CN111862251B - Method, device, storage medium and electronic equipment for medical image reconstruction technology - Google Patents
Method, device, storage medium and electronic equipment for medical image reconstruction technology Download PDFInfo
- Publication number
- CN111862251B CN111862251B CN202010565805.6A CN202010565805A CN111862251B CN 111862251 B CN111862251 B CN 111862251B CN 202010565805 A CN202010565805 A CN 202010565805A CN 111862251 B CN111862251 B CN 111862251B
- Authority
- CN
- China
- Prior art keywords
- image
- generator
- training
- discriminator
- trained
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 76
- 238000003860 storage Methods 0.000 title claims abstract description 14
- 238000005516 engineering process Methods 0.000 title abstract description 11
- 238000012549 training Methods 0.000 claims abstract description 129
- 239000011159 matrix material Substances 0.000 claims description 73
- 239000013598 vector Substances 0.000 claims description 20
- 230000006870 function Effects 0.000 claims description 18
- 238000004590 computer program Methods 0.000 claims description 16
- 230000003416 augmentation Effects 0.000 claims description 7
- 238000004422 calculation algorithm Methods 0.000 claims description 7
- 230000008569 process Effects 0.000 description 11
- 238000002595 magnetic resonance imaging Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 238000012795 verification Methods 0.000 description 5
- 238000013527 convolutional neural network Methods 0.000 description 4
- 230000004044 response Effects 0.000 description 4
- 238000010276 construction Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000003384 imaging method Methods 0.000 description 3
- 208000024827 Alzheimer disease Diseases 0.000 description 2
- 230000004913 activation Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000011478 gradient descent method Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 230000003190 augmentative effect Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000002591 computed tomography Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000002059 diagnostic imaging Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 210000001320 hippocampus Anatomy 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 231100000915 pathological change Toxicity 0.000 description 1
- 230000036285 pathological change Effects 0.000 description 1
- 238000004904 shortening Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 210000003478 temporal lobe Anatomy 0.000 description 1
- 238000003325 tomography Methods 0.000 description 1
- 238000002604 ultrasonography Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/003—Reconstruction from projections, e.g. tomography
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
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)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The application is applicable to the technical field of image reconstruction, and provides a method, a device, a storage medium and electronic equipment for medical image reconstruction technology, which comprise the following steps: acquiring a sample training data set, wherein the sample training data set comprises a first image and a second image, the first image is a low-resolution image, and the second image is a high-resolution image corresponding to the first image; the generator performs image reconstruction on the first image to obtain a third image, wherein the third image is a reconstructed image corresponding to the first image, and the generator is a pre-trained width learning network; inputting the second image and the third image into the discriminator, adjusting model parameters of the generator according to the output result of the discriminator until the preset training condition is met, obtaining a trained generator, and reconstructing the medical image by using the trained generator. The application can shorten the training time of the medical image reconstruction model and improve the efficiency of medical image reconstruction.
Description
Technical Field
The application belongs to the technical field of image reconstruction, and particularly relates to a method, a device, a storage medium and electronic equipment for medical image reconstruction technology.
Background
In recent 30 years, high-end medical imaging technologies and equipment represented by magnetic resonance imaging (Magnetic Resonance Imaging, MRI), computerized tomography (Computed Tomography, CT) and ultrasound have been developed, functions and performances have been improved, and inspection technologies and methods have been innovated. However, the quality of MRI/CT images is not always satisfactory, and there are problems such as blurring, partial omission, etc. in a large number of images, and such images often need to be reconstructed or enhanced before they can be used for diagnosis.
In the existing medical image reconstruction technology, a convolutional neural network (Convolutional Neural Networks, CNN) is mostly adopted to construct and train a reconstruction deep learning reconstruction model for image reconstruction. However, model training using a deep learning reconstruction model of a convolutional neural network is long, model performance is poor, and medical image reconstruction efficiency is low.
Disclosure of Invention
The embodiment of the application provides a method, a device, a storage medium and electronic equipment for medical image reconstruction technology, which can solve the problems of long training time of an image reconstruction model, poor model performance and low medical image reconstruction efficiency in the prior art.
In a first aspect, an embodiment of the present application provides a method for medical image reconstruction techniques, including:
Acquiring a sample training data set, wherein the sample training data set comprises a first image and a second image, the first image is a low-resolution image, and the second image is a high-resolution image corresponding to the first image;
The generator performs image reconstruction on the first image to obtain a third image, wherein the third image is a reconstructed image corresponding to the first image, and the generator is a pre-trained width learning network;
Inputting the second image and the third image into a discriminator, and adjusting model parameters of the generator according to the output result of the discriminator until a preset training condition is met, so as to obtain a trained generator;
medical image reconstruction is performed using the trained generator.
In a possible implementation manner of the first aspect, the step of performing image reconstruction on the first image by the generator to obtain a third image includes:
Extracting linear characteristics of the first image to obtain characteristic nodes;
Carrying out nonlinear characteristic enhancement on the characteristic nodes to obtain enhancement nodes;
and performing image reconstruction according to the feature nodes, the enhancement nodes and the pre-training model parameters of the pre-trained width learning network to generate a third image.
In a possible implementation manner of the first aspect, the step of reconstructing an image according to the feature node, the enhancement node, and the pre-trained model parameters of the pre-trained breadth-learning network to generate a third image includes:
constructing an input matrix based on the feature nodes and the enhancement nodes;
constructing a pre-training parameter matrix based on the pre-training model parameters;
determining an output matrix of the pre-trained generator according to the input matrix and the pre-trained parameter matrix;
and reconstructing an image based on the output matrix to obtain a third image.
In a possible implementation manner of the first aspect, the step of adjusting model parameters of the generator according to an output result of the arbiter includes:
calculating the loss values of the third image and the second image according to a preset loss function;
and adjusting model parameters of the generator according to the loss value.
In a possible implementation manner of the first aspect, the width learning network includes an output layer, and the step of adjusting model parameters of the generator according to the loss value includes:
calculating a gradient vector of the loss value at the output layer according to a back propagation algorithm;
and adjusting model parameters of the generator according to the gradient vector.
In a possible implementation manner of the first aspect, the pre-training of the breadth-learning network includes obtaining pre-training model parameters, specifically including:
extracting linear characteristics of the first image to obtain a first image characteristic node;
performing nonlinear characteristic enhancement on the first image characteristic node to obtain a first image enhancement node;
Constructing a model input matrix based on the first image feature node and the first image enhancement node;
Extracting linear characteristics of the second image to obtain second image characteristic nodes;
Constructing a model output matrix according to the second image characteristic nodes;
And obtaining the pre-training model parameters of the width learning network according to the model input matrix and the model output matrix.
In a second aspect, an embodiment of the present application provides an apparatus for medical image reconstruction techniques, including:
The system comprises a sample data acquisition unit, a first image acquisition unit and a second image acquisition unit, wherein the sample data acquisition unit is used for acquiring a sample training data set, the sample training data set comprises a first image and a second image, the first image is a low-resolution image, and the second image is a high-resolution image corresponding to the first image;
The first training unit is used for carrying out image reconstruction on the first image by the generator to obtain a third image, wherein the third image is a reconstructed image corresponding to the first image, and the generator is a pre-trained width learning network;
the second training unit is used for inputting the second image and the third image into a discriminator, and adjusting model parameters of the generator according to the output result of the discriminator until a preset training condition is met, so as to obtain a generator after training is completed;
And the model application unit is used for reconstructing medical images by using the trained generator.
In a third aspect, an embodiment of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor implementing a method of medical image reconstruction technique as described in the first aspect when the computer program is executed by the processor.
In a fourth aspect, embodiments of the present application provide a computer readable storage medium storing a computer program which, when executed by a processor, implements a method of medical image reconstruction technique as described in the first aspect above.
In a fifth aspect, embodiments of the present application provide a computer program product for, when run on an electronic device, causing the electronic device to perform the method of medical image reconstruction technique as described in the first aspect above.
In the embodiment of the application, a sample training data set is obtained, wherein the sample training data set comprises a first image and a second image, the first image is a low-resolution image, the second image is a high-resolution image corresponding to the first image, a generator is used for carrying out image reconstruction on the first image to obtain a third image, the third image is a reconstructed image corresponding to the first image, and the generator is a pre-trained width learning network; inputting the second image and the third image into a discriminator, and adjusting model parameters of the generator according to the output result of the discriminator until a preset training condition is met, so as to obtain a trained generator. The application optimizes the image reconstruction model based on the generation countermeasure network, pre-trains the generator before training the image reconstruction model, so that the generator has partial fitting capacity, can greatly save the training iteration times, shortens the training time of the model, further adjusts model parameters by combining with the verification of the discriminator, improves the performance of the image reconstruction model, enhances the timeliness of the application of the image reconstruction model, can meet more requirements in practical application, and can improve the efficiency of medical image reconstruction by using the trained generator for medical image reconstruction.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the embodiments or the description of the prior art will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method of implementing a medical image reconstruction technique provided by an embodiment of the present application;
FIG. 2 is a flowchart of a particular implementation of a generator reconstructing an image in a method of medical image reconstruction techniques provided by embodiments of the present application;
FIG. 3 is a flowchart of a specific implementation of a process of training a BLS (wide learning network) in a method of medical image reconstruction technique according to an embodiment of the present application;
FIG. 4 is a flowchart of a specific implementation of model parameter adjustment of a generator in a method of medical image reconstruction techniques provided by an embodiment of the present application;
FIG. 5 is a schematic diagram of a model training scenario of a method of medical image reconstruction techniques provided by an embodiment of the present application;
FIG. 6 is a block diagram of an apparatus for medical image reconstruction techniques provided by an embodiment of the present application;
fig. 7 is a schematic diagram of an electronic device according to an embodiment of the present application.
Detailed Description
In the following description, for purposes of explanation and not limitation, specific details are set forth such as the particular system architecture, techniques, etc., in order to provide a thorough understanding of the embodiments of the present application. It will be apparent, however, to one skilled in the art that the present application may be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present application with unnecessary detail.
As used in the present description and the appended claims, the term "if" may be interpreted as "when..once" or "in response to a determination" or "in response to detection" depending on the context. Similarly, the phrase "if a determination" or "if a [ described condition or event ] is detected" may be interpreted in the context of meaning "upon determination" or "in response to determination" or "upon detection of a [ described condition or event ]" or "in response to detection of a [ described condition or event ]".
Furthermore, the terms "first," "second," "third," and the like in the description of the present specification and in the appended claims, are used for distinguishing between descriptions and not necessarily for indicating or implying a relative importance.
Reference in the specification to "one embodiment" or "some embodiments" or the like means that a particular feature, structure, or characteristic described in connection with the embodiment is included in one or more embodiments of the application. Thus, appearances of the phrases "in one embodiment," "in some embodiments," "in other embodiments," and the like in the specification are not necessarily all referring to the same embodiment, but mean "one or more but not all embodiments" unless expressly specified otherwise. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless expressly specified otherwise.
The method of the medical image reconstruction technology provided by the embodiment of the application is suitable for various types of terminal equipment or servers, and can specifically comprise MRI imaging equipment, mobile phones, tablet computers, wearable equipment, notebook computers, vehicle-mounted equipment, augmented Reality (AR) equipment, virtual Reality (VR) equipment, personal digital assistants (Personal DIGITAL ASSISTANT, PDA), digital televisions and other electronic equipment.
In order to shorten the training time of the model for reconstructing the medical image and improve the model performance and the efficiency of reconstructing the medical image, the embodiment of the application optimizes the image reconstruction model based on the generation countermeasure network, and the generator is formed by a pre-trained width learning network, so that when the image reconstruction model is trained, the generator already has partial fitting capacity, and the accuracy of model parameters of the image reconstruction model is further improved through the verification of the discriminator, thereby shortening the model training time and improving the performance of the image reconstruction model.
The method of medical image reconstruction techniques provided by the present application is exemplified below in connection with specific embodiments.
Fig. 1 shows an implementation flow of a method of medical image reconstruction technology provided by an embodiment of the present application, where an execution end is an electronic device, and the method flow may include the following steps S101 to S104.
S101: a sample training dataset is obtained, the sample training dataset comprises a first image and a second image, wherein the first image is a low-resolution image, and the second image is a high-resolution image corresponding to the first image.
The first image is a low resolution image before image reconstruction, and the second image is a high resolution image corresponding to the first image. In the embodiment of the present application, the first image may be a three-primary-color image (RGB image) or an original image (RAW image) acquired by an image sensor.
In one possible embodiment, the sample training data set is divided into a first sample training data set and a second sample training data set according to a specified proportion, wherein the first sample training data set is used for training the image reconstruction model, and the second sample training data set is used for verifying the trained image reconstruction model.
In one possible implementation manner, the sample training data set is divided into a first sample training data set, a second sample training data set and a third sample training data set according to a specified proportion, wherein the first sample training data set is used for training the image reconstruction model, the second sample training data set is used for verifying the iterative training process, and the third training sample data set is used for verifying the image reconstruction model after the iterative training is finished.
S102: and the generator performs image reconstruction on the first image to obtain a third image, wherein the third image is a reconstructed image corresponding to the first image, and the generator is a pre-trained width learning network.
In an embodiment of the application, the generator in the image reconstruction model is used for reconstructing the image.
In one possible implementation manner, before the first image is input into the generator for image reconstruction, the first image is preprocessed to obtain a preprocessed first image, and the preprocessed first image is input into the generator for image reconstruction.
In some embodiments, the preprocessing includes image normalization processing, and the first image subjected to the image normalization processing is input into a generator for image reconstruction.
In an embodiment of the application, the generator for image reconstruction is a pre-trained breadth learning network (Board LEARNING SYSTEM, BLS). The pre-trained width learning network enables a generator for image reconstruction to have certain fitting capacity before training, and model parameters can be quickly converged.
As a possible embodiment of the present application, as shown in fig. 2, the step of performing image reconstruction on the first image by the generator to obtain a third image includes:
A1: and extracting the linear characteristics of the first image to obtain characteristic nodes.
In the embodiment of the application, the scale of the input is compressed by extracting the features through the feature layer of the generator.
A2: and carrying out nonlinear characteristic enhancement on the characteristic nodes to obtain enhanced nodes.
In the embodiment of the application, the enhancement layer of the generator enhances the characteristic nodes through the nonlinear activation function, and the nonlinear fitting capacity of the model is increased.
A3: and performing image reconstruction according to the feature nodes, the enhancement nodes and the pre-training model parameters of the pre-trained width learning network to generate a third image.
In some possible embodiments, the step A3 specifically includes:
A31: and constructing an input matrix based on the characteristic nodes and the enhancement nodes.
A32: and constructing a pre-training parameter matrix based on the pre-training model parameters.
A33: and determining an output matrix of the pre-trained generator according to the input matrix and the pre-trained parameter matrix.
A34: and reconstructing an image based on the output matrix to obtain a third image.
As a possible implementation manner of the present application, fig. 3 shows a pre-training process of the BLS of the width learning network provided by the embodiment of the present application, and the pre-training of the BLS network, that is, the initialization of the generator. The BLS network includes a feature layer and an enhancement layer, and specifically, the pre-training of the BLS network includes obtaining pre-training model parameters, as detailed below:
b1: and extracting the linear characteristics of the first image to obtain a first image characteristic node.
In the embodiment of the application, the linear characteristics of the first image are extracted through the characteristic layer of the generator, and the first image characteristic node is obtained. The feature layer is composed of a plurality of feature windows, each feature window has a plurality of feature nodes, and each node can be understood as a column vector. In the embodiment of the application, the feature layer is actually an array of column vectors, and the function of the feature layer is to extract the features of the input first image, and is actually to compress the scale of the input image.
B2: and carrying out nonlinear characteristic enhancement on the first image characteristic node to obtain a first image enhancement node.
In the embodiment of the application, the first image enhancement node is obtained by carrying out nonlinear feature enhancement on the first image feature node through the enhancement layer. The enhancement layer enhances the first image feature node through a nonlinear activation function, so that nonlinear fitting capacity of the model is enhanced.
B3: and constructing a model input matrix based on the first image characteristic node and the first image enhancement node.
Specifically, an augmentation matrix is constructed according to the characteristic nodes and the augmentation nodes to form the model input matrix.
In the embodiment of the present application, the image resolution of the first image is n1×n2, and the data size of the model input matrix is n1×n2+n3 rows, each row being n1×n2. Wherein N1 is the number of feature windows, N2 is the number of feature nodes for each feature window, N3 is the number of enhancement nodes, and n1×n2 is the dimension of the feature nodes. The dimensions of each row of the model input matrix are determined from the image resolution of the first image.
In the embodiment of the application, the feature window is actually a preprocessing process for input data, and no matter how much input data exists, the input data can be mapped to N1 feature windows (each feature window has N2 dimensions), so that the input dimensions are reduced, and the complexity of a model is reduced.
B4: and extracting the linear characteristics of the second image to obtain a second image characteristic node.
B5: and constructing a model output matrix according to the second image characteristic nodes.
In the embodiment of the present application, the image resolution of the second image is b1×b2, and the dimension of each row of the model output matrix is b1×b2, that is, the dimension of each row of the model output matrix is determined according to the image resolution of the second image.
B6: and obtaining the pre-training model parameters of the BLS network according to the model input matrix and the model output matrix.
In the embodiment of the application, the pre-training model parameters are obtained according to the regression formula y=wx. Wherein, define X as model input matrix, Y as model output matrix, W as pre-training parameter matrix, namely the matrix of pre-training model parameter construction. Specifically, the pre-training parameter matrix is derived from w=yx +, where X + is the pseudo-inverse found by ridge regression. And obtaining the pre-training model parameters of the BLS network according to the pre-training parameter matrix.
Taking an application scenario as an example, a generator is constructed, an input layer of the generator is composed of a BLS network, and an input layer of the BLS network is composed of a feature layer and an enhancement layer. Specifically, m pairs of image groups, each including m low-resolution images and m high-resolution images, each including 1 low-resolution image and 1 high-resolution image corresponding to the low-resolution image, are acquired from the training sample data set. The images in the m pairs of image groups are normalized to normalize the similarity of the images to between [ -1,1 ]. The resolution of the low-resolution image is N1 multiplied by N2, m low-resolution images are input into a feature layer, a two-dimensional matrix N1 multiplied by N2 of the low-resolution image is rearranged into a one-dimensional vector N1 multiplied by N2 through the feature layer, then N1 feature windows are generated according to preset parameters N1 and N2, each feature window comprises N2 feature nodes, N1 multiplied by N2 feature nodes are obtained, and the dimension of each feature node is N1 multiplied by N2. And carrying out nonlinear characteristic enhancement on the characteristic nodes through the enhancement layer, and generating N3 enhancement nodes according to a preset parameter N3, wherein the dimension of each enhancement node is N1 multiplied by N2. And constructing an augmentation matrix according to the characteristic nodes and the augmentation nodes to obtain a model input matrix X, wherein the data scale of X is N1×N2+N3 rows, and each row is n1×n2. And constructing an output matrix according to m high-resolution images, wherein the resolution of each high-resolution image is b1×b2, and rearranging the two-dimensional matrix b1×b2 of the high-resolution image into a one-dimensional vector b1×b2 through a feature layer to generate a model output matrix. The pre-training parameter matrix w=yx + is calculated according to regression formula y=wx, where X + is the pseudo-inverse found by ridge regression. And obtaining the pre-training parameters based on the pre-training parameter matrix, and completing the pre-training of the generator.
In some embodiments, a fully connected layer is added to the pre-trained BLS as actually needed, the fully connected layer being used to enhance the generation capabilities of the generator.
In the embodiment of the application, the BLS network is utilized to construct the generator, and the BLS network, namely the generator, is pre-trained to obtain the pre-training parameters of the generator, so that the generator has certain fitting capacity before iterative training by utilizing the discriminator, and further the time of iterative training can be shortened.
S103: inputting the second image and the third image into a discriminator, and adjusting model parameters of the generator according to the output result of the discriminator until a preset training condition is met, so as to obtain a trained generator.
In the embodiment of the present application, the preset training condition may be a preset iteration number. Inputting the second image and the third image into a discriminator, adjusting the model parameters of the generator according to the output result of the discriminator, finishing one-time training, repeatedly executing the steps of inputting the first image in the sample training data set into the generator for image reconstruction to obtain the third image, inputting the second image and the third image into the discriminator, and adjusting the model parameters of the generator according to the output result of the discriminator until the preset iteration times, for example, 200 times, finishing training to obtain the trained generator.
In an embodiment of the present application, the arbiter is configured to verify the third image generated by the generator.
In one possible implementation, the arbiter is initialized. Specifically, m low-resolution images are input into a generator which is initialized, a reconstructed image with the size of m is generated, the low-resolution images and the reconstructed image are rearranged into one dimension according to the low-resolution images, and an augmentation matrix is constructed, so that an input matrix is obtained. And splicing all 0 and all 1 vectors with the same scale of m to construct an output matrix, solving a parameter matrix of the discriminator according to a regression formula, and then obtaining parameters of the discriminator.
As a possible implementation manner of the present application, as shown in fig. 4, the step of adjusting the model parameters of the generator according to the output result of the discriminator specifically includes:
C1: and calculating the loss values of the third image and the second image according to a preset loss function.
In the embodiment of the present application, the preset loss function may be a cross entropy function.
C2: and adjusting model parameters of the generator according to the loss value.
In some possible embodiments, the width learning network includes an output layer, and the step C2 specifically includes:
C21: and calculating a gradient vector of the loss value at the output layer according to a back propagation algorithm.
Specifically, a back propagation algorithm is used to calculate the gradient vector of the loss value in the output matrix corresponding to the output layer. The specific process comprises the steps of calculating bias derivatives of each parameter of a hidden layer nearest to an inner output layer according to a loss value, calculating a gradient by using a gradient formula, wherein the gradient formula is as follows (1):
Where i k is the unit vector in the kth direction.
C22: and adjusting model parameters of the generator according to the gradient vector.
Specifically, the gradient descent method is used to calculate updated model parameters and the chain rule is used to similarly update each hidden layer forward.
Taking an application scene as an example, as shown in fig. 5, the generator inputs a low-resolution image, and if the low-resolution image is not trained for the first time, initializing a generator BLS network to obtain a generator with pre-training parameters; if the training is not the first training, obtaining a reconstructed image through a generator, inputting a high-resolution image corresponding to the low-resolution image and the reconstructed image into a discriminator, calculating loss values of the reconstructed image and the high-resolution image according to a preset loss function, updating model parameters of the generator according to a back propagation algorithm, completing one-time training, and completing the training when the preset iteration times are reached.
As a possible embodiment of the application, the parameters of the arbiter are updated. Specifically, the third image and the second image generated by the generator are input to the discriminator, the loss value of the discriminator is calculated according to a preset loss function, the gradient vector of the loss value of the discriminator at the output layer of the discriminator is calculated according to a back propagation algorithm, the model parameters of the discriminator are calculated and updated by using a gradient descent method, and the hidden layers of the discriminator are updated forwards similarly by using a chain rule.
In the embodiment of the application, the parameters of the discriminator are synchronously updated and adjusted based on the reconstructed image generated by the generator during training, so that the discrimination verification capability of the discriminator is further improved, and the optimization effect of the generator during training can be effectively enhanced.
S104: medical image reconstruction is performed using the trained generator.
In the embodiment of the present application, the image reconstruction is performed by the generator obtained by training in the steps S101 to S103, and the method is particularly applicable to medical image reconstruction. In an actual application scene, the trained generator can be used for MRI imaging, is used for observing pathological changes of brain structures such as temporal lobes and hippocampus of Alzheimer's Disease (AD) patients, and can improve the efficiency of MRI reconstruction tasks.
From the above, in the embodiment of the present application, by acquiring a sample training dataset, where the sample training dataset includes a first image and a second image, the first image is a low-resolution image, the second image is a high-resolution image corresponding to the first image, a generator performs image reconstruction on the first image to obtain a third image, the third image is a reconstructed image corresponding to the first image, and the generator is a pre-trained width learning network; inputting the second image and the third image into a discriminator, and adjusting model parameters of the generator according to the output result of the discriminator until a preset training condition is met, so as to obtain a trained generator. The application optimizes the image reconstruction model based on the generation countermeasure network, pre-trains the generator before training the image reconstruction model, so that the generator has partial fitting capacity, can greatly save the training iteration times, shortens the training time of the model, further adjusts model parameters by combining with the verification of the discriminator, improves the performance of the image reconstruction model, enhances the timeliness of the application of the image reconstruction model, can meet more requirements in practical application, and can improve the efficiency of medical image reconstruction by using the trained generator for medical image reconstruction.
It should be understood that the sequence number of each step in the foregoing embodiment does not mean that the execution sequence of each process should be determined by the function and the internal logic, and should not limit the implementation process of the embodiment of the present application.
Corresponding to the method of medical image reconstruction technique described in the above embodiments, fig. 6 shows a block diagram of an apparatus of medical image reconstruction technique provided in an embodiment of the present application, and for convenience of explanation, only a portion related to the embodiment of the present application is shown.
Referring to fig. 6, the apparatus of the medical image reconstruction technique includes: a sample data acquisition unit 61, a first training unit 62, a second training unit 63, wherein:
A sample data obtaining unit 61, configured to obtain a sample training data set, where the sample training data set includes a first image and a second image, and the first image is a low resolution image, and the second image is a high resolution image corresponding to the first image;
A first training unit 62, configured to perform image reconstruction on the first image by using a generator, so as to obtain a third image, where the third image is a reconstructed image corresponding to the first image, and the generator is a pre-trained width learning network;
and the second training unit 63 is configured to input the second image and the third image into a discriminator, and adjust model parameters of the generator according to an output result of the discriminator until a preset training condition is satisfied, thereby obtaining a trained generator.
A model application unit 64 for medical image reconstruction using the trained generator.
In one possible implementation, the first training unit 62 includes:
the characteristic node acquisition module is used for extracting the linear characteristics of the first image to obtain characteristic nodes;
the enhancement node acquisition module is used for carrying out nonlinear feature enhancement on the feature nodes to obtain enhancement nodes;
and the image reconstruction module is used for reconstructing an image according to the feature node, the enhancement node and the pre-training model parameters of the pre-trained width learning network to generate a third image.
In one possible implementation, the image reconstruction module is specifically configured to:
constructing an input matrix based on the feature nodes and the enhancement nodes;
constructing a pre-training parameter matrix based on the pre-training model parameters;
determining an output matrix of the pre-trained generator according to the input matrix and the pre-trained parameter matrix;
and reconstructing an image based on the output matrix to obtain a third image.
In one possible implementation manner, the second training unit 63 specifically includes:
The loss calculation module is used for calculating loss values of the third image and the second image according to a preset loss function;
and the parameter adjustment module is used for adjusting the model parameters of the generator according to the loss value.
In one possible implementation manner, the width learning network includes an output layer, and the parameter adjustment module is specifically configured to:
calculating a gradient vector of the loss value at the output layer according to a back propagation algorithm;
and adjusting model parameters of the generator according to the gradient vector.
In a possible implementation manner, the apparatus of the medical image reconstruction technology further includes a pre-training unit, where the pre-training unit specifically includes:
the first feature extraction module is used for extracting linear features of the first image to obtain first image feature nodes;
The first enhancement module is used for carrying out nonlinear feature enhancement on the first image feature node to obtain a first image enhancement node;
the input construction module is used for constructing a model input matrix based on the first image characteristic node and the first image enhancement node;
the second feature extraction module is used for extracting the linear features of the second image to obtain second image feature nodes;
The output construction module is used for constructing a model output matrix according to the second image characteristic nodes;
And the parameter calculation module is used for obtaining the pre-training model parameters of the width learning network according to the model input matrix and the model output matrix.
From the above, in the embodiment of the present application, by acquiring a sample training dataset, where the sample training dataset includes a first image and a second image, the first image is a low-resolution image, the second image is a high-resolution image corresponding to the first image, a generator performs image reconstruction on the first image to obtain a third image, the third image is a reconstructed image corresponding to the first image, and the generator is a pre-trained width learning network; inputting the second image and the third image into a discriminator, and adjusting model parameters of the generator according to the output result of the discriminator until a preset training condition is met, so as to obtain a trained generator. The application optimizes the image reconstruction model based on the generation countermeasure network, pre-trains the generator before training the image reconstruction model, so that the generator has partial fitting capacity, can greatly save the training iteration times, shortens the training time of the model, further adjusts model parameters by combining with the verification of the discriminator, improves the performance of the image reconstruction model, enhances the timeliness of the application of the image reconstruction model, can meet more requirements in practical application, and can improve the efficiency of medical image reconstruction by using the trained generator for medical image reconstruction.
It should be noted that, because the content of information interaction and execution process between the above devices/units is based on the same concept as the method embodiment of the present application, specific functions and technical effects thereof may be referred to in the method embodiment section, and will not be described herein.
Embodiments of the present application also provide a computer readable storage medium storing computer readable instructions which, when executed by a processor, implement the steps of a method of any of the medical image reconstruction techniques as represented in fig. 1 to 5.
The embodiment of the application also provides an electronic device, which comprises a memory, a processor and computer readable instructions stored in the memory and executable on the processor, wherein the steps of the method of any one of the medical image reconstruction techniques shown in fig. 1 to 5 are realized when the processor executes the computer readable instructions.
Embodiments of the present application also provide a computer program product which, when run on an electronic device, causes the electronic device to perform the steps of a method of implementing any of the medical image reconstruction techniques as represented in fig. 1 to 5.
Fig. 7 is a schematic diagram of an electronic device according to an embodiment of the present application. As shown in fig. 7, the electronic device 7 of this embodiment includes: a processor 70, a memory 71, and computer readable instructions 72 stored in the memory 71 and executable on the processor 70. The processor 70, when executing the computer readable instructions 72, implements the steps of the method embodiments of the respective medical image reconstruction techniques described above, such as steps S101 to S104 shown in fig. 1. Or the processor 70, when executing the computer readable instructions 72, performs the functions of the modules/units of the apparatus embodiments described above, such as the units 61 through 64 of fig. 6.
Illustratively, the computer readable instructions 72 may be partitioned into one or more modules/units that are stored in the memory 71 and executed by the processor 70 to complete the present application. The one or more modules/units may be a series of computer readable instruction segments capable of performing particular functions describing the execution of the computer readable instructions 72 in the electronic device 7.
The electronic device 7 may be an imaging device or a server. The electronic device 7 may include, but is not limited to, a processor 70, a memory 71. It will be appreciated by those skilled in the art that fig. 7 is merely an example of the electronic device 7 and is not meant to be limiting of the electronic device 7, and may include more or fewer components than shown, or may combine certain components, or different components, e.g., the electronic device 7 may further include input-output devices, network access devices, buses, etc.
The Processor 70 may be a central processing unit (CentraL Processing Unit, CPU), or may be another general purpose Processor, a digital signal Processor (DIGITAL SIGNAL Processor, DSP), an application specific integrated Circuit (AppLication SPECIFIC INTEGRATED Circuit, ASIC), an off-the-shelf programmable gate array (FieLd-ProgrammabLe GATE ARRAY, FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, or the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The memory 71 may be an internal storage unit of the electronic device 7, such as a hard disk or a memory of the electronic device 7. The memory 71 may also be an external storage device of the electronic device 7, such as a plug-in hard disk, a smart memory card (SMART MEDIA CARD, SMC), a Secure DigitaL (SD) card, a flash memory card (FLASH CARD) or the like, which are provided on the electronic device 7. Further, the memory 71 may also include both an internal storage unit and an external storage device of the electronic device 7. The memory 71 is used to store the computer readable instructions and other programs and data required by the electronic device. The memory 71 may also be used for temporarily storing data that has been output or is to be output.
It should be noted that, because the content of information interaction and execution process between the above devices/units is based on the same concept as the method embodiment of the present application, specific functions and technical effects thereof may be referred to in the method embodiment section, and will not be described herein.
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-described division of the functional units and modules is illustrated, and in practical application, the above-described functional distribution may be performed by different functional units and modules according to needs, i.e. the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-described functions. The functional units and modules in the embodiment may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit, where the integrated units may be implemented in a form of hardware or a form of a software functional unit. In addition, the specific names of the functional units and modules are only for distinguishing from each other, and are not used for limiting the protection scope of the present application. The specific working process of the units and modules in the above system may refer to the corresponding process in the foregoing method embodiment, which is not described herein again.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the present application may implement all or part of the flow of the method of the above embodiments, and may be implemented by a computer program to instruct related hardware, where the computer program may be stored in a computer readable storage medium, and when the computer program is executed by a processor, the computer program may implement the steps of each of the method embodiments described above. Wherein the computer program comprises computer program code which may be in source code form, object code form, executable file or some intermediate form etc. The computer readable medium may include at least: any entity or device capable of carrying computer program code to a photographing device/terminal apparatus, recording medium, computer Memory, read-Only Memory (ROM), random access Memory (RAM, random Access Memory), electrical carrier signals, telecommunications signals, and software distribution media. Such as a U-disk, removable hard disk, magnetic or optical disk, etc. In some jurisdictions, computer readable media may not be electrical carrier signals and telecommunications signals in accordance with legislation and patent practice.
In the foregoing embodiments, the descriptions of the embodiments are emphasized, and in part, not described or illustrated in any particular embodiment, reference is made to the related descriptions of other embodiments.
The above embodiments are only for illustrating the technical solution of the present application, and not for limiting the same; although the application has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present application, and are intended to be included in the scope of the present application.
Claims (10)
1. A method of medical image reconstruction techniques, comprising:
Acquiring a sample training data set, wherein the sample training data set comprises a first image and a second image, the first image is a low-resolution image, and the second image is a high-resolution image corresponding to the first image;
The generator performs image reconstruction on the first image to obtain a third image, wherein the third image is a reconstructed image corresponding to the first image, and the generator is a pre-trained width learning network;
Initializing a discriminator, wherein the discriminator is used for verifying a third image generated by the generator, inputting m low-resolution images into the initialized generator, generating a reconstructed image with the scale of m, rearranging the low-resolution images and the reconstructed image into a one-dimensional vector, and constructing an augmentation matrix to obtain an input matrix; the all 0 and all 1 vectors with the same scale of m are spliced to construct an output matrix, and the parameter matrix of the discriminator is obtained according to a regression formula to obtain the parameters of the discriminator;
Inputting the second image and the third image into a discriminator, and adjusting model parameters of the generator according to the output result of the discriminator until a preset training condition is met, so as to obtain a trained generator;
medical image reconstruction is performed using the trained generator.
2. The method of claim 1, wherein the step of the generator performing image reconstruction on the first image to obtain a third image comprises:
Extracting linear characteristics of the first image to obtain characteristic nodes;
Carrying out nonlinear characteristic enhancement on the characteristic nodes to obtain enhancement nodes;
and performing image reconstruction according to the feature nodes, the enhancement nodes and the pre-training model parameters of the pre-trained width learning network to generate a third image.
3. The method of claim 2, wherein the step of generating a third image from image reconstruction from pre-trained model parameters of the feature node, the enhancement node, and the pre-trained breadth-learning network comprises:
constructing an input matrix based on the feature nodes and the enhancement nodes;
constructing a pre-training parameter matrix based on the pre-training model parameters;
determining an output matrix of the pre-trained generator according to the input matrix and the pre-trained parameter matrix;
and reconstructing an image based on the output matrix to obtain a third image.
4. The method of claim 1, wherein the step of adjusting model parameters of the generator based on the output of the arbiter comprises:
calculating the loss values of the third image and the second image according to a preset loss function;
and adjusting model parameters of the generator according to the loss value.
5. The method of claim 4, wherein the width learning network comprises an output layer, and wherein the step of adjusting model parameters of the generator according to the loss value comprises:
calculating a gradient vector of the loss value at the output layer according to a back propagation algorithm;
and adjusting model parameters of the generator according to the gradient vector.
6. The method according to any one of claims 1 to 5, wherein the pre-training of the breadth-learning network comprises obtaining pre-training model parameters, in particular comprising:
extracting linear characteristics of the first image to obtain a first image characteristic node;
performing nonlinear characteristic enhancement on the first image characteristic node to obtain a first image enhancement node;
Constructing a model input matrix based on the first image feature node and the first image enhancement node;
Extracting linear characteristics of the second image to obtain second image characteristic nodes;
Constructing a model output matrix according to the second image characteristic nodes;
And obtaining the pre-training model parameters of the width learning network according to the model input matrix and the model output matrix.
7. An apparatus for medical image reconstruction techniques, comprising:
The system comprises a sample data acquisition unit, a first image acquisition unit and a second image acquisition unit, wherein the sample data acquisition unit is used for acquiring a sample training data set, the sample training data set comprises a first image and a second image, the first image is a low-resolution image, and the second image is a high-resolution image corresponding to the first image;
The first training unit is used for carrying out image reconstruction on the first image by the generator to obtain a third image, wherein the third image is a reconstructed image corresponding to the first image, and the generator is a pre-trained width learning network; initializing a discriminator, wherein the discriminator is used for verifying a third image generated by the generator, inputting m low-resolution images into the initialized generator, generating a reconstructed image with the scale of m, rearranging the low-resolution images and the reconstructed image into a one-dimensional vector, and constructing an augmentation matrix to obtain an input matrix; the all 0 and all 1 vectors with the same scale of m are spliced to construct an output matrix, and the parameter matrix of the discriminator is obtained according to a regression formula to obtain the parameters of the discriminator;
the second training unit is used for inputting the second image and the third image into a discriminator, and adjusting model parameters of the generator according to the output result of the discriminator until a preset training condition is met, so as to obtain a generator after training is completed;
And the model application unit is used for reconstructing medical images by using the trained generator.
8. The apparatus of claim 7, wherein the first training unit comprises:
the characteristic node acquisition module is used for extracting the linear characteristics of the first image to obtain characteristic nodes;
the enhancement node acquisition module is used for carrying out nonlinear feature enhancement on the feature nodes to obtain enhancement nodes;
and the image reconstruction module is used for reconstructing an image according to the feature node, the enhancement node and the pre-training model parameters of the pre-trained width learning network to generate a third image.
9. An electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements a method of medical image reconstruction technique according to any one of claims 1 to 6 when the computer program is executed.
10. A computer readable storage medium storing a computer program, characterized in that the computer program, when executed by a processor, implements a method of medical image reconstruction technique according to any one of claims 1 to 6.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010565805.6A CN111862251B (en) | 2020-06-19 | 2020-06-19 | Method, device, storage medium and electronic equipment for medical image reconstruction technology |
| PCT/CN2020/129565 WO2021253722A1 (en) | 2020-06-19 | 2020-11-17 | Medical image reconstruction technology method and apparatus, storage medium and electronic device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010565805.6A CN111862251B (en) | 2020-06-19 | 2020-06-19 | Method, device, storage medium and electronic equipment for medical image reconstruction technology |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111862251A CN111862251A (en) | 2020-10-30 |
| CN111862251B true CN111862251B (en) | 2024-05-03 |
Family
ID=72986947
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010565805.6A Active CN111862251B (en) | 2020-06-19 | 2020-06-19 | Method, device, storage medium and electronic equipment for medical image reconstruction technology |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN111862251B (en) |
| WO (1) | WO2021253722A1 (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111862251B (en) * | 2020-06-19 | 2024-05-03 | 中国科学院深圳先进技术研究院 | Method, device, storage medium and electronic equipment for medical image reconstruction technology |
| CN112509091B (en) * | 2020-12-10 | 2023-11-14 | 上海联影医疗科技股份有限公司 | Medical image reconstruction method, device, equipment and medium |
| CN114742807A (en) * | 2022-04-24 | 2022-07-12 | 北京医准智能科技有限公司 | Chest radiography identification method and device based on X-ray image, electronic equipment and medium |
| CN115544126A (en) * | 2022-12-05 | 2022-12-30 | 南方电网数字电网研究院有限公司 | Frequency-up reconstruction method and device for photovoltaic data, computer equipment and storage medium |
| CN116863016A (en) * | 2023-05-31 | 2023-10-10 | 北京长木谷医疗科技股份有限公司 | Medical image reconstruction method and device for generating countermeasure network based on deep learning |
| CN116402825B (en) * | 2023-06-09 | 2023-08-11 | 华东交通大学 | Bearing fault infrared diagnosis method, system, electronic equipment and storage medium |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109146784A (en) * | 2018-07-27 | 2019-01-04 | 徐州工程学院 | A kind of image super-resolution rebuilding method based on multiple dimensioned generation confrontation network |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10540798B1 (en) * | 2019-01-10 | 2020-01-21 | Capital One Services, Llc | Methods and arrangements to create images |
| CN110070174B (en) * | 2019-04-10 | 2020-11-27 | 厦门美图之家科技有限公司 | Stable training method for generating confrontation network |
| CN111160198B (en) * | 2019-12-23 | 2023-06-27 | 北方工业大学 | Object identification method and system based on width learning |
| CN111862251B (en) * | 2020-06-19 | 2024-05-03 | 中国科学院深圳先进技术研究院 | Method, device, storage medium and electronic equipment for medical image reconstruction technology |
-
2020
- 2020-06-19 CN CN202010565805.6A patent/CN111862251B/en active Active
- 2020-11-17 WO PCT/CN2020/129565 patent/WO2021253722A1/en active Application Filing
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109146784A (en) * | 2018-07-27 | 2019-01-04 | 徐州工程学院 | A kind of image super-resolution rebuilding method based on multiple dimensioned generation confrontation network |
Non-Patent Citations (4)
| Title |
|---|
| Broad Learning System: An Effective and Efficient Incremental Learning System Without the Need for Deep Architecture;C. L. Philip Chen等;IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS;第29卷(第1期);正文第1-15页 * |
| 基于宽度学习的集成超分辨率重建方法;曾宪华等;计算机工程与设计;第37卷(第9期);第2527-2530页 * |
| 基于秩极小化理论的单幅图像超分辨率复原;车云;;哈尔滨商业大学学报(自然科学版);20161015;第32卷(第05期);第594-597页 * |
| 天绘一号卫星图像处理方法及应用研究;黄伟佳等;测绘与空间地理信息;第36卷(第7期);第115-116页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111862251A (en) | 2020-10-30 |
| WO2021253722A1 (en) | 2021-12-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111862251B (en) | Method, device, storage medium and electronic equipment for medical image reconstruction technology | |
| Lan et al. | MADNet: a fast and lightweight network for single-image super resolution | |
| WO2020168844A1 (en) | Image processing method, apparatus, equipment, and storage medium | |
| CN111369440B (en) | Model training and image super-resolution processing method, device, terminal and storage medium | |
| EP3963516B1 (en) | Teaching gan (generative adversarial networks) to generate per-pixel annotation | |
| Kim et al. | Deeply aggregated alternating minimization for image restoration | |
| CN112488923A (en) | Image super-resolution reconstruction method and device, storage medium and electronic equipment | |
| CN114863225B (en) | Image processing model training method, image processing model generation device, image processing model equipment and image processing model medium | |
| CN112150470B (en) | Image segmentation method, device, medium and electronic equipment | |
| CN110874855B (en) | Collaborative imaging method and device, storage medium and collaborative imaging equipment | |
| CN110246200B (en) | Magnetic resonance cardiac cine imaging method and device and magnetic resonance scanner | |
| Zhao et al. | A deep variational Bayesian framework for blind image deblurring | |
| CN109961435B (en) | Brain image acquisition method, device, equipment and storage medium | |
| CN110135428B (en) | Image segmentation processing method and device | |
| CN115115724A (en) | Image processing method, image processing device, computer equipment and storage medium | |
| Hua et al. | Dynamic scene deblurring with continuous cross-layer attention transmission | |
| CN116503239A (en) | Vehicle-mounted video data abnormality processing and simulation method | |
| CN110677671A (en) | Image compression method and device and terminal equipment | |
| Wu et al. | Non‐uniform image blind deblurring by two‐stage fully convolution network | |
| Tan et al. | Denoising real bursts with squeeze‐and‐excitation residual network | |
| CN109949383B (en) | High dynamic optical projection tomography method and device | |
| CN114494484A (en) | Training method of data recognition model, data recognition method, device and equipment | |
| Jia et al. | Learning Rich Information for Quad Bayer Remosaicing and Denoising | |
| CN110689513B (en) | Color image fusion method and device and terminal equipment | |
| CN112465007B (en) | Training method of target recognition model, target recognition method and terminal equipment |
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 |