CN109712207A - V-Net Depth Imaging method - Google Patents
V-Net Depth Imaging method Download PDFInfo
- Publication number
- CN109712207A CN109712207A CN201811411535.2A CN201811411535A CN109712207A CN 109712207 A CN109712207 A CN 109712207A CN 201811411535 A CN201811411535 A CN 201811411535A CN 109712207 A CN109712207 A CN 109712207A
- Authority
- CN
- China
- Prior art keywords
- layer
- network
- depth
- layers
- net
- 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
Abstract
The present invention relates to a kind of V-Net Depth Imaging methods, V-Net depth network structure is named as using one kind, it is a kind of 33 layer network structures, the functional module being linked in sequence by three forms, i.e. initial image-forming module, depth characteristic analysis and extraction module and Depth Imaging module, the connection between network layer use full connection, part connection, residual error connection and jump four kinds of connection types of connection, the intersection entropy loss item and L that loss function is exported in network2It is added to the intersection entropy loss item of layer 5 output on the basis of regularization constraint item, realizes the convergence process for accelerating network while initial imaging；V-Net depth network structure remains the spatial positional information of dielectric distribution in information process by the addition of pixel and shearing.
Description
Technical field
The invention belongs to tomography fields, propose a kind of novel supervised depth net of 33 layers using multi-connection mode
Network structure is used for electricity tomographic image reconstruction.
Technical background
Electricity chromatography imaging technique is the process tomographic imaging technology based on different electrical characteristics sensitive mechanisms, image-forming principle
It is space sensitive electrod-array under alternating voltage or current excitation, in the measurement sensitivity field that tested object field is formed, is tested object field
The spatial variations of interior media distribution or movement generate modulating action to sensitivity field, the electricity projection that sensor space array obtains
Information changes, and the dielectric distribution of measurand can be rebuild in conjunction with corresponding imaging algorithm, realizes visualization measurement.
Image reconstruction algorithm is electrical layer by the anti-true distribution for pushing away medium in measured zone of boundary electricity projection signal
Analyse the important branch of imaging technique, the also referred to as inverse problem of electricity imaging.Recent decades, researcher propose and have developed perhaps
More image reconstruction algorithms can substantially be divided into Class of Iterative algorithm and non-iterative class algorithm.Non-iterative linearized algorithm common are
Linear back projection, a step newton algorithm for reconstructing and a step Landweber algorithm for reconstructing etc., such method image taking speed are very fast, smart
It spends lower.When the distribution of reference conductivity rate severe deviations occurs with true distribution of conductivity, such first approximation model is not enough to
Reflection problem it is non-linear.In response to this problem using repeatedly approximate, iterative approximation method, i.e. iterative linearized algorithm.Iteration
It is higher to linearize class algorithm reconstruction precision, can be used for quantitative analysis, but requires to carry out direct problem, a spirit when each iteration
The solution of sensitive matrix, inverse problem, computational efficiency is low, is not able to satisfy the demand of real time imagery.In addition, regularization class method is solution
The certainly universal method of pathosis problem, such method depend on the selection of regularization.The priori regularization letter that artificial experience is extracted
It ceases the limited areas imaging for making this method and imaging precision is limited.Find and explore it is a kind of can automatically extract feature, while it is simultaneous
The image reconstruction algorithm for caring for the generalization ability of image taking speed, imaging precision and model, is a new research hotspot and direction.
Deep learning due to can in network training process self study useful feature, pass through the conversion and feature of layer-by-layer feature space
Effective extraction, the accuracy of regression forecasting can be efficiently promoted, to become a kind of new image reconstruction algorithm.
Summary of the invention
The purpose of the present invention is being directed to the image reconstruction problem of electricity tomography, a kind of V- based on deep learning is proposed
Net Depth Imaging method, in network model training process not only can efficient self study with from extracting useful feature, but also
The calculating for avoiding sensitivity matrix time-consuming in traditional Class of Iterative algorithm based on sensitivity, it is quick to can solve industrial process
Visual demand.Technical scheme is as follows:
A kind of V-Net Depth Imaging method is named as V-Net depth network structure using one kind, is a kind of 33 layers of net
Network structure, the functional module being linked in sequence by three form, i.e., initial image-forming module, depth characteristic analysis and extraction module and depth
Image-forming module is spent, the connection between network layer uses full connection, part connection, residual error connection and jump four kinds of connection sides of connection
Formula, the intersection entropy loss item and L that loss function is exported in network2Layer 5 output is added on the basis of regularization constraint item
Intersect entropy loss item, realizes the convergence process for accelerating network while initial imaging；V-Net depth network structure is in information processing
The spatial positional information of dielectric distribution is remained by the addition of pixel and shearing in the process, steps are as follows:
First step establishes the M group data for trained and test depth network, includes two sequences in every group of dataWherein, V is the electricity tomography boundary survey sequence for characterizing dielectric distribution projection,For quilt
Survey the true distribution series of medium inside region；
Second step designs the structure of V-Net depth network, and specific design scheme is as follows:
(1) input layer: input layer is electricity tomography boundary survey sequence V in V-Net Depth Imaging network structure, defeated
Entering layer matrix is 3 dimension matrixes, wherein the length of input layer matrix and the wide length and width for being equal to measurement sequence V, the 3rd dimension of input matrix
Indicate that the characteristic pattern number of electricity tomography boundary survey sequence V, the characteristic pattern number being originally inputted are 1；
(2) connection by the way of connecting entirely between 5 layer network layers before, and neuron number is followed successively by fully connected network network layers
812,406,250,406 and 812, the characteristic pattern number of fully connected network network layers output is 1；
First 5 layers of full articulamentum will characterize pipeline in electricity chromatographic imaging system in V-Net depth network training process
The boundary survey data Nonlinear Mapping of cross section information is the two-dimensional image vegetarian refreshments of cross-sectional image, realizes initial imaging, considers simultaneously
The pixel of image is increased to 1024 by addition pixel by the spatial positional information of dielectric distribution, is convenient for Depth Imaging network design
The extension of network structure in the process；
(3) the 6-19 layer depth structure of V-Net depth network structure gradually extends, and is mainly made of 5 convolution blocks, point
Not Wei 6-7 layers, 9-10 layers, 12-13 layers, 15-16 layers, 18-19 layers, containing there are two identical rulers in each convolution block
Spend the convolutional layer of 3 × 3 convolution kernels；It is connected between different convolution blocks by maximum pond layer, completes down-sampling, pond step-length is 2
× 2, the local maxima information in maximum pondization operation keeping characteristics space ignores other features, uses maximum 4 layers of layer of pond altogether,
Respectively the 8th layer, 11th layer, the 14th layer, the 17th layer；Based on convolution technique by the feature of 1 initial width low level during this
Figure, has gradually extracted the characteristic pattern of 1024 panel height levels, while analyzing the size of dielectric distribution in boundary survey and field domain
With position feature information, i.e. the analysis and extraction of features process of reconstruction image；
(4) Stepwize Shrink of the 20-33 layer depth network structure of V-Net depth network structure, mainly by 4 convolution blocks
It constitutes, respectively 21-22 layers, 24-25 layers, 27-28 layers, 30-31 layers, containing there are two identical rulers in each convolution block
Spend the convolutional layer of 3 × 3 convolution kernels；It is realized from de-convolution operation to up-sampling between different convolution blocks, uses warp lamination 4 altogether
Layer, respectively the 20th layer, the 23rd layer, the 26th layer, the 29th layer；The convolution kernel of 32nd layer of use 1 × 1 realizes the drop of Depth Imaging
Dimension；It is based on convolution technique during this, the number of network characterization figure gradually decreases, the recovery of simultaneous image pixel
Journey, i.e. Depth Imaging process；
(5) residual error connects: using residual error in network between the 5th layer and the 32nd layer and connects, so that depth network is in forward direction
The property of communication process operation becomes addition of matrices by matrix multiplication；In back-propagation process, no matter network travels to which
Layer, the high-rise biggish ingredient of gradient components, that is, gradient can be directly transmitted through；
(6) jump connection: the 7th layer and the 29th layer in network, the 10th layer and the 26th layer, the 13rd layer and the 23rd layer, the 16th layer
With the 20th layer between using jump connection, jump connection keeping characteristics extraction process is ignored due to being operated using maximum pondization
Local message, the information bank of complete depth image reconstruction process, so that the image rebuild is more accurate；
The design of third step loss function is as follows:
It is constituted shown in V-Net Depth Imaging network losses function such as formula (1) by three, respectively the 5th layer of output electricity of network
The intersection entropy loss L of conductance distributionout5(w), network output is the intersection entropy loss L of the 33rd layer of distribution of conductivityout33(w) and
L2Regularization term, a, b, c are respectively Lout5(w)、Lout33(w)、L2Weight coefficient:
L (w)=a*Lout5(w)+b*Lout33(w)+c*L2(w) (1)
Wherein, intersect loss item Lout5(w) and Lout33(w) it is calculated using formula (2), σjiIt is defeated for jth layer network layer
The predicted value of dielectric distribution out；
When 4th step carries out electricity tomographic image reconstruction, the boundary survey sequence conduct of electricity chromatographic imaging system
The input of trained V-Net network model, the output of V-Net network are the specific distribution of medium in tested object field.
Novel V-Net Depth Imaging method proposed by the present invention, the self study boundary survey information in a manner of supervised training
Nonlinear Mapping relationship between distribution with medium in field domain.Using full connection, part between V-Net Depth Imaging network layer
Connection, residual error connection and jump four kinds of connection types of connection constitute sequential connection it is initial be imaged, feature extraction and depth at
As three functional blocks.In addition, considering the spatial positional information of dielectric distribution in V-Net Depth Imaging network structure.Advantage is such as
Under:
1) novelty of the V-Net Depth Imaging algorithm in structure be using full connection, part connection, residual error connection with
Four kinds of connection types of jumping constitute the functional block of three sequential connections of initial imaging, feature extraction and Depth Imaging, realize
Nonlinear Mapping between boundary survey and tested object field dielectric distribution, improves the reconstruction precision of image.
2) V-Net Depth Imaging algorithm can effectively derive from study in the training process and extract different characteristic space certainly
Characteristics of image, the algorithm have certain anti-noise ability and model generalization ability.
3) V-Net Depth Imaging algorithm does not need to solve time-consuming sensitivity matrix, can be fast according to trained model
Speed solves the dielectric distribution of tested region.
Detailed description of the invention
The following drawings describes the selected embodiment of the present invention, is exemplary drawings and non exhaustive or restricted,
In:
Fig. 1 V-Net Depth Imaging network structure；
Fig. 2 five times of cross validation schematic diagrames used in inventive embodiments；
The imaging results of emulation testing data and addition different noise levels Fig. 3 of the invention；
Experiment image scene and imaging results Fig. 4 of the invention；
(a) Single bubble tests imaging results；
(b) two different size bubble imaging results；
(c) three different size bubble imaging results.
Specific embodiment
V-Net Depth Imaging method is by taking electrical resistance tomography (ERT) as an example, for solving the problems, such as the image reconstruction of ERT.It should
Method can not only be empty in different features compared with tradition is based on the regularization class image reconstruction algorithm method of sensitivity matrix
Between middle self study with from extracting useful feature, and do not need to solve time-consuming sensitivity matrix, as long as the boundary of ERT is surveyed
Sequence inputting is measured into trained V-Net network model, so that it may quick, accurate to rebuild medium in ERT field domain
True distribution, the algorithm for reconstructing have good model generalization ability and noiseproof feature.
The functional module that 33 layers of V-Net Depth Imaging network structure of shape approximation V-type are linked in sequence by three forms, i.e.,
Initial image-forming module, depth characteristic analysis and extraction module and Depth Imaging module.Connection in network structure between network layer
Using full connection, part connection, residual error connection and jump four kinds of connection types of connection.The loss function of V-Net network is in net
The intersection entropy loss item and L of network output2The intersection entropy loss item of layer 5 output is added on the basis of regularization constraint item, it is real
Now accelerate the convergence process of network while initial imaging.In addition, V-Net depth network structure passes through in information process
The addition and shearing of pixel remain the spatial positional information of dielectric distribution.Above-mentioned network structure feature enables Depth Imaging algorithm
Enough solve the problems, such as the image reconstruction from data to image.
Steps are as follows for the realization of V-Net Depth Imaging algorithm:
It include two sequences in every group of data 1. establishing the M group data for trained and test depth networkWherein, V is the boundary survey sequence for characterizing dielectric distribution projection,Inside tested region
The true distribution series of medium.
2. designing the structure of V-Net Depth Imaging network, specific design scheme is as follows:
(1) input layer: input layer is electricity tomography boundary survey sequence V in V-Net Depth Imaging network structure, defeated
Entering layer matrix is 3 dimension matrixes, wherein the length of input layer matrix and the wide length and width for being equal to measurement sequence V, the 3rd dimension of input matrix
Indicate that the characteristic pattern number of measurement sequence V, the characteristic pattern number being originally inputted are 1.
(2) connection by the way of connecting entirely between 5 layer network layers before, and neuron number is followed successively by fully connected network network layers
812,406,250,406 and 812, the characteristic pattern number of fully connected network network layers output is 1.
First 5 layers of full articulamentum will characterize pipeline in electricity chromatographic imaging system in V-Net depth network training process
The boundary survey data Nonlinear Mapping of cross section information is the two-dimensional image vegetarian refreshments of cross-sectional image, realizes initial imaging, considers simultaneously
The pixel of image is increased to 1024 (32 × 32) by addition pixel by the spatial positional information of dielectric distribution, is convenient for Depth Imaging
The extension of network structure in network design process.
(3) the 6-19 layer depth structure of V-Net depth network structure gradually extends, and is mainly made of 5 convolution blocks, point
Not Wei 6-7 layers, 9-10 layers, 12-13 layers, 15-16 layers, 18-19 layers, containing there are two identical rulers in each convolution block
Spend the convolutional layer of 3 × 3 convolution kernels；It is connected between different convolution blocks by maximum pond layer, completes down-sampling, pond step-length is 2
× 2, the local maxima information in maximum pondization operation keeping characteristics space has ignored other features, altogether using maximum pond layer 4
Layer, respectively the 8th layer, 11th layer, the 14th layer, the 17th layer；Based on convolution technique by the spy of 1 initial width low level during this
Sign figure, gradually extracted the characteristic pattern of 1024 panel height levels, at the same analyze boundary survey in field domain dielectric distribution it is big
Small and position feature information, i.e. the analysis and extraction of features process of reconstruction image.
(4) Stepwize Shrink of the 20-33 layer depth network structure of V-Net depth network structure, mainly by 4 convolution blocks
It constitutes, respectively 21-22 layers, 24-25 layers, 27-28 layers, 30-31 layers, containing there are two identical rulers in each convolution block
Spend the convolutional layer of 3 × 3 convolution kernels；It is realized from de-convolution operation to up-sampling between different convolution blocks, uses warp lamination 4 altogether
Layer, respectively the 20th layer, the 23rd layer, the 26th layer, the 29th layer；The convolution kernel of 32nd layer of use 1 × 1 realizes the drop of Depth Imaging
Dimension；It is based on convolution technique during this, the number of network characterization figure gradually decreases, the recovery of simultaneous image pixel
Journey, i.e. Depth Imaging process.
(5) residual error connects: using residual error in network between the 5th layer and the 32nd layer and connects, so that depth network is in forward direction
The property of communication process operation becomes addition of matrices by matrix multiplication, and calculating becomes simpler.In back-propagation process, net
Which layer no matter network travel to, and the high-rise biggish ingredient of gradient components, that is, gradient can be directly transmitted through.Such propagation
Mode, which makes the gradient of the depth network in back-propagation process decay, to be further inhibited, and the performance of network is more stable.
(6) jump connection: the 7th layer and the 29th layer in network, the 10th layer and the 26th layer, the 13rd layer and the 23rd layer, the 16th layer
With the 20th layer between using jump connection.Jump connection remains characteristic extraction procedure due to ignoring using maximum pondization operation
Local message, the complete information bank of depth image reconstruction process, so that the image rebuild is more accurate.
The design of the loss function of 3.V-Net Depth Imaging network is as follows:
It is constituted shown in V-Net Depth Imaging network losses function such as formula (1) by three, respectively the 5th layer of output electricity of network
The intersection entropy loss L of conductance distributionout5(w), network output is the intersection entropy loss L of the 33rd layer of distribution of conductivityout33(w) and
L2Regularization term.A, b, c are respectively Lout5(w)、Lout33(w)、L2Weight coefficient.
L (w)=a*Lout5(w)+b*Lout33(w)+c*L2(w) (1)
Wherein, intersect loss item Lout5(w) and Lout33(w) it is calculated using formula (2).True point of medium
Cloth, σjiFor the predicted value of the dielectric distribution of jth layer network layer output.
4. be used for electricity tomographic image reconstruction using V-Net Depth Imaging algorithm, electricity chromatographic imaging system
Input of the boundary survey sequence as trained V-Net network model, the output of V-Net network are to be situated between in tested object field
The specific distribution of matter.
It is described in detail below to manufacture and operate step of the invention, it is intended to be described as the embodiment of the present invention, be not
The unique forms that can be manufactured or be utilized can realize that the embodiment of identical function should also be included in the scope of the present invention to other
It is interior.
Below with reference to specification annex, the preferred embodiments of the present invention are described in detail.
(1) 40000 group data sets are constructed using 16 electrode ERT simulation model of adjacent current excitation-voltage measurement, be used for
Training and measurement V-Net Depth Imaging model, wherein including a boundary survey voltage value sequence of ERT system in each group of data
True distribution of conductivity sequence in column and a field domain.Include different size, different numbers, the bubble of different location in data set
Distribution.
(2) 40000 training set datas are divided into 5 equal portions, will wherein 4 equal portions totally 36000 pairs of data as training set
Training V-Net network, remaining 1 part generalization ability of totally 4000 pairs of data as test set test model, successively alternately.I.e.
5 models have been respectively trained in 5 different data sets using 5 times of cross validation methods shown in Fig. 2, have been carried out simultaneously
The test of model.Select one of data set as the input of network, training process is as follows:
(a) V-Net depth network structure proposed by the present invention is constructed for the simulation data base of ERT.The network structure is total
There are 33 layers, as shown in figure 1 each layer of each rectangle expression network, wherein input of the ERT boundary survey contact potential series as network,
It is the number that ERT measure contact potential series that the number of input layer, which is 208, the digital representation layer right above each rectangle
Network exports the number of characteristic pattern, and the ERT contact potential series characteristic pattern number of network inputs is 1；First 5 layers are full articulamentum, each
The number of neuron in the digital representation of the rectangle lower left layer network is respectively 812 from the 1st layer to the 5th layer, 406,250,
406 and 812, the number of characteristic pattern is all 1 in fully connected network network layers；Totally 27 layers of the locally-attached network layer of V-Net network, wherein
Comprising 9 convolution blocks, each convolution block includes the convolutional layer that two layers of convolution kernel is 3 × 3, amounts to 18 layers of convolutional layer；Feature extraction
5 convolution blocks are used in the process, and the characteristic pattern number that a convolutional layer is connected in each convolution block is equal, according to the sequence of network layer
The characteristic pattern number of convolutional layer is followed successively by 64,128,256,512,1024 in different convolution blocks, uses pond between each convolution block
Change the maximum pond layer that region is 2 × 2 to connect, use 4 layers of maximum pond layer altogether, characteristic pattern number is followed successively by 64,128,
256,512；Depth Imaging process uses 4 convolution blocks, according to the feature of convolutional layer in the sequence difference convolution block of network layer
Figure number is followed successively by 512,256,128,64, uses warp lamination between each convolution block, uses 4 layers of warp lamination altogether,
The number of characteristic pattern is followed successively by 512,256,128,64；32nd layer uses convolution kernel for 1 × 1 convolutional layer, by depth reconstruction
Image number is reduced to 1 by 64, and the last layer is the output layer of whole network.The activation primitive of every layer network is repaired using unsaturation
Linear positive function Relu.
(b) parameters in network are initialized:
The weight w of each layer of networkm0: random number (mean value 0, variance 0.01)；Deviation bm0: 0.1；
Initial learning rate: η0=0.01；Learning rate attenuation rate: ρ=0.99；Lot number: batch=100；
Every coefficient in loss function: a=1, b=1, c=0.0001；Momentum: γ=0.99；
Total the number of iterations: steps=60000；
(b) input of the ERT boundary survey contact potential series as network, input matrix are 100 × 208 × 1, input stream
Initial conductivity, which is rebuild, through 5 layers of realization before V-Net feedforward network is distributed σ5i, initial conductivity be distributed σ5iInformation passes through V-Net net
The self study that 6-19 layer of network with from after excavating imaging depth feature, finally 20-33 layers of completion depth of V-Net network at
As process reconstructs more accurate ERT field domain internal conductance rate distribution σ in the output end of network33i.Calculate separately network the 5th, 33
The intersection entropy loss L of layer output conductance rate distributionout5(w) and Lout33(w)。
(c) loss function of V-Net network is calculated
L (w)=a*Lout 5(w)+b*Lout 33(w)+c*L2(w) (2)
(d) loss function is calculated to the gradient of each parameter using chain type Rule for derivation in network backpropagation, in conjunction with
Learning rate updates the weight w of each layer network using small lot momentum stochastic gradient descent methodmWith deviation bm, renewal equation is such as
Formula (3).
Wherein learning rate η updates in the way of exponential damping in formula (4):
η=η0*ρsteps/batch (4)
(e) step (a)~(c) is repeated, the number of iterations of network training is equal to steps, and model training stops, and saves mould
Type.
(3) trained 5 models are tested in respective measurement concentration respectively, and is calculated according to formula (5)-(6)
Corresponding image error and related coefficient out select image error minimum, and the maximum model of related coefficient is as V-Net network
Final mask.
WhereinFor true conductivityAverage value,For neural network forecast conductivityσiAverage value.
(4) using the experiment boundary survey contact potential series of different distributions as the network inputs of (3) step preference pattern, network
Output be field domain internal conductance rate true distribution.
Emulation experiment is carried out to verify validity and the noise immunity of inventive algorithm.During emulation experiment in test set
The random noise of 20-60dB is added respectively, and the simulation imaging result of different medium distribution is as shown in figure 3, first is classified as true point
Cloth, second is classified as the imaging results for not adding noise, and third arranges the imaging results being sequentially increased to the 4th column noise.It can by Fig. 3
Know, V-Net Depth Imaging algorithm can be used to solve the problems, such as the image reconstruction of ERT, while the algorithm has certain noise immunity.
Shown in Fig. 4, is tested for different medium distributed model and V-Net Depth Imaging algorithm is verified.Due to
The conductivity of PVC material and the conductivity of water choose various sizes of PVC stick simulating in experiment and are tested area there are biggish difference
The measured medium of different distributions in domain, tap water is as the background media in tested region.The PVC of two scales is chosen in experiment
Stick, diameter are respectively 21.4mm and 30.0mm, and the pipe diameter of tested region is 125mm.There are biggish in experimentation
System noise and random noise, this experiment not only demonstrate feasibility of the V-Net Depth Imaging algorithm in tomographic process
With applicability, while the noiseproof feature of the algorithm and the generalization ability of model are demonstrated.
(a) is the experiment scene for single isolated bubbles in pipeline in Fig. 4, quick and precisely using V-Net Depth Imaging algorithm
The distribution of conductivity figure reconstructed.The image error and related coefficient of the imaging results are respectively 4%, 99%.
(b) is the experiment scene for two bubbles of different sizes in pipeline in Fig. 4, using V-Net Depth Imaging algorithm
The distribution of conductivity figure quick and precisely reconstructed.
(c) is the experiment scene for three bubbles of different sizes in pipeline in Fig. 4, using V-Net Depth Imaging algorithm
The distribution of conductivity figure quick and precisely reconstructed.
Claims (1)
- It is a kind of 33 layer networks 1. a kind of V-Net Depth Imaging method is named as V-Net depth network structure using one kind Structure, the functional module being linked in sequence by three form, i.e., initial image-forming module, depth characteristic analysis and extraction module and depth Image-forming module, the connection between network layer use full connection, part connection, residual error connection and jump four kinds of connection sides of connection Formula, the intersection entropy loss item and L that loss function is exported in network2Layer 5 output is added on the basis of regularization constraint item Intersect entropy loss item, realizes the convergence process for accelerating network while initial imaging；V-Net depth network structure is in information processing The spatial positional information of dielectric distribution is remained by the addition of pixel and shearing in the process, steps are as follows:First step establishes the M group data for trained and test depth network, includes two sequences in every group of dataWherein, V is the electricity tomography boundary survey sequence for characterizing dielectric distribution projection,For quilt Survey the true distribution series of medium inside region；Second step designs the structure of V-Net depth network, and specific design scheme is as follows:(1) input layer: input layer is electricity tomography boundary survey sequence V, input layer in V-Net Depth Imaging network structure Matrix is 3 dimension matrixes, wherein the length of input layer matrix and the wide length and width for being equal to measurement sequence V, the 3rd dimension table of input matrix show The characteristic pattern number of electricity tomography boundary survey sequence V, the characteristic pattern number being originally inputted are 1；(2) before between 5 layer network layers connection by the way of connecting entirely, in fully connected network network layers neuron number be followed successively by 812, 406,250,406 and 812, the characteristic pattern number of fully connected network network layers output is 1；First 5 layers of full articulamentum will characterize pipeline section in electricity chromatographic imaging system in V-Net depth network training process The boundary survey data Nonlinear Mapping of information is the two-dimensional image vegetarian refreshments of cross-sectional image, realizes initial imaging, while considering medium The pixel of image is increased to 1024 by addition pixel by the spatial positional information of distribution, is convenient for Depth Imaging network design process The extension of middle network structure；(3) the 6-19 layer depth structure of V-Net depth network structure gradually extends, and is mainly made of 5 convolution blocks, respectively 6-7 layers, 9-10 layers, 12-13 layers, 15-16 layers, 18-19 layers, containing there are two same scales 3 in each convolution block The convolutional layer of × 3 convolution kernels；It being connected between different convolution blocks by maximum pond layer, completes down-sampling, pond step-length is 2 × 2, The local maxima information in maximum pondization operation keeping characteristics space, ignores other features, altogether using maximum 4 layers of layer of pond, respectively For the 8th layer, 11th layer, the 14th layer, the 17th layer；Based on convolution technique by the characteristic pattern of 1 initial width low level during this, by Step has extracted the characteristic pattern of 1024 panel height levels, while analyzing the size of dielectric distribution and position in boundary survey and field domain Characteristic information, i.e. the analysis and extraction of features process of reconstruction image；(4) Stepwize Shrink of the 20-33 layer depth network structure of V-Net depth network structure, mainly by 4 convolution block structures At respectively 21-22 layers, 24-25 layers, 27-28 layers, 30-31 layers, containing there are two same scales in each convolution block The convolutional layer of 3 × 3 convolution kernels；It is realized from de-convolution operation to up-sampling between different convolution blocks, uses warp lamination 4 altogether Layer, respectively the 20th layer, the 23rd layer, the 26th layer, the 29th layer；The convolution kernel of 32nd layer of use 1 × 1 realizes the drop of Depth Imaging Dimension；It is based on convolution technique during this, the number of network characterization figure gradually decreases, the recovery of simultaneous image pixel Journey, i.e. Depth Imaging process；(5) residual error connects: using residual error in network between the 5th layer and the 32nd layer and connects, so that depth network is in propagated forward The property of process operation becomes addition of matrices by matrix multiplication；In back-propagation process, which layer no matter network travel to, high The biggish ingredient of gradient components, that is, gradient of layer can be directly transmitted through；(6) jump connection: in network the 7th layer with the 29th layer, the 10th layer with the 26th layer, the 13rd layer with the 23rd layer, the 16th layer and the It is connected between 20 layers using jump, the part that jump connection keeping characteristics extraction process is ignored due to being operated using maximum pondization Information, the information bank of complete depth image reconstruction process, so that the image rebuild is more accurate；The design of third step loss function is as follows:It is constituted shown in V-Net Depth Imaging network losses function such as formula (1) by three, respectively the 5th layer of output conductance rate of network The intersection entropy loss L of distributionout5(w), network output is the intersection entropy loss L of the 33rd layer of distribution of conductivityout33(w) and L2Just Then change item, a, b, c are respectively Lout5(w)、Lout33(w)、L2Weight coefficient:L (w)=a*Lout5(w)+b*Lout33(w)+c*L2(w) (1)Wherein, intersect loss item Lout5(w) and Lout33(w) it is calculated using formula (2), σjiFor the output of jth layer network layer The predicted value of dielectric distribution；When 4th step carries out electricity tomographic image reconstruction, the boundary survey sequence of electricity chromatographic imaging system, which is used as, has been instructed The input for the V-Net network model perfected, the output of V-Net network are the specific distribution of medium in tested object field.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811411535.2A CN109712207B (en) | 2018-11-24 | 2018-11-24 | V-Net depth imaging method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811411535.2A CN109712207B (en) | 2018-11-24 | 2018-11-24 | V-Net depth imaging method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109712207A true CN109712207A (en) | 2019-05-03 |
| CN109712207B CN109712207B (en) | 2023-04-07 |
Family
ID=66255074
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811411535.2A Active CN109712207B (en) | 2018-11-24 | 2018-11-24 | V-Net depth imaging method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109712207B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110580727A (en) * | 2019-08-27 | 2019-12-17 | 天津大学 | depth V-shaped dense network imaging method with increased information flow and gradient flow |
| CN111951400A (en) * | 2020-08-06 | 2020-11-17 | 天津大学 | Narrow-beam X-ray excitation luminescence tomography method based on U-net network |
| CN113870377A (en) * | 2021-10-21 | 2021-12-31 | 天津科技大学 | Lung imaging method based on V-ResNet |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017215284A1 (en) * | 2016-06-14 | 2017-12-21 | 山东大学 | Gastrointestinal tumor microscopic hyper-spectral image processing method based on convolutional neural network |
| WO2018028255A1 (en) * | 2016-08-11 | 2018-02-15 | 深圳市未来媒体技术研究院 | Image saliency detection method based on adversarial network |
| CN108062756A (en) * | 2018-01-29 | 2018-05-22 | 重庆理工大学 | Image, semantic dividing method based on the full convolutional network of depth and condition random field |
| CN108171701A (en) * | 2018-01-15 | 2018-06-15 | 复旦大学 | Conspicuousness detection method based on U networks and confrontation study |
-
2018
- 2018-11-24 CN CN201811411535.2A patent/CN109712207B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017215284A1 (en) * | 2016-06-14 | 2017-12-21 | 山东大学 | Gastrointestinal tumor microscopic hyper-spectral image processing method based on convolutional neural network |
| WO2018028255A1 (en) * | 2016-08-11 | 2018-02-15 | 深圳市未来媒体技术研究院 | Image saliency detection method based on adversarial network |
| CN108171701A (en) * | 2018-01-15 | 2018-06-15 | 复旦大学 | Conspicuousness detection method based on U networks and confrontation study |
| CN108062756A (en) * | 2018-01-29 | 2018-05-22 | 重庆理工大学 | Image, semantic dividing method based on the full convolutional network of depth and condition random field |
Non-Patent Citations (3)
| Title |
|---|
| XIZI SONG: "Sensitivity Matrix for Ultrasound Modulated Electrical Impedance Tomography" * |
| 丁永维;董峰;: "电阻层析成像技术正则化图像重建算法研究" * |
| 赵佳;董峰;: "非完整ERT数据的两相层状流分布图像重建" * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110580727A (en) * | 2019-08-27 | 2019-12-17 | 天津大学 | depth V-shaped dense network imaging method with increased information flow and gradient flow |
| CN110580727B (en) * | 2019-08-27 | 2023-04-18 | 天津大学 | Depth V-shaped dense network imaging method with increased information flow and gradient flow |
| CN111951400A (en) * | 2020-08-06 | 2020-11-17 | 天津大学 | Narrow-beam X-ray excitation luminescence tomography method based on U-net network |
| CN113870377A (en) * | 2021-10-21 | 2021-12-31 | 天津科技大学 | Lung imaging method based on V-ResNet |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109712207B (en) | 2023-04-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109598768B (en) | Electrical tomography image reconstruction method based on convolutional neural network | |
| Tan et al. | Image reconstruction based on convolutional neural network for electrical resistance tomography | |
| Lefik et al. | Artificial neural network as an incremental non-linear constitutive model for a finite element code | |
| Fan et al. | Numerical solutions of waves-current interactions by generalized finite difference method | |
| CN109712207A (en) | V-Net Depth Imaging method | |
| CN106503801A (en) | Electrical impedance tomography method based on deep learning | |
| CN109674471A (en) | A kind of electrical impedance imaging method and system based on generation confrontation network | |
| CN110580727B (en) | Depth V-shaped dense network imaging method with increased information flow and gradient flow | |
| Dehghan et al. | Meshless simulation of stochastic advection–diffusion equations based on radial basis functions | |
| CN106203625A (en) | A kind of deep-neural-network training method based on multiple pre-training | |
| CN110222828A (en) | A kind of Unsteady Flow method for quick predicting based on interacting depth neural network | |
| Liu et al. | Application of neural networks in multiphase flow through porous media: Predicting capillary pressure and relative permeability curves | |
| Alkayem et al. | A new self-adaptive quasi-oppositional stochastic fractal search for the inverse problem of structural damage assessment | |
| Bera et al. | A MATLAB-based boundary data simulator for studying the resistivity reconstruction using neighbouring current pattern | |
| CN108665546B (en) | Multi-point geostatistical three-dimensional modeling method combined with deep learning | |
| Furnari et al. | Asynchronous cellular automata subsurface flow simulations in two-and three-dimensional heterogeneous soils | |
| CN111241725B (en) | Structure response reconstruction method for generating countermeasure network based on conditions | |
| CN114155354B (en) | Method and device for reconstructing capacitance tomography based on graph convolution network | |
| Gu et al. | Anomalous sub-diffusion equations by the meshless collocation method | |
| CN113870377A (en) | Lung imaging method based on V-ResNet | |
| Vollant et al. | Optimal estimator and artificial neural network as efficient tools for the subgrid-scale scalar flux modeling | |
| CN111444614A (en) | Flow field reconstruction method based on graph convolution | |
| Swetha et al. | Flow and transport parameter estimation of a confined aquifer using simulation–optimization model | |
| Nagel et al. | Studying first passage problems using neural networks: A case study in the slit-well microfluidic device | |
| CN117150243B (en) | Fault isolation and estimation method based on fault influence decoupling network |
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 |