CN105574820A - Deep learning-based adaptive ultrasound image enhancement method - Google Patents
Deep learning-based adaptive ultrasound image enhancement method Download PDFInfo
- Publication number
- CN105574820A CN105574820A CN201510885535.6A CN201510885535A CN105574820A CN 105574820 A CN105574820 A CN 105574820A CN 201510885535 A CN201510885535 A CN 201510885535A CN 105574820 A CN105574820 A CN 105574820A
- Authority
- CN
- China
- Prior art keywords
- neural network
- data
- image
- deep neural
- ultrasonoscopy
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000002604 ultrasonography Methods 0.000 title claims abstract description 10
- 230000003044 adaptive effect Effects 0.000 title abstract description 4
- 238000013135 deep learning Methods 0.000 title abstract 2
- 238000013528 artificial neural network Methods 0.000 claims abstract description 23
- 238000012549 training Methods 0.000 claims abstract description 16
- 230000015572 biosynthetic process Effects 0.000 claims abstract description 15
- 238000003062 neural network model Methods 0.000 claims abstract description 12
- 238000012545 processing Methods 0.000 claims abstract description 9
- 239000000284 extract Substances 0.000 claims abstract description 5
- 230000004927 fusion Effects 0.000 claims abstract description 4
- 230000008569 process Effects 0.000 claims description 8
- 230000006870 function Effects 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 5
- 238000001914 filtration Methods 0.000 claims description 5
- 238000005070 sampling Methods 0.000 claims description 4
- 238000005728 strengthening Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 2
- 238000003672 processing method Methods 0.000 claims description 2
- 238000012546 transfer Methods 0.000 claims description 2
- 238000003384 imaging method Methods 0.000 abstract description 2
- 230000000694 effects Effects 0.000 abstract 1
- 230000001629 suppression Effects 0.000 abstract 1
- 239000000523 sample Substances 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000002059 diagnostic imaging Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 210000000056 organ Anatomy 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
-
- 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
- G06N3/088—Non-supervised learning, e.g. competitive learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/19—Recognition using electronic means
- G06V30/192—Recognition using electronic means using simultaneous comparisons or correlations of the image signals with a plurality of references
- G06V30/194—References adjustable by an adaptive method, e.g. learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10132—Ultrasound image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
Abstract
The invention provides a deep learning-based adaptive ultrasound image enhancement method. The method is carried out according to the following steps: training a deep neural network; reading ultrasound image data; dividing the image data into blocks and inputting the blocks into the trained deep neural network, and classifying the deep neural network into a uniform formation region and a structure region; carrying out corresponding data processing on ultrasound block images respectively; and carrying out fusion and whole outputting on the corresponding parts. The step of training the deep neural network also comprises the following steps: collecting picture data, and carrying out grouping pretreatment on the picture data; building a RBM model, and determining the number of layers and the training mode of the model; extracting features, and feeding a group of pictures in the grouped pictures into the RBM to calculate and extract image features; building the neural network, building a deep neural network model by the image features extracted in the previous step; and adopting specific image enhancement and speckle suppression according to classification and identification results, so that the imaging effect of the ultrasound images is effectively improved.
Description
Technical field
The present invention relates to image enchancing method, be specifically the self-adaptation ultrasonoscopy Enhancement Method of a kind of degree of depth study, belong to ultrasonoscopy processing technology field.
Background technology
Ultrasound medical imaging because of its have intuitively, the advantage such as convenient, safe, quick is widely used in clinical, but due to the physical characteristics of ultrasonic imaging and the association attributes of ultrasonic probe, ultrasonoscopy is while reflection human organ profile, also usually occur irregular blotches and pseudomorphism, this have impact on the quality of image and the identification of focus greatly.For this situation, ultrasonoscopy main at present strengthens algorithm anisotropy parameter, wavelet transformation, medium filtering etc., but these methods act on entire image, fuzzy border while there will be filtering noise unavoidably, or also enhance noise while strengthening border.So good disposal route first analyzes image, processes respectively after edge and tissue detect again.In prior art to the analysis of image mainly through the method such as compute gradient, structure tensor, but this method needs setting hard-threshold usually, lacks good adaptive ability.
Summary of the invention
To the present invention is directed in ultrasonoscopy processing procedure the inaccurate problem of graphical analysis, disclose a kind of can the self-adaptation ultrasonoscopy Enhancement Method of the degree of depth study of identifying processing ultrasonoscopy voluntarily.
For realizing above technical purpose, the present invention will take following technical scheme:
Based on a self-adaptation ultrasonoscopy Enhancement Method for degree of depth study, carry out in accordance with the following steps: step (1) training deep neural network; Step (2) reads ultrasound image data; View data is divided into fritter by step (3), inputs the deep neural network trained, is categorized as uniform formation region and structural region; Step (4) carries out corresponding data processing respectively to this ultrasonic small images; Corresponding partial fusion view picture exports by step (5); Wherein, step (1) comprises the steps: that steps A gathers image data, carries out grouping pre-service to image data; Step B builds RBM model, the number of plies of Confirming model, training patterns; Step C feature extraction, sends in RBM to calculate by the picture group sheet in steps A and extracts characteristics of image; Step D neural network, the characteristics of image utilizing step C to extract sets up deep neural network model.
The technical scheme that the present invention limits further is:
Further, steps A is specially: system random acquisition M from ultrasonic device opens ultrasonoscopy, wherein M >=10; Each pictures in image set is divided into the fritter of n*n pixel, the principle choosing fritter is the feature that fritter comprises classification; The fritter picture split marks according to uniform formation region and structural region by artificial mode, and the fritter picture marked is unit classified and stored according to initial whole pictures.
Further, in step B, the described RBM model number of plies is set to three layers, and adopt gibbs ALTERNATE SAMPLING method successively to train, iterations is 50 times.Step C feature extraction, sends in RBM to calculate by the picture group sheet in steps A and extracts characteristics of image.The foundation of step D neural network model, sends into restriction Boltzmann machine network and carries out calculating the characteristic extracting training image, set up deep neural network model by these initial parameters in units of initial whole pictures by the fritter picture split.
Further, step (3) comprising:
From ultrasonic device, obtain ultrasonoscopy be divided into fritter, input the deep neural network trained and identify, be categorized as uniform formation region and structural region.Because deep neural network has been trained by above-mentioned steps, each parameter is preserved, then can ensure the process that the classification of image can be real-time.
Further, step (4) comprising: according to the result identified, take different image processing methods.Be identified as the small images in uniform formation region, adopt gaussian filtering to carry out denoising.Be identified as structural region, carry out image enhaucament, formula is:
，
Wherein,
represent the gray-scale value after strengthening,
represent fritter gray average,
represent gray scale regulation coefficient.
Further, deep neural network model carries out pre-training by degree of depth belief network DBN, comprising:
Adopt unsupervised learning mode from bottom to top to train, namely use is without the sample data of label from bottom, and past top layer is in layer trained.
RBM is a kind of energy transfer model, and the energy state defining every one deck is:
，
Wherein,
be the relevant parameter of w, a, b, k is kth layer RBM,
be
layer i unit and
link weight parameter between layer j unit, and a and b is respectively
layer and
the bigoted parameter of layer.And just know accordingly
probability be:
，
Wherein,
it is normalization coefficient.According to probability distribution above, be easy to obtain following conditional probability:
，
，
Wherein,
, by the probability logarithm of hidden layer to the differentiate of W, can obtain:
，
Wherein,
represent the expectation of Data distribution8,
represent the Data distribution8 after S Gibbs model.Can obtain further:
，
Wherein,
momentum,
it is learning rate.The unbiased sample of realistic model is often difficult to obtain, and general employing contrast three sampling of method to reconstruct data is all similar to and upgrades network weight.The input of lower one deck comes from the output of last layer, transmit, and the input of the bottom generally comes from observational variable with this, and the namely initial characteristic data of object, as the pixel value of image.The pixel value being input as n*n small images of the bottom of the present invention.
Further, the tuning of described degree of depth belief network (DBN), comprising:
Adopt top-down supervised learning mode to carry out tuning, namely use the sample data of label to train,
Error is top-down to be transmitted, and carries out tuning, comprising network:
Export the error function of data and reconstruct data according to target, utilize backpropagation (BP) algorithm to readjust the parameter of network, finally make network reach the process of global optimum.The error function that target exports data and reconstruct data is
，
Wherein, y is that target exports,
reconstruct data,
represent 2 normal form of reconstructed error, error function is asked to the local derviation of weights, the updated value of weight can be obtained.
The present invention is owing to taking above technical scheme, tool has the following advantages: the present invention utilizes degree of depth learning art to analyze ultrasonoscopy, the eigenwert meeting image is found by the means of no manual intervention, carry out again identifying, classifying, afterwards for different classification, take different Processing Algorithm, this method has better robustness and adaptive ability.
Accompanying drawing explanation
Fig. 1 is whole implementation protocol procedures figure of the present invention;
Fig. 2 is training pattern schematic diagram of the present invention;
Fig. 3 is the particular flow sheet of present invention process embodiment;
Fig. 4 is a B ultrasonic figure before treatment;
Fig. 5 is the B ultrasonic figure after a process.
Embodiment
Accompanying drawing discloses the schematic flow sheet of preferred embodiment involved in the present invention without limitation; Technical scheme of the present invention is explained below with reference to accompanying drawing.
Based on a ultrasonoscopy Enhancement Method for degree of depth study, its basic step is as follows:
Training deep neural network.
Read ultrasound image data.
By the deep neural network that view data input trains, be categorized as uniform formation region and structural region.
Respectively corresponding data processing is carried out to this ultrasonic small images.
Corresponding partial fusion view picture is exported.
Wherein train the step of deep neural network as follows:
Gathering m from ultrasonic device and open different ultrasonoscopys, can be the image obtained under the different conditions such as diverse location, not consubstantiality mould.Current image is divided into two parts, and a part is as training plan image set, and a part is as test pattern image set.
These images are divided into the fritter of n*n pixel, and carry out label.Such as label is designated as uniform formation region and structural region.
The image of training sample is sent into restriction Boltzmann machine network and carry out pre-service, say the weight initialization BP neural network parameter obtained, train deep neural network model.
Be below elaborate a kind of ultrasonoscopy Enhancement Method based on degree of depth study of the present invention, process flow diagram is as Fig. 1, Fig. 3.
From ultrasonic device, obtain ultrasonoscopy, be divided into fractionlet, input the network trained and identify, the result according to identifying takes different image processing algorithms.
The method that acquisition m opens different ultrasonoscopys (20 ultrasonoscopys are chosen in this experiment) is under different condition such as scanning different positions or different positions etc.
All images are divided into the fritter to n*n pixel, and the principle choosing fritter is the feature that fritter comprises classification, but can not affect the accuracy rate identified too greatly.
Carry out mark to all fritters, it is labeled as uniform formation region and structural region.
As Fig. 1, the image of training sample is sent into restriction Boltzmann machine network and carry out pre-service, by the weight initialization BP neural network parameter obtained, train deep neural network model, and with the deep neural network model trained, the step that the result of feature extraction carries out Images Classification identification comprised:
The picture element matrix X={X0 of n*n fritter ultrasonoscopy, X1...Xn} are sent into restriction Boltzmann machine network and carry out data prediction, the restriction Boltzmann machine network of three layers of 8*8-50-30-10 structure is selected in experiment, and the maximum iteration time of every layer network is set to 50;
Pretreated data are added reverse BP algorithm of neural network to learn, experiment iterations is set to 100, and selecting the desired output Y={Y0 of network, Y1}, represent uniform formation region and structural region respectively, such as, is uniform formation region then Y={1,0};
Set up deep neural network system model by study, and utilize described deep neural network system to carry out the analysis identification of ultrasonoscopy.
As accompanying drawing 3, from ultrasonic device, obtain ultrasound image data, be divided into the fritter of n*n, by the deep neural network that fritter input trains, carry out the analysis identification of image, process respectively according to recognition result, in this experiment, test pattern is divided into the fritter of 8*8, is identified as the small images in uniform formation region, carry out gaussian filtering, for structural region, carry out algorithm for image enhancement, finally merge view picture and export.
As accompanying drawing 4 and Fig. 5 contrast are easy to find out, picture organization edge is after treatment clear, Be very effective.
Claims (7)
1., based on a self-adaptation ultrasonoscopy Enhancement Method for degree of depth study, carry out in accordance with the following steps:
Step (1) training deep neural network;
Step (2) reads ultrasound image data;
View data is divided into fritter by step (3), inputs the deep neural network trained, is categorized as uniform formation region and structural region;
Step (4) carries out corresponding data processing respectively to this ultrasonic small images;
Corresponding partial fusion view picture exports by step (5);
It is characterized in that, step (1) comprises the steps:
Steps A gathers image data, carries out grouping pre-service to image data; Step B builds RBM model, the number of plies of Confirming model, training patterns; Step C feature extraction, sends in RBM to calculate by the picture group sheet in steps A and extracts characteristics of image;
Step D neural network, the characteristics of image utilizing step C to extract sets up deep neural network model.
2. the self-adaptation ultrasonoscopy Enhancement Method based on degree of depth study according to claim 1, it is characterized in that, steps A is specially: system random acquisition M from ultrasonic device opens ultrasonoscopy, wherein M >=10; Each pictures in image set is divided into the fritter of n*n pixel, the principle choosing fritter is the feature that fritter comprises classification; The fritter picture split marks according to uniform formation region and structural region by artificial mode, and the fritter picture marked is unit classified and stored according to initial whole pictures.
3. the self-adaptation ultrasonoscopy Enhancement Method based on degree of depth study according to claim 1, it is characterized in that, in step B, the described RBM model number of plies is set to three layers, and adopt gibbs ALTERNATE SAMPLING method successively to train, iterations is 50 times; Step C feature extraction, sends in RBM to calculate by the picture group sheet in steps A and extracts characteristics of image; The foundation of step D neural network model, sends into restriction Boltzmann machine network and carries out calculating the characteristic extracting training image, set up deep neural network model by these initial parameters in units of initial whole pictures by the fritter picture split.
4. the self-adaptation ultrasonoscopy Enhancement Method based on degree of depth study according to claim 1, it is characterized in that, step (3) comprising: from ultrasonic device, obtain ultrasonoscopy be divided into fritter, input the deep neural network trained to identify, be categorized as uniform formation region and structural region.
5. the self-adaptation ultrasonoscopy Enhancement Method based on degree of depth study according to claim 1, it is characterized in that, step (4) comprising: according to the result identified, take different image processing methods, be identified as the small images in uniform formation region, adopt gaussian filtering to carry out denoising; Be identified as structural region, carry out image enhaucament, formula is:
，
Wherein,
represent the gray-scale value after strengthening,
represent fritter gray average,
represent gray scale regulation coefficient.
6. the self-adaptation ultrasonoscopy Enhancement Method based on degree of depth study according to claim 1 or 3, it is characterized in that, deep neural network model carries out pre-training by degree of depth belief network DBN, comprising:
Adopt unsupervised learning mode from bottom to top to train, namely use is without the sample data of label from bottom, and past top layer is in layer trained;
RBM is a kind of energy transfer model, and the energy state defining every one deck is:
，
Wherein,
be the relevant parameter of w, a, b, k is kth layer RBM,
be
layer i unit and
link weight parameter between layer j unit, and a and b is respectively
layer and
the bigoted parameter of layer; And just know accordingly
probability be:
，
Wherein,
it is normalization coefficient; According to probability distribution above, be easy to obtain following conditional probability:
，
，
Wherein,
, by the probability logarithm of hidden layer to the differentiate of W, can obtain:
，
Wherein,
represent the expectation of Data distribution8,
represent the Data distribution8 after S Gibbs model; Can obtain further:
，
Wherein,
momentum,
it is learning rate; The unbiased sample of realistic model is often difficult to obtain, and general employing contrast three sampling of method to reconstruct data is all similar to and upgrades network weight; The input of lower one deck comes from the output of last layer, transmit, and the input of the bottom generally comes from observational variable with this, and the namely initial characteristic data of object, as the pixel value of image; The pixel value being input as n*n small images of the bottom of the present invention.
7. the self-adaptation ultrasonoscopy Enhancement Method based on degree of depth study according to claim 6, it is characterized in that, the tuning of described degree of depth belief network (DBN), comprising:
Adopt top-down supervised learning mode to carry out tuning, namely use the sample data of label to train,
Error is top-down to be transmitted, and carries out tuning, comprising network: the error function exporting data and reconstruct data according to target, utilizes backpropagation (BP) algorithm to readjust the parameter of network, finally make network reach the process of global optimum; The error function that target exports data and reconstruct data is:
，
Wherein, y is that target exports,
reconstruct data,
represent 2 normal form of reconstructed error, error function is asked to the local derviation of weights, the updated value of weight can be obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510885535.6A CN105574820A (en) | 2015-12-04 | 2015-12-04 | Deep learning-based adaptive ultrasound image enhancement method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510885535.6A CN105574820A (en) | 2015-12-04 | 2015-12-04 | Deep learning-based adaptive ultrasound image enhancement method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105574820A true CN105574820A (en) | 2016-05-11 |
Family
ID=55884915
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510885535.6A Pending CN105574820A (en) | 2015-12-04 | 2015-12-04 | Deep learning-based adaptive ultrasound image enhancement method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105574820A (en) |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106203488A (en) * | 2016-07-01 | 2016-12-07 | 福州大学 | A kind of galactophore image Feature fusion based on limited Boltzmann machine |
| CN106331433A (en) * | 2016-08-25 | 2017-01-11 | 上海交通大学 | Video denoising method based on deep recursive neural network |
| CN106503801A (en) * | 2016-10-18 | 2017-03-15 | 天津工业大学 | Electrical impedance tomography method based on deep learning |
| CN106548169A (en) * | 2016-11-02 | 2017-03-29 | 重庆中科云丛科技有限公司 | Fuzzy literal Enhancement Method and device based on deep neural network |
| CN108154169A (en) * | 2017-12-11 | 2018-06-12 | 北京小米移动软件有限公司 | Image processing method and device |
| CN108229347A (en) * | 2016-12-22 | 2018-06-29 | Tcl集团股份有限公司 | For the method and apparatus of the deep layer displacement of the plan gibbs structure sampling of people's identification |
| WO2018157868A1 (en) * | 2017-03-02 | 2018-09-07 | 江清泉 | Ultrasonic imaging system and imaging method thereof |
| CN108985366A (en) * | 2018-07-06 | 2018-12-11 | 武汉兰丁医学高科技有限公司 | B ultrasound image recognition algorithm and system based on convolution depth network |
| CN109102515A (en) * | 2018-07-31 | 2018-12-28 | 浙江杭钢健康产业投资管理有限公司 | A kind of method for cell count based on multiple row depth convolutional neural networks |
| CN109658354A (en) * | 2018-12-20 | 2019-04-19 | 上海联影医疗科技有限公司 | A kind of image enchancing method and system |
| CN109833061A (en) * | 2017-11-24 | 2019-06-04 | 无锡祥生医疗科技股份有限公司 | The method of optimization ultrasonic image-forming system parameter based on deep learning |
| CN109965905A (en) * | 2019-04-11 | 2019-07-05 | 复旦大学 | A kind of radiography region detection imaging method based on deep learning |
| CN110033417A (en) * | 2019-04-12 | 2019-07-19 | 江西财经大学 | A kind of image enchancing method based on deep learning |
| CN110197714A (en) * | 2018-02-27 | 2019-09-03 | 希森美康株式会社 | The generation method of deep learning algorithm after method for analyzing image, device, study |
| WO2019168699A1 (en) * | 2018-03-01 | 2019-09-06 | Fujifilm Sonosite, Inc. | Method and apparatus for annotating ultrasound examinations |
| CN110555827A (en) * | 2019-08-06 | 2019-12-10 | 上海工程技术大学 | Ultrasonic imaging information computer processing system based on deep learning drive |
| CN110694149A (en) * | 2019-10-16 | 2020-01-17 | 山东大学齐鲁医院 | Ultrasonic-assisted muscle identification method and system and auxiliary injection device |
| WO2020016836A1 (en) * | 2018-07-18 | 2020-01-23 | Advenio Tecnosys Pvt. Ltd. | System and method for managing the quality of an image |
| CN112508833A (en) * | 2020-12-09 | 2021-03-16 | 东南大学 | Medical image enhancement method and device based on deep learning |
| CN113570499A (en) * | 2021-07-21 | 2021-10-29 | 此刻启动(北京)智能科技有限公司 | Self-adaptive image toning method, system, storage medium and electronic equipment |
| CN114782283A (en) * | 2022-06-16 | 2022-07-22 | 深圳华声医疗技术股份有限公司 | Ultrasonic image enhancement method and device, ultrasonic equipment and storage medium |
| US11475569B2 (en) | 2018-05-30 | 2022-10-18 | Shanghai United Imaging Healthcare Co., Ltd. | Systems and methods for image processing |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101630405A (en) * | 2009-08-14 | 2010-01-20 | 重庆市勘测院 | Multi-focusing image fusion method utilizing core Fisher classification and redundant wavelet transformation |
| CN102243711A (en) * | 2011-06-24 | 2011-11-16 | 南京航空航天大学 | Neighbor embedding-based image super-resolution reconstruction method |
| CN103632351A (en) * | 2013-12-16 | 2014-03-12 | 武汉大学 | All-weather traffic image enhancement method based on brightness datum drift |
| CN103917166A (en) * | 2011-08-17 | 2014-07-09 | Vp诊断公司 | A method and system of characterization of carotid plaque |
| CN104408483A (en) * | 2014-12-08 | 2015-03-11 | 西安电子科技大学 | Deep neural network-based SAR texture image classification method |
| CN105046664A (en) * | 2015-07-13 | 2015-11-11 | 广东工业大学 | Image denoising method based on self-adaptive EPLL algorithm |
-
2015
- 2015-12-04 CN CN201510885535.6A patent/CN105574820A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101630405A (en) * | 2009-08-14 | 2010-01-20 | 重庆市勘测院 | Multi-focusing image fusion method utilizing core Fisher classification and redundant wavelet transformation |
| CN102243711A (en) * | 2011-06-24 | 2011-11-16 | 南京航空航天大学 | Neighbor embedding-based image super-resolution reconstruction method |
| CN103917166A (en) * | 2011-08-17 | 2014-07-09 | Vp诊断公司 | A method and system of characterization of carotid plaque |
| CN103632351A (en) * | 2013-12-16 | 2014-03-12 | 武汉大学 | All-weather traffic image enhancement method based on brightness datum drift |
| CN104408483A (en) * | 2014-12-08 | 2015-03-11 | 西安电子科技大学 | Deep neural network-based SAR texture image classification method |
| CN105046664A (en) * | 2015-07-13 | 2015-11-11 | 广东工业大学 | Image denoising method based on self-adaptive EPLL algorithm |
Non-Patent Citations (2)
| Title |
|---|
| 张春霞 等: "受限波尔兹曼机", 《工程数学学报》 * |
| 钱民: "基于机器学习方法的前列腺病理切片图像识别", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (35)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106203488A (en) * | 2016-07-01 | 2016-12-07 | 福州大学 | A kind of galactophore image Feature fusion based on limited Boltzmann machine |
| CN106203488B (en) * | 2016-07-01 | 2019-09-13 | 福州大学 | A kind of galactophore image Feature fusion based on limited Boltzmann machine |
| CN106331433A (en) * | 2016-08-25 | 2017-01-11 | 上海交通大学 | Video denoising method based on deep recursive neural network |
| CN106331433B (en) * | 2016-08-25 | 2020-04-24 | 上海交通大学 | Video denoising method based on deep recurrent neural network |
| CN106503801A (en) * | 2016-10-18 | 2017-03-15 | 天津工业大学 | Electrical impedance tomography method based on deep learning |
| CN106548169A (en) * | 2016-11-02 | 2017-03-29 | 重庆中科云丛科技有限公司 | Fuzzy literal Enhancement Method and device based on deep neural network |
| CN106548169B (en) * | 2016-11-02 | 2019-04-23 | 重庆中科云从科技有限公司 | Fuzzy literal Enhancement Method and device based on deep neural network |
| CN108229347A (en) * | 2016-12-22 | 2018-06-29 | Tcl集团股份有限公司 | For the method and apparatus of the deep layer displacement of the plan gibbs structure sampling of people's identification |
| CN108229347B (en) * | 2016-12-22 | 2022-01-18 | Tcl科技集团股份有限公司 | Method and apparatus for deep replacement of quasi-Gibbs structure sampling for human recognition |
| WO2018157868A1 (en) * | 2017-03-02 | 2018-09-07 | 江清泉 | Ultrasonic imaging system and imaging method thereof |
| CN109833061A (en) * | 2017-11-24 | 2019-06-04 | 无锡祥生医疗科技股份有限公司 | The method of optimization ultrasonic image-forming system parameter based on deep learning |
| US11564661B2 (en) | 2017-11-24 | 2023-01-31 | Chison Medical Technologies Co., Ltd. | Method for optimizing ultrasonic imaging system parameter based on deep learning |
| CN108154169A (en) * | 2017-12-11 | 2018-06-12 | 北京小米移动软件有限公司 | Image processing method and device |
| CN110197714B (en) * | 2018-02-27 | 2024-02-20 | 希森美康株式会社 | Image analysis method and device and method for generating deep learning algorithm after learning |
| CN110197714A (en) * | 2018-02-27 | 2019-09-03 | 希森美康株式会社 | The generation method of deep learning algorithm after method for analyzing image, device, study |
| WO2019168699A1 (en) * | 2018-03-01 | 2019-09-06 | Fujifilm Sonosite, Inc. | Method and apparatus for annotating ultrasound examinations |
| US11382601B2 (en) | 2018-03-01 | 2022-07-12 | Fujifilm Sonosite, Inc. | Method and apparatus for annotating ultrasound examinations |
| US11475569B2 (en) | 2018-05-30 | 2022-10-18 | Shanghai United Imaging Healthcare Co., Ltd. | Systems and methods for image processing |
| CN108985366A (en) * | 2018-07-06 | 2018-12-11 | 武汉兰丁医学高科技有限公司 | B ultrasound image recognition algorithm and system based on convolution depth network |
| WO2020016836A1 (en) * | 2018-07-18 | 2020-01-23 | Advenio Tecnosys Pvt. Ltd. | System and method for managing the quality of an image |
| CN109102515A (en) * | 2018-07-31 | 2018-12-28 | 浙江杭钢健康产业投资管理有限公司 | A kind of method for cell count based on multiple row depth convolutional neural networks |
| CN109658354A (en) * | 2018-12-20 | 2019-04-19 | 上海联影医疗科技有限公司 | A kind of image enchancing method and system |
| CN109965905B (en) * | 2019-04-11 | 2020-02-11 | 复旦大学 | Contrast region detection imaging method based on deep learning |
| CN109965905A (en) * | 2019-04-11 | 2019-07-05 | 复旦大学 | A kind of radiography region detection imaging method based on deep learning |
| CN110033417B (en) * | 2019-04-12 | 2023-06-13 | 江西财经大学 | Image enhancement method based on deep learning |
| CN110033417A (en) * | 2019-04-12 | 2019-07-19 | 江西财经大学 | A kind of image enchancing method based on deep learning |
| CN110555827A (en) * | 2019-08-06 | 2019-12-10 | 上海工程技术大学 | Ultrasonic imaging information computer processing system based on deep learning drive |
| CN110555827B (en) * | 2019-08-06 | 2022-03-29 | 上海工程技术大学 | Ultrasonic imaging information computer processing system based on deep learning drive |
| CN110694149A (en) * | 2019-10-16 | 2020-01-17 | 山东大学齐鲁医院 | Ultrasonic-assisted muscle identification method and system and auxiliary injection device |
| CN110694149B (en) * | 2019-10-16 | 2021-06-22 | 山东大学齐鲁医院 | Ultrasonic-assisted muscle identification method and system and auxiliary injection device |
| CN112508833A (en) * | 2020-12-09 | 2021-03-16 | 东南大学 | Medical image enhancement method and device based on deep learning |
| CN112508833B (en) * | 2020-12-09 | 2024-03-29 | 东南大学 | Medical image enhancement method and device based on deep learning |
| CN113570499A (en) * | 2021-07-21 | 2021-10-29 | 此刻启动(北京)智能科技有限公司 | Self-adaptive image toning method, system, storage medium and electronic equipment |
| CN114782283A (en) * | 2022-06-16 | 2022-07-22 | 深圳华声医疗技术股份有限公司 | Ultrasonic image enhancement method and device, ultrasonic equipment and storage medium |
| CN114782283B (en) * | 2022-06-16 | 2022-09-02 | 深圳华声医疗技术股份有限公司 | Ultrasonic image enhancement method and device, ultrasonic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105574820A (en) | Deep learning-based adaptive ultrasound image enhancement method | |
| CN107610087B (en) | Tongue coating automatic segmentation method based on deep learning | |
| KR101977174B1 (en) | Apparatus, method and computer program for analyzing image | |
| CN108830326B (en) | Automatic segmentation method and device for MRI (magnetic resonance imaging) image | |
| CN113592845A (en) | Defect detection method and device for battery coating and storage medium | |
| CN107909566A (en) | A kind of image-recognizing method of the cutaneum carcinoma melanoma based on deep learning | |
| CN103996018B (en) | Face identification method based on 4DLBP | |
| CN107993221B (en) | Automatic identification method for vulnerable plaque of cardiovascular Optical Coherence Tomography (OCT) image | |
| CN109977780A (en) | A kind of detection and recognition methods of the diatom based on deep learning algorithm | |
| CN105740758A (en) | Internet video face recognition method based on deep learning | |
| CN101794372B (en) | Method for representing and recognizing gait characteristics based on frequency domain analysis | |
| CN108426994A (en) | Digital holographic microscopy data are analyzed for hematology application | |
| CN112784879A (en) | Medical image segmentation or classification method based on small sample domain self-adaption | |
| CN109102515A (en) | A kind of method for cell count based on multiple row depth convolutional neural networks | |
| CN101558996A (en) | Gait recognition method based on orthogonal projection three-dimensional reconstruction of human motion structure | |
| CN101667289B (en) | Retinal image segmentation method based on NSCT feature extraction and supervised classification | |
| CN104346617A (en) | Cell detection method based on sliding window and depth structure extraction features | |
| CN105760877A (en) | Wool and cashmere identification algorithm based on gray level co-occurrence matrix model | |
| CN110448335A (en) | A kind of fetus head circumference full-automatic measuring method and device based on ultrasound image | |
| Liu et al. | A classification method of glass defect based on multiresolution and information fusion | |
| CN110276763A (en) | It is a kind of that drawing generating method is divided based on the retinal vessel of confidence level and deep learning | |
| CN107330883A (en) | A kind of medical image lesion region positioning and sorting technique | |
| Samiappan et al. | Classification of carotid artery abnormalities in ultrasound images using an artificial neural classifier. | |
| CN105976397A (en) | Target tracking method based on half nonnegative optimization integration learning | |
| CN113780242A (en) | Cross-scene underwater sound target classification method based on model transfer learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160511 |
|
| RJ01 | Rejection of invention patent application after publication |