric convolution layer of the decoder is composed of a deconvolution function, a batch normalization function and a linear rectification function.
In some preferred embodiments, the insulator segmentation model is trained by:
a100, acquiring an image containing an insulator, and constructing a sample set by a preset image enhancement method, wherein the sample set comprises an input image sample and a real segmentation image of the insulator contained in the input image sample; splitting the sample set into a training sample set and a testing sample set;
step A200, obtaining an insulator segmentation image through an insulator segmentation model based on the input image sample in the training sample set; generating a segmentation image by taking the insulator as an insulator;
step A300, generating a segmentation image and an insulator real segmentation image corresponding to a training sample according to the insulator, generating a discriminator of a confrontation network cGAN through a condition to obtain a segmentation result of each region in the insulator segmentation image, and acquiring a loss value of an insulator segmentation model;
step A400, obtaining the current iteration number, if the loss value is smaller than a preset training loss value threshold value or the iteration number is larger than a preset training iteration number, outputting a trained insulator segmentation model, taking the trained insulator segmentation model as a first model, and skipping to step A500; otherwise, updating the parameters of the insulator segmentation model through a back propagation algorithm based on the loss value, adding 1 to the iteration times, and skipping to the step A200;
step A500, acquiring insulator segmentation images of all input image samples in the test sample set through the first model, and comparing the insulator segmentation images with real segmentation images of insulators contained in the test sample set to acquire an mIoU evaluation value;
and step A600, if the mIoU evaluation value is larger than a preset evaluation value, taking the first model as a finally trained insulator segmentation model, otherwise, skipping to the step A200.
In some preferred embodiments, in step a100, "the sample set is constructed by a preset image enhancement method", which includes:
acquiring an image containing an insulator as a preprocessing image sample;
based on a preset brightness multiple set, randomly selecting a brightness multiple to carry out brightness processing on the preprocessed image sample to obtain a brightness processed image sample;
rotating the preprocessed image sample to obtain a plurality of rotated image samples;
scaling the luma processed image sample and the rotated processed image sample to a preset size; based on the scaled image, a sample set is constructed.
In some preferred embodiments, the arbiter of the condition generating countermeasure network cGAN consists of five convolutional layers; the first layer of convolutional layer is composed of a convolution function and a Leaky ReLU function, the last layer of convolutional layer is composed of a convolution function, and the other three layers of convolutional layers are composed of a convolution function, a Leaky ReLU function and a batch normalization function.
In some preferred embodiments, the condition generates a matrix with an output of 16 × 16 for the arbiter of the antagonistic network cGAN.
In a second aspect of the present invention, an insulator segmentation system for generating a countermeasure network based on improved conditions is provided, the system including an acquisition module and an output module;
the acquisition module is configured to acquire an image containing an insulator as an input image;
the output module is configured to obtain an insulator segmentation image through an insulator segmentation model based on the input image;
the insulator segmentation model is constructed by a generator for generating a countermeasure network cGAN based on conditions; the generator is constructed based on a self-encoder, which comprises an encoder and a decoder; the encoder comprises an asymmetric convolutional layer and a maximum pooling layer; the decoder comprises an asymmetric convolutional layer and an upper sampling layer; the training sample of the insulator segmentation model comprises an input image sample and a real segmentation image of the insulator contained in the input image sample.
In a third aspect of the present invention, a storage device is provided, in which a plurality of programs are stored, the programs being loaded and executed by a processor to implement the above-described insulator segmentation method for generating a countermeasure network based on improved conditions.
In a fourth aspect of the present invention, a processing apparatus is provided, which includes a processor, a storage device; a processor adapted to execute various programs; a storage device adapted to store a plurality of programs; the program is adapted to be loaded and executed by a processor to implement the above-described insulator segmentation method for generating a countermeasure network based on improved conditions.
The invention has the beneficial effects that:
according to the invention, the countermeasure network is generated through improved conditions, and the segmentation precision and efficiency of the insulator are improved. The invention constructs the encoder and the decoder containing the asymmetrical convolution layer to form the self-encoder network, and the self-encoder network is used as a generator for generating the anti-network cGAN under the improved condition, thereby reducing the calculation amount of insulator segmentation and improving the segmentation efficiency.
Meanwhile, the output of a discriminator of a conditional generation countermeasure network cGAN is changed into a matrix of 16 multiplied by 16, the segmentation result of each region of the insulator segmentation image output by a generator can be distinguished, the generator (insulator segmentation model) is updated based on the result output by the discriminator, compared with the 0 and 1 values output by the existing discriminator, the trained insulator model has higher precision, and high-quality segmentation in the insulator image with complex background and various types and postures is realized.
Drawings
Other features, objects and advantages of the present application will become more apparent upon reading of the following detailed description of non-limiting embodiments thereof, made with reference to the accompanying drawings.
Fig. 1 is a schematic flow chart of an insulator segmentation method for generating a countermeasure network based on improved conditions according to an embodiment of the present invention;
FIG. 2 is a block diagram of an insulator segmentation system for generating a countermeasure network based on improved conditions in accordance with an embodiment of the present invention;
FIG. 3 is a flow chart illustrating a method for training an insulator model according to an embodiment of the invention;
FIG. 4 is an exemplary diagram of an improved conditional generation countermeasure network in accordance with one embodiment of the present invention;
FIG. 5 is an exemplary graph of a comparison of test results for different network models in accordance with one embodiment of the present invention;
FIG. 6 is an exemplary graph of the detection results of the present invention for one embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some, but not all embodiments of the present invention. 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.
The present application will be described in further detail with reference to the following drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the relevant invention and not restrictive of the invention. It should be noted that, for convenience of description, only the portions related to the related invention are shown in the drawings.
It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.
The insulator segmentation method for generating the countermeasure network based on the improved conditions, disclosed by the invention, as shown in figure 1, comprises the following steps:
step S100, acquiring an image containing an insulator as an input image;
step S200, based on the input image, obtaining an insulator segmentation image through an insulator segmentation model;
the insulator segmentation model is constructed by a generator for generating a countermeasure network cGAN based on conditions; the generator is constructed based on a self-encoder, which comprises an encoder and a decoder; the encoder comprises an asymmetric convolutional layer and a maximum pooling layer; the decoder comprises an asymmetric convolutional layer and an upper sampling layer; the training sample of the insulator segmentation model comprises an input image sample and a real segmentation image of the insulator contained in the input image sample.
In order to more clearly explain the insulator segmentation method for generating a countermeasure network based on improved conditions, the following describes in detail the steps of an embodiment of the method according to the present invention with reference to the accompanying drawings.
In the following preferred embodiment, a method for training an insulator segmentation model is first described in detail, and an insulator segmentation image is obtained by an insulator segmentation method for generating a countermeasure network based on improved conditions.
1. Training method of insulator segmentation model
The insulator segmentation model, as shown in fig. 3, includes the following steps:
a100, acquiring an image containing an insulator, and constructing a sample set by a preset image enhancement method, wherein the sample set comprises an input image sample and a real segmentation image of the insulator contained in the input image sample; splitting the sample set into a training sample set and a testing sample set;
step A200, obtaining an insulator segmentation image through an insulator segmentation model based on the input image sample in the training sample set; generating a segmentation image by taking the insulator as an insulator;
step A300, generating a segmentation image and an insulator real segmentation image corresponding to a training sample according to the insulator, generating a discriminator of a confrontation network cGAN through a condition to obtain a segmentation result of each region in the insulator segmentation image, and acquiring a loss value of an insulator segmentation model;
step A400, obtaining the current iteration number, if the loss value is smaller than a preset training loss value threshold value or the iteration number is larger than a preset training iteration number, outputting a trained insulator segmentation model, taking the trained insulator segmentation model as a first model, and skipping to step A500; otherwise, updating the parameters of the insulator segmentation model through a back propagation algorithm based on the loss value, adding 1 to the iteration times, and skipping to the step A200;
step A500, acquiring insulator segmentation images of all input image samples in the test sample set through the first model, and comparing the insulator segmentation images with real segmentation images of insulators contained in the test sample set to acquire an mIoU evaluation value;
and step A600, if the mIoU evaluation value is larger than a preset evaluation value, taking the first model as a finally trained insulator segmentation model, otherwise, skipping to the step A200.
The following is a detailed development of the above training procedure, and specifically follows:
a100, acquiring an image containing an insulator, and constructing a sample set by a preset image enhancement method, wherein the sample set comprises an input image sample and a real segmentation image of the insulator contained in the input image sample; splitting the sample set into a training sample set and a testing sample set.
Step A101, acquiring an image containing the insulator as an original image.
Step A102, in order to enrich the samples, the invention carries out brightness processing and rotation processing on the original image.
The brightness processing method comprises the following steps: the original image brightness is processed by randomly selecting two values in the range of 0.5 to 1.5 times of the original image brightness.
And (3) rotation treatment: the original image is rotated every 24 deg. to obtain 15 enhanced images per image. The invention preferably rotates every 24 degrees, and in other preferred embodiments, the rotation can be set according to actual conditions.
In the present invention, 350 insulator images were processed using the image enhancement method described above, resulting in 6000 enhanced images.
And step A103, zooming the enhanced image to 256 multiplied by 256, naming the image according to 1.jpg to 6000.jpg, labeling the sample set image to obtain an insulator segmentation graph, and finishing the manufacturing of the image sample set.
The sample set includes the input image samples, and the true segmentation images of the insulators contained therein.
And splitting the sample set into a training sample set and a testing sample set, wherein 5000 pieces of samples are used as the training sample set, and 1000 pieces of samples are used as the testing sample set.
Step A200, obtaining an insulator segmentation image through an insulator segmentation model based on the input image sample in the training sample set; the insulator is used as an insulator to generate a segmentation image.
In the present embodiment, the condition generation countermeasure network cGAN is composed based on a generator and a discriminator, as shown in fig. 4. The insulator segmentation model is constructed based on a generator of a conditional generation countermeasure network cGAN, wherein the generator is a self-encoder and comprises an encoder and a decoder, namely an encoding part and a decoding part. The sampling index of the encoded part is saved and passed to the upsampling layer, thereby reducing information loss. And taking the real segmentation image of the insulator and the image sample corresponding to the real segmentation image as well as the insulator segmentation image output by the generator and the input image sample corresponding to the insulator segmentation image as a real image pair and a false image pair. Based on the insulator segmented image output by the generator in the true and false image pairs and the true segmented image of the insulator, a 16 × 16 matrix is output by the discriminator as input, each element of the matrix representing whether the corresponding patch belongs to true or false.
Each encoder in the generator contains an asymmetric convolutional layer and a max-pooling layer. The asymmetric convolution layer is composed of a convolution function, a batch normalization function and a linear rectification function, namely the structure of Conv + BN + ReLU can deepen the complexity of the network and reduce network parameters. The decoder comprises an asymmetric convolution layer and an upper sampling layer; the decoder and the encoder are symmetrical, and a deconvolution function, a batch normalization function and a linear rectification function are adopted, namely the structure of Conv + BN + ReLU. As shown in table 1:
TABLE 1
In table 1, Set is a frame including Input, Encoder1-Encoder5 (Encoder), Decoder1-Decoder5 (Decoder), Layer name indicates the name of each Layer in the generator, RGB image is an Input image, Conv indicates a convolutional Layer, MP indicates a Max pooling Layer, Deconv indicates a deconvolution Layer, UP indicates an UpSampling Layer, Type of layers is a Layer Type, fs is a convolution kernel, window is a window size, Max-pooling is a Max-pooling, UpSampling is UpSampling, and Output size indicates an Output size.
Inputting the input image samples in the training sample set into a generator, wherein the resolution of the input image samples is 256 multiplied by 3, and obtaining an insulator segmentation image; the insulator is used as an insulator to generate a segmentation image.
When the insulator segmentation model is trained, initializing each parameter of a condition generation countermeasure network cGANThe batch of inferior feed training models was set to 8, the initial learning rate was set to 0.0001, and the optimizer parameters were β1＝0.9，β2The maximum number of training iterations is set to 50000, 0.99.
Step A300, generating a segmentation image and an insulator real segmentation image corresponding to the training sample according to the insulator, obtaining segmentation results of each region in the insulator segmentation image through a discriminator of a conditional generation countermeasure network cGAN, and obtaining a loss value of an insulator segmentation model.
In this embodiment, the discriminator for the condition generation countermeasure network cGAN mainly comprises five encoders, namely five convolutional layers, the first convolutional layer is composed of a convolution function and a leakage ReLU function; the last layer is composed of convolution functions, the other three layers of convolution layers are composed of convolution functions, Leaky ReLU functions and batch normalization functions, namely Conv + Leaky ReLU + BN, the convolution kernel adopts 4 multiplied by 4, and the step length is 2. As shown in table 2:
TABLE 2
In table 2, RGB image and generated image are the generated image of the generator and the real insulator segmentation image corresponding to the training sample.
The input of the discriminator is a generating diagram of the generator and an insulator real segmentation image corresponding to the training sample, and a matrix with the size of 16 multiplied by 16 is output, wherein each element of the matrix represents whether the corresponding patch belongs to true or false.
And generating a segmentation image and an insulator real segmentation image corresponding to the training sample according to the insulator, and generating a discriminator of the countermeasure network cGAN through the condition to obtain the segmentation result of each region in the insulator segmentation image. And carrying out feedforward conduction on the input sample in the network, comparing the generated segmentation graph with the real segmentation graph to obtain a training error, and continuously operating the read training sample through a generator and a discriminator according to the current network parameter value until the training loss values of the generator and the discriminator are output by the network.
Step A400, obtaining the current iteration number, if the loss value is smaller than a preset training loss value threshold value or the iteration number is larger than a preset training iteration number, outputting a trained insulator segmentation model, taking the trained insulator segmentation model as a first model, and skipping to step A500; otherwise, updating the parameters of the insulator segmentation model through a back propagation algorithm based on the loss value, adding 1 to the iteration times, and skipping to the step A200.
In this embodiment, whether the training is continued or terminated is determined according to the current iteration number or the loss value threshold, if the training is terminated, the trained insulator segmentation model is obtained, otherwise, based on the loss value, the variation of the antagonistic network cGAN is generated by obtaining conditions through back propagation, the parameters are updated, the iteration number is increased by 1, and the step a200 is skipped.
Step A500, obtaining the insulator segmentation images of all input image samples in the test sample set through the first model, and comparing the insulator segmentation images with the real segmentation images of the insulators contained in the test sample set to obtain an mIoU evaluation value.
In this embodiment, based on the insulator segmentation model trained in step S400, the model is tested through the test sample set. The method comprises the steps of generating insulator segmentation images of all input image samples in a test sample set, comparing the insulator segmentation images with real segmentation images of insulators contained in the test sample set to obtain an mIoU evaluation value, and counting average test time.
And step A600, if the mIoU evaluation value is larger than a preset evaluation value, taking the first model as a finally trained insulator segmentation model, otherwise, skipping to the step A200.
Generally, the higher the mlio u evaluation value is, the better the average segmentation effect is, so that if the obtained mlio u value is greater than the preset evaluation value, the model effect of the current training is better, otherwise, the retraining is performed.
In order to evaluate the insulator segmentation effect of the improved condition generation countermeasure network, the invention compares the network with other classical network models, and the experimental result is shown in table 3:
TABLE 3
In table 3, Models represent the network model used in the experiment, Ours is the improved network extracted by the present invention, that is, the present network in fig. 5 and 6, Pix2Pix, SegNet, Unet, and FCN are pixel-to-pixel Models, semantic segmentation networks, deep learning segmentation networks, and full convolution neural networks, Trainable Para (M) represents the number of training parameters, and Time represents the average test Time, and it can be seen from table 3 that the split mlio u evaluation value in the present invention is the highest, which indicates that the average splitting effect is the best. Meanwhile, the required parameters of the network are minimum, so that the complexity of the network is reduced to a great extent, and the segmentation efficiency is improved. Fig. 5 shows a final segmentation result graph. The network of the invention can improve the segmentation precision of the insulator because of adding the discriminator, and can complete pixel-level segmentation at tiny details.
In order to verify the segmentation capability of the network to insulators with different scales, an insulator segmentation experiment is performed by selecting an image with a complex background and a background object far larger than the size of the insulator as a test object. As shown in fig. 6, even though the background is very complicated and the object is larger than the size of the insulator, the present network can still accurately identify the position of the insulator and segment it with high precision. Therefore, the problem that insulator detection is difficult in a complex environment is solved by the network.
2. Insulator segmentation method for generating countermeasure network based on improved conditions
Step S100, acquiring an image including an insulator as an input image.
In the present embodiment, an image including an insulator is acquired as an input image. The image containing the insulator can be shot manually or obtained by aerial photography or other ways.
And S200, acquiring an insulator segmentation image through an insulator segmentation model based on the input image.
In this embodiment, based on the acquired image including the insulator, an insulator segmentation image is acquired through a trained insulator segmentation model.
An insulator segmentation system for generating a countermeasure network based on improved conditions according to a second embodiment of the present invention, as shown in fig. 2, includes: an acquisition module 100 and an output module 200;
the acquiring module 100 is configured to acquire an image including an insulator as an input image;
the output module 200 is configured to obtain an insulator segmentation image through an insulator segmentation model based on the input image;
the insulator segmentation model is constructed by a generator for generating a countermeasure network cGAN based on conditions; the generator is constructed based on a self-encoder, which comprises an encoder and a decoder; the encoder comprises an asymmetric convolutional layer and a maximum pooling layer; the decoder comprises an asymmetric convolutional layer and an upper sampling layer; the training sample of the insulator segmentation model comprises an input image sample and a real segmentation image of the insulator contained in the input image sample.
It can be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working process and related description of the system described above may refer to the corresponding process in the foregoing method embodiment, and details are not described herein again.
It should be noted that, the insulator segmentation system for generating a countermeasure network based on improved conditions provided in the foregoing embodiment is only illustrated by the division of the above functional modules, and in practical applications, the above functions may be allocated to different functional modules according to needs, that is, the modules or steps in the embodiment of the present invention are further decomposed or combined, for example, the modules in the foregoing embodiment may be combined into one module, or may be further split into multiple sub-modules, so as to complete all or part of the above described functions. The names of the modules and steps involved in the embodiments of the present invention are only for distinguishing the modules or steps, and are not to be construed as unduly limiting the present invention.
A storage device according to a third embodiment of the present invention stores therein a plurality of programs adapted to be loaded by a processor and to implement the above-described insulator segmentation method for generating a countermeasure network based on improved conditions.
A processing apparatus according to a fourth embodiment of the present invention includes a processor, a storage device; a processor adapted to execute various programs; a storage device adapted to store a plurality of programs; the program is adapted to be loaded and executed by a processor to implement the above-described insulator segmentation method for generating a countermeasure network based on improved conditions.
It can be clearly understood by those skilled in the art that, for convenience and brevity, the specific working processes and related descriptions of the storage device and the processing device described above may refer to the corresponding processes in the foregoing method examples, and are not described herein again.
Those of skill in the art would appreciate that the various illustrative modules, method steps, and modules described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that programs corresponding to the software modules, method steps may be located in Random Access Memory (RAM), memory, Read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. To clearly illustrate this interchangeability of electronic hardware and software, various illustrative components and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as electronic hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
The terms "first," "second," and the like are used for distinguishing between similar elements and not necessarily for describing or implying a particular order or sequence.
The terms "comprises," "comprising," or any other similar term are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
So far, the technical solutions of the present invention have been described in connection with the preferred embodiments shown in the drawings, but it is easily understood by those skilled in the art that the scope of the present invention is obviously not limited to these specific embodiments. Equivalent changes or substitutions of related technical features can be made by those skilled in the art without departing from the principle of the invention, and the technical scheme after the changes or substitutions can fall into the protection scope of the invention.
Claims (9)
1. An insulator segmentation method for creating a countermeasure network based on improved conditions, the method comprising the steps of:
step S100, acquiring an image containing an insulator as an input image;
step S200, based on the input image, obtaining an insulator segmentation image through an insulator segmentation model;
the insulator segmentation model is constructed by a generator for generating a countermeasure network cGAN based on conditions; the generator is constructed based on a self-encoder, which comprises an encoder and a decoder; the encoder comprises an asymmetric convolutional layer and a maximum pooling layer; the decoder comprises an asymmetric convolutional layer and an upper sampling layer; the training sample of the insulator segmentation model comprises an input image sample and a real segmentation image of the insulator contained in the input image sample.
2. The improved conditional generation countermeasure network-based insulator segmentation method of claim 1, wherein the asymmetric convolution layer of the encoder is composed of a convolution function, a batch normalization function, a linear rectification function; the asymmetric convolution layer of the decoder is composed of a deconvolution function, a batch normalization function and a linear rectification function.
3. The method for insulator segmentation based on improved conditional generation countermeasure network of claim 1, wherein the insulator segmentation model is trained by:
a100, acquiring an image containing an insulator, and constructing a sample set by a preset image enhancement method, wherein the sample set comprises an input image sample and a real segmentation image of the insulator contained in the input image sample; splitting the sample set into a training sample set and a testing sample set;
step A200, obtaining an insulator segmentation image through an insulator segmentation model based on the input image sample in the training sample set; generating a segmentation image by taking the insulator as an insulator;
step A300, generating a segmentation image and an insulator real segmentation image corresponding to a training sample according to the insulator, generating a discriminator of a confrontation network cGAN through a condition to obtain a segmentation result of each region in the insulator segmentation image, and acquiring a loss value of an insulator segmentation model;
step A400, obtaining the current iteration number, if the loss value is smaller than a preset training loss value threshold value or the iteration number is larger than a preset training iteration number, outputting a trained insulator segmentation model, taking the trained insulator segmentation model as a first model, and skipping to step A500; otherwise, updating the parameters of the insulator segmentation model through a back propagation algorithm based on the loss value, adding 1 to the iteration times, and skipping to the step A200;
step A500, acquiring insulator segmentation images of all input image samples in the test sample set through the first model, and comparing the insulator segmentation images with real segmentation images of insulators contained in the test sample set to acquire an mIoU evaluation value;
and step A600, if the mIoU evaluation value is larger than a preset evaluation value, taking the first model as a finally trained insulator segmentation model, otherwise, skipping to the step A200.
4. The method for insulator segmentation based on improved conditional generation countermeasure network of claim 3, wherein in step a100, "construct a sample set by a preset image enhancement method", the method comprises:
acquiring an image containing an insulator as a preprocessing image sample;
based on a preset brightness multiple set, randomly selecting a brightness multiple to carry out brightness processing on the preprocessed image sample to obtain a brightness processed image sample;
rotating the preprocessed image sample to obtain a plurality of rotated image samples;
scaling the luma processed image sample and the rotated processed image sample to a preset size; based on the scaled image, a sample set is constructed.
5. The insulator segmentation method based on the improved conditional generation countermeasure network of claim 3, wherein the discriminator of the conditional generation countermeasure network cGAN is composed of five convolutional layers; the first layer of convolutional layer is composed of a convolution function and a Leaky ReLU function, the last layer of convolutional layer is composed of a convolution function, and the other three layers of convolutional layers are composed of a convolution function, a Leaky ReLU function and a batch normalization function.
6. The method of insulator segmentation based on improved conditional generation countermeasure network of claim 5, wherein the conditional generation countermeasure network cGAN's discriminator has an output of a matrix of size 16 x 16.
7. An insulator segmentation system for generating a countermeasure network based on improved conditions is characterized by comprising an acquisition module and an output module;
the acquisition module is configured to acquire an image containing an insulator as an input image;
the output module is configured to obtain an insulator segmentation image through an insulator segmentation model based on the input image;
the insulator segmentation model is constructed by a generator for generating a countermeasure network cGAN based on conditions; the generator is constructed based on a self-encoder, which comprises an encoder and a decoder; the encoder comprises an asymmetric convolutional layer and a maximum pooling layer; the decoder comprises an asymmetric convolutional layer and an upper sampling layer; the training sample of the insulator segmentation model comprises an input image sample and a real segmentation image of the insulator contained in the input image sample.
8. A storage device having stored therein a plurality of programs, wherein said program applications are loaded and executed by a processor to implement the method of insulator segmentation for generating a countermeasure network based on improved conditions as claimed in any of claims 1-6.
9. A processing device comprising a processor, a storage device; a processor adapted to execute various programs; a storage device adapted to store a plurality of programs; characterized in that said program is adapted to be loaded and executed by a processor to implement the method of insulator segmentation for generating a countermeasure network based on improved conditions according to any of claims 1-6.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911066870.8A CN110827297A (en) | 2019-11-04 | 2019-11-04 | Insulator segmentation method for generating countermeasure network based on improved conditions |
| PCT/CN2019/117494 WO2021088101A1 (en) | 2019-11-04 | 2019-11-12 | Insulator segmentation method based on improved conditional generative adversarial network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911066870.8A CN110827297A (en) | 2019-11-04 | 2019-11-04 | Insulator segmentation method for generating countermeasure network based on improved conditions |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110827297A true CN110827297A (en) | 2020-02-21 |
Family
ID=69552793
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911066870.8A Pending CN110827297A (en) | 2019-11-04 | 2019-11-04 | Insulator segmentation method for generating countermeasure network based on improved conditions |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN110827297A (en) |
| WO (1) | WO2021088101A1 (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111429464A (en) * | 2020-03-11 | 2020-07-17 | 深圳先进技术研究院 | Medical image segmentation method, medical image segmentation device and terminal equipment |
| CN112085162A (en) * | 2020-08-12 | 2020-12-15 | 北京师范大学 | Magnetic resonance brain tissue segmentation method and device based on neural network, computing equipment and storage medium |
| CN112733756A (en) * | 2021-01-15 | 2021-04-30 | 成都大学 | Remote sensing image semantic segmentation method based on W divergence countermeasure network |
| CN112927220A (en) * | 2021-03-25 | 2021-06-08 | 郑州大学 | Method and device for detecting insulator fault based on counterstudy and terminal equipment |
| CN113762120A (en) * | 2021-08-27 | 2021-12-07 | 南京南瑞信息通信科技有限公司 | Insulator image segmentation method and device, electronic equipment and storage medium |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113627073B (en) * | 2021-07-01 | 2023-09-19 | 武汉大学 | Underwater vehicle flow field result prediction method based on improved Unet++ network |
| CN113534837B (en) * | 2021-07-06 | 2024-03-22 | 深圳市科卫泰实业发展有限公司 | Unmanned aerial vehicle flight state fusion control system and control method based on generation countermeasure network |
| CN117315354B (en) * | 2023-09-27 | 2024-04-02 | 南京航空航天大学 | Insulator anomaly detection method based on multi-discriminant composite coding GAN network |
| CN117914656B (en) * | 2024-03-13 | 2024-05-10 | 北京航空航天大学 | End-to-end communication system design method based on neural network |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107680090A (en) * | 2017-10-11 | 2018-02-09 | 电子科技大学 | Based on the electric transmission line isolator state identification method for improving full convolutional neural networks |
| CN108961235A (en) * | 2018-06-29 | 2018-12-07 | 山东大学 | A kind of disordered insulator recognition methods based on YOLOv3 network and particle filter algorithm |
| CN109166126A (en) * | 2018-08-13 | 2019-01-08 | 苏州比格威医疗科技有限公司 | A method of paint crackle is divided on ICGA image based on condition production confrontation network |
| CN110060774A (en) * | 2019-04-29 | 2019-07-26 | 赵蕾 | A kind of thyroid nodule recognition methods based on production confrontation network |
| US20190252073A1 (en) * | 2018-02-12 | 2019-08-15 | Ai.Skopy, Inc. | System and method for diagnosing gastrointestinal neoplasm |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106960178B (en) * | 2017-02-23 | 2020-02-07 | 中国科学院自动化研究所 | Training method of insulator recognition model and insulator recognition and positioning method |
| CN108520503B (en) * | 2018-04-13 | 2020-12-22 | 湘潭大学 | Face defect image restoration method based on self-encoder and generation countermeasure network |
| CN109753929B (en) * | 2019-01-03 | 2022-10-28 | 华东交通大学 | High-speed rail insulator inspection image recognition method |
| US10387753B1 (en) * | 2019-01-23 | 2019-08-20 | StradVision, Inc. | Learning method and learning device for convolutional neural network using 1×1 convolution for image recognition to be used for hardware optimization, and testing method and testing device using the same |
| CN110222837A (en) * | 2019-04-28 | 2019-09-10 | 天津大学 | A kind of the network structure ArcGAN and method of the picture training based on CycleGAN |
| CN110223304B (en) * | 2019-05-20 | 2023-01-24 | 山东大学 | Image segmentation method and device based on multipath aggregation and computer-readable storage medium |
-
2019
- 2019-11-04 CN CN201911066870.8A patent/CN110827297A/en active Pending
- 2019-11-12 WO PCT/CN2019/117494 patent/WO2021088101A1/en active Application Filing
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107680090A (en) * | 2017-10-11 | 2018-02-09 | 电子科技大学 | Based on the electric transmission line isolator state identification method for improving full convolutional neural networks |
| US20190252073A1 (en) * | 2018-02-12 | 2019-08-15 | Ai.Skopy, Inc. | System and method for diagnosing gastrointestinal neoplasm |
| CN108961235A (en) * | 2018-06-29 | 2018-12-07 | 山东大学 | A kind of disordered insulator recognition methods based on YOLOv3 network and particle filter algorithm |
| CN109166126A (en) * | 2018-08-13 | 2019-01-08 | 苏州比格威医疗科技有限公司 | A method of paint crackle is divided on ICGA image based on condition production confrontation network |
| CN110060774A (en) * | 2019-04-29 | 2019-07-26 | 赵蕾 | A kind of thyroid nodule recognition methods based on production confrontation network |
Non-Patent Citations (1)
| Title |
|---|
| 蒋芸 等: "基于条件生成对抗网络的咬翼片图像分割", 《计算机工程》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111429464A (en) * | 2020-03-11 | 2020-07-17 | 深圳先进技术研究院 | Medical image segmentation method, medical image segmentation device and terminal equipment |
| CN111429464B (en) * | 2020-03-11 | 2023-04-25 | 深圳先进技术研究院 | Medical image segmentation method, medical image segmentation device and terminal equipment |
| CN112085162A (en) * | 2020-08-12 | 2020-12-15 | 北京师范大学 | Magnetic resonance brain tissue segmentation method and device based on neural network, computing equipment and storage medium |
| CN112085162B (en) * | 2020-08-12 | 2024-02-09 | 北京师范大学 | Neural network-based magnetic resonance brain tissue segmentation method, device, computing equipment and storage medium |
| CN112733756A (en) * | 2021-01-15 | 2021-04-30 | 成都大学 | Remote sensing image semantic segmentation method based on W divergence countermeasure network |
| CN112927220A (en) * | 2021-03-25 | 2021-06-08 | 郑州大学 | Method and device for detecting insulator fault based on counterstudy and terminal equipment |
| CN113762120A (en) * | 2021-08-27 | 2021-12-07 | 南京南瑞信息通信科技有限公司 | Insulator image segmentation method and device, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2021088101A1 (en) | 2021-05-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110827297A (en) | Insulator segmentation method for generating countermeasure network based on improved conditions | |
| CN110443143B (en) | Multi-branch convolutional neural network fused remote sensing image scene classification method | |
| CN110321910B (en) | Point cloud-oriented feature extraction method, device and equipment | |
| CN111126386B (en) | Sequence domain adaptation method based on countermeasure learning in scene text recognition | |
| CN108664981B (en) | Salient image extraction method and device | |
| CN111476719A (en) | Image processing method, image processing device, computer equipment and storage medium | |
| CN111914654B (en) | Text layout analysis method, device, equipment and medium | |
| CN113095333B (en) | Unsupervised feature point detection method and unsupervised feature point detection device | |
| CN111932577B (en) | Text detection method, electronic device and computer readable medium | |
| CN112149526B (en) | Lane line detection method and system based on long-distance information fusion | |
| CN110852199A (en) | Foreground extraction method based on double-frame coding and decoding model | |
| CN114842180B (en) | Point cloud completion method, device, equipment and medium | |
| CN114742985A (en) | Hyperspectral feature extraction method and device and storage medium | |
| CN113065525A (en) | Age recognition model training method, face age recognition method and related device | |
| CN115147862A (en) | Benthonic animal automatic identification method, system, electronic device and readable storage medium | |
| CN115187530A (en) | Method, device, terminal and medium for identifying ultrasonic automatic breast full-volume image | |
| CN107729885B (en) | Face enhancement method based on multiple residual error learning | |
| CN117636298A (en) | Vehicle re-identification method, system and storage medium based on multi-scale feature learning | |
| CN113487550B (en) | Target detection method and device based on improved activation function | |
| CN113538387B (en) | Multi-scale inspection image identification method and device based on deep convolutional neural network | |
| CN112200275B (en) | Artificial neural network quantification method and device | |
| CN112800952B (en) | Marine organism identification method and system based on improved SSD algorithm | |
| CN113762120A (en) | Insulator image segmentation method and device, electronic equipment and storage medium | |
| CN115937565A (en) | Hyperspectral image classification method based on self-adaptive L-BFGS algorithm | |
| CN113920311A (en) | Remote sensing image segmentation method and system based on edge auxiliary information |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200221 |