CN109146845A - Head image sign point detecting method based on convolutional neural networks - Google Patents
Head image sign point detecting method based on convolutional neural networks Download PDFInfo
- Publication number
- CN109146845A CN109146845A CN201810774478.8A CN201810774478A CN109146845A CN 109146845 A CN109146845 A CN 109146845A CN 201810774478 A CN201810774478 A CN 201810774478A CN 109146845 A CN109146845 A CN 109146845A
- Authority
- CN
- China
- Prior art keywords
- image
- point
- image block
- index point
- stage
- 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 27
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 15
- 238000012549 training Methods 0.000 claims abstract description 31
- 238000013461 design Methods 0.000 claims abstract description 4
- 238000001914 filtration Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000009432 framing Methods 0.000 claims description 3
- 230000000306 recurrent effect Effects 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 2
- 230000009286 beneficial effect Effects 0.000 abstract description 3
- 238000005516 engineering process Methods 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 8
- 238000001514 detection method Methods 0.000 description 7
- 238000013135 deep learning Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- 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
- G06T2207/20024—Filtering details
- G06T2207/20032—Median filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- 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]
-
- 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
- G06T2207/20112—Image segmentation details
- G06T2207/20132—Image cropping
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30016—Brain
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of head image sign point detecting method based on convolutional neural networks, includes the following steps: data prediction, the selection of training set, the design of network structure and the selection of loss function.Compared with the relevant technologies, the head image sign point detecting method provided by the invention based on convolutional neural networks has the following beneficial effects: can influence with effective noise reduction function data to training result；Influence of the farther away point of range image block to prediction result can be reduced, precision of prediction is finally improved；Improve the positioning accuracy of index point.
Description
Technical field
The present invention relates to a kind of image processing methods, are based particularly on the head image sign point detection side of deep learning
Method.
Background technique
Image key points detection is widely used in object detection, image registration and target identification, solves in medical image
Index point detection is cutd open to be widely used in medical diagnosis on disease, medical image segmentation.Traditional medicine image anatomic points label needs to lead
Domain expert is labeled, and spends a large amount of human resources, and inefficiency, and the index point detection of deep learning is broadly divided into two at present
Class, the method based on the Return Law and the method based on classification.
The first kind is the method based on classification, SercanThe method of the index point detection of equal realizations is to indicate
Point surrounding neighbors take image block, one network of each index point training, which is used to do two classification problems, judge that image block is
No includes index point, to find out candidate marker point position, finally finds out best preferred mark from candidate point using shape
Will point.This method has only used the statistics such as angle, distance to believe the problem is that when selecting optimal candidate point using shape
Breath, it is understood that there may be deviation.
Second class is the method based on recurrence, and Jun Zhang etc. proposes two stages mission-oriented network structure, this method
First stage establishes the association between image block, end-to-end prediction indication point using image block training network parameter, second stage
Location information.This method has preferable effect for a large amount of index point problems, not too much ideal to a small amount of index point effect, and
Directly taking image block using image information entropy, there are still the image blocks that largely can not be applied to finally predict.
Accordingly, it is desirable to provide a kind of new head image sign point detecting method based on convolutional neural networks to solve on
State technical problem.
Summary of the invention
The present invention provides a kind of head image sign point detecting method based on convolutional neural networks, is obviously improved
The positioning accuracy of index point.
To achieve the goals above, technical scheme is as follows:
A kind of head image sign point detecting method based on convolutional neural networks, includes the following steps:
S1, data prediction are cut the image in training set according to given framing mask, are calculated using Sobel
Son extracts image outline information, is carried out smoothly using median filtering to image, is carried out using Canny operator to smoothed image
Processing；
The selection of S2, training set take image block on the processed image of Canny operator at random, if the information of image block
Entropy is greater than preset threshold value, then takes corresponding image block as training on the original image, and the label of first stage network is with 0/1
Label, the label index point to the manhatton distance at the image block center of second stage network；
The design of S3, network structure predict index point position that the first stage uses using two stages network structure
Sorter network trains the image block containing index point, calculates each mark neighborhood of a point；Second stage uses recurrence net
Network takes image block in each index point contiguous range, each index point final position is respectively trained；
The selection of S4, loss function, using the weighting loss based on distance, loss function are as follows:
Wherein N indicates index point number,Indicate the weight of loss；α indicates the adjustment factor of weight, TiIndicate the
The actual position of i point, PiIndicate i-th point of predicted position.
As an improvement of the present invention, in step sl, data prediction specifically comprises the following steps:
Image is cut out according to given profile information；
Use the profile information of Sobel operator extraction image；
Image is smoothed using median filtering；
Smoothed image is further processed using Canny operator, obtains the marginal information of image.
As an improvement of the present invention, in step s3, the first stage takes image block, network to 19 index points respectively
Structure takes class VGG network structure, predicts the image block containing index point and the image block without index point, is determined by calculation
Each mark vertex neighborhood；Each mark vertex neighborhood that second stage is determined according to the first stage, each mark vertex neighborhood training one
A sub-network, network structure take class VGG network structure, predict the location information of each index point.
As an improvement of the present invention, in step s 4, the weighting loss is weighting Euclid's loss.
Compared with the relevant technologies, the head image sign point detecting method provided by the invention based on convolutional neural networks
It has the following beneficial effects:
1, replace image to do training set using image block, and provide the method for data screening, i.e., after image cropping,
Image outline is detected using Sobel operator first, smooth operation is carried out to image using mean filter, is calculated using Canny
Son calculates smoothed out image, calculates marginal information, and last use information entropy takes image block, can effectively reduce and make an uproar
Influence of the sound data to training result；
2, during inverse iteration is calculated and lost, compared with Euclid is used only as loss function, the present invention
The improved Euclid loss proposed can reduce influence of the farther away point of range image block to prediction result, finally improve
Precision of prediction；
3, training network uses two stage training program, and whether first stage training image blocks contain index point, according to
Image block containing index point, calculates each mark neighborhood of a point, second stage according to the neighborhood information of each index point again
It is secondary to take image block, using Recurrent networks, the final position of each index point is respectively trained, improves the positioning accuracy of index point.
Detailed description of the invention
To describe the technical solutions in the embodiments of the present invention more clearly, make required in being described below to embodiment
Attached drawing is briefly described, and drawings in the following description are only some embodiments of the invention, common for this field
For technical staff, without creative efforts, it can also be obtained according to these attached drawings other attached drawings,
In:
Fig. 1 is the flow chart of the head image sign point detecting method provided by the invention based on convolutional neural networks；
Fig. 2A~2E is the pretreatment figure of image；
Fig. 3 A~3D is the image block and label figure for making training set；
Fig. 4 A and 4B are each phase Network structure chart；
Fig. 5 is training result figure.
Specific embodiment
The technical scheme in the embodiments of the invention will be clearly and completely described below, it is clear that described implementation
Example is only a part of the embodiments of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, this field is common
Technical staff's all other embodiment obtained without making creative work belongs to the model that the present invention protects
It encloses.
Referring to Fig. 1, the present invention provides a kind of head image sign point detecting method based on convolutional neural networks, including
Following steps:
S1, data prediction are cut the image in training set according to given framing mask, are calculated using Sobel
Son extracts image outline information, is carried out smoothly using median filtering to image, is carried out using Canny operator to smoothed image
Processing；
Data prediction specifically comprises the following steps:
Image as shown in Figure 2 A is cut out according to given profile information, obtains the institute cut as shown in Figure 2 B
State image；
Specifically, by being cut to image, the part that can be removed edge contour information unrelated with training.
Smoothed image is handled using Sobel operator, obtains the profile information of image, as shown in Figure 2 C；
The image cut is carried out smoothly using median filtering, smoothed image is as shown in Figure 2 D；
Smoothed image is further processed using Canny operator, obtains the marginal information of image, such as Fig. 2 E institute
Show.
Specifically, the Canny operator removes the unwanted marginal point of described image by setting threshold value.
The selection of S2, training set take image block on the processed image of Canny operator at random, if the information of image block
Entropy is greater than preset threshold value, then takes corresponding image block as training set on the original image, and the label of first stage network is used
0/1 label, the manhatton distance of the web tab index point to the image block of second stage；
Specifically, the original image refer to do not cut before image.As shown in figure 3, in the first stage, it is random to select
The 500 big small image block for being 84 × 84 is taken, calculates separately the comentropy of each image, setting threshold value is 0.2, if image information
Entropy is greater than threshold value, then is used to do training set, less than the image block of threshold value, is not used in the training of final result；Second stage phase
Same principle takes 38 × 38 image block.
The design of S3, network structure predict index point position that the first stage uses using two stages network structure
Sorter network trains the image block containing index point, calculates each mark neighborhood of a point；Second stage uses recurrence net
Network takes image block in each index point contiguous range, each index point final position is respectively trained；
Specifically, network structure provided by the invention is as shown in Figure 4, wherein the first stage respectively takes 19 index points
Image block, as shown in Figure 4 A, network structure take class VGG network structure, predict the image block containing index point and without mark
The image block of point, is determined by calculation each mark vertex neighborhood；Second stage is adjacent according to each index point that the first stage determines
Domain, one sub-network of each mark vertex neighborhood training, as shown in Figure 4 B, network structure takes class VGG network structure, and prediction is each
The location information of index point.
Specific training result is as shown in figure 5, Fig. 5 the first row is the actual position of 3 groups of x-ray head image sign points, second
Row is the training result final using two-stage network, it can be found that using the available more accurate detection of two-stage network
Position.
The selection of S4, loss function, using the weighting loss based on distance, loss function are as follows:
Wherein N indicates index point number,Indicate the weight of loss；α indicates the adjustment factor of weight, TiIndicate the
The actual position of i point, PiIndicate i-th point of predicted position.
Specifically, the weighting loss is weighting Euclid's loss.Result is trained for cutting image block, is repaired
Loss function is changed, has made the distance dependent of loss with index point range image block.If the gap mistake between predicted value and true value
Greatly, its influence to final result known to loss weight is smaller, can effectively reduce because of the individual difference band between image
The error come.
Compared with the relevant technologies, the head image sign point detecting method provided by the invention based on convolutional neural networks
It has the following beneficial effects:
1, replace image to do training set using image block, and provide the method for data screening, i.e., after image cropping,
Image outline is detected using Sobel operator, smooth operation is carried out to image using mean filter, uses Canny operator pair
Smoothed out image is calculated, and marginal information is calculated, and last use information entropy takes image block, can be with effective noise reduction function number
According to the influence to training result；
2, during inverse iteration is calculated and lost, compared with Euclid is used only as loss function, the present invention
The improved Euclid loss proposed can reduce influence of the farther away point of range image block to prediction result, finally improve
Precision of prediction；
3, training network uses two stage training program, and whether first stage training image blocks contain index point, according to
Image block containing index point, calculates each mark neighborhood of a point, second stage according to the neighborhood information of each index point again
It is secondary to take image block, using Recurrent networks, the final position of each index point is respectively trained, improves the positioning accuracy of index point.
The above description is only an embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair
Equivalent structure or equivalent flow shift made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant skills
Art field, is included within the scope of the present invention.
Claims (4)
1. a kind of head image sign point detecting method based on convolutional neural networks, which comprises the steps of:
S1, data prediction are cut the image in training set according to given framing mask, are mentioned using Sobel operator
Take image edge information, image is carried out using median filtering it is smooth, using Canny operator to smoothed image at
Reason；
The selection of S2, training set take image block on the processed image of Canny operator at random, if the comentropy of image block is big
In preset threshold value, then take corresponding image block as training set, 0/1 mark of label of first stage network on the original image
Label, the label index point to the manhatton distance at the image block center of second stage network；
The design of S3, network structure predict index point position that the first stage uses classification using two stages network structure
Network trains the image block containing index point, calculates each mark neighborhood of a point；Second stage uses Recurrent networks,
Each index point contiguous range takes image block, and each index point final position is respectively trained；
The selection of S4, loss function, using the weighting loss based on distance, loss function are as follows:
Wherein N indicates index point number,Indicate the weight of loss；α indicates the adjustment factor of weight, TiIt indicates i-th
The actual position of point, PiIndicate i-th point of predicted position.
2. the head image sign point detecting method according to claim 1 based on convolutional neural networks, which is characterized in that
In step sl, data prediction specifically comprises the following steps:
Image is cut out according to given profile information；
Use Sobel operator extraction image outline information；
Image is smoothed using median filter；
Smoothed image is further processed using Canny operator, obtains the marginal information of image.
3. the head image sign point detecting method according to claim 2 based on convolutional neural networks, which is characterized in that
In step s3, the first stage takes image block to 19 index points respectively, and network structure takes class VGG network structure, and prediction contains
There are the image block of index point and the image block without index point, each mark vertex neighborhood is determined by calculation；Second stage according to
Each mark vertex neighborhood that first stage determines, one sub-network of each mark vertex neighborhood training, network structure take class VGG net
Network structure predicts the location information of each index point.
4. the head image sign point detecting method according to claim 1 based on convolutional neural networks, which is characterized in that
In step s 4, the weighting loss is weighting Euclid's loss.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810774478.8A CN109146845A (en) | 2018-07-16 | 2018-07-16 | Head image sign point detecting method based on convolutional neural networks |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810774478.8A CN109146845A (en) | 2018-07-16 | 2018-07-16 | Head image sign point detecting method based on convolutional neural networks |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109146845A true CN109146845A (en) | 2019-01-04 |
Family
ID=64800748
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810774478.8A Pending CN109146845A (en) | 2018-07-16 | 2018-07-16 | Head image sign point detecting method based on convolutional neural networks |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109146845A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110276289A (en) * | 2019-06-17 | 2019-09-24 | 厦门美图之家科技有限公司 | Generate the method and human face characteristic point method for tracing of Matching Model |
| CN111599432A (en) * | 2020-05-29 | 2020-08-28 | 上海优医基医疗影像设备有限公司 | Three-dimensional craniofacial image feature point mark analysis system and method |
| CN111933253A (en) * | 2020-07-14 | 2020-11-13 | 北京邮电大学 | Neural network-based marking point marking method and device for bone structure image |
| CN113158960A (en) * | 2021-05-06 | 2021-07-23 | 吴国军 | Medical image recognition model construction and recognition method and device |
| CN113241155A (en) * | 2021-03-31 | 2021-08-10 | 正雅齿科科技(上海)有限公司 | Method and system for acquiring mark points in lateral skull tablet |
| CN114820517A (en) * | 2022-04-26 | 2022-07-29 | 杭州隐捷适生物科技有限公司 | System and method for automatically detecting key points of lateral skull tablets based on deep learning |
| WO2023275420A1 (en) * | 2021-07-02 | 2023-01-05 | Panacea Cooperative Research S. Coop. | Method for forensic identification by automatically comparing a 3d model of a skull with one or more photos of a face |
| CN115797730A (en) * | 2023-01-29 | 2023-03-14 | 有方（合肥）医疗科技有限公司 | Model training method and device, and head shadow measurement key point positioning method and device |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120182294A1 (en) * | 2009-07-30 | 2012-07-19 | Universidad De Granada | Forensic identification system using craniofacial superimposition based on soft computing |
| CN103824049A (en) * | 2014-02-17 | 2014-05-28 | 北京旷视科技有限公司 | Cascaded neural network-based face key point detection method |
| CN105868769A (en) * | 2015-01-23 | 2016-08-17 | 阿里巴巴集团控股有限公司 | Method and device for positioning face key points in image |
| CN106485215A (en) * | 2016-09-29 | 2017-03-08 | 西交利物浦大学 | Face occlusion detection method based on depth convolutional neural networks |
| CN106599830A (en) * | 2016-12-09 | 2017-04-26 | 中国科学院自动化研究所 | Method and apparatus for positioning face key points |
| CN106845398A (en) * | 2017-01-19 | 2017-06-13 | 北京小米移动软件有限公司 | Face key independent positioning method and device |
| CN107403201A (en) * | 2017-08-11 | 2017-11-28 | 强深智能医疗科技（昆山）有限公司 | Tumour radiotherapy target area and jeopardize that organ is intelligent, automation delineation method |
| CN107609525A (en) * | 2017-09-19 | 2018-01-19 | 吉林大学 | Remote Sensing Target detection method based on Pruning strategy structure convolutional neural networks |
| CN107704817A (en) * | 2017-09-28 | 2018-02-16 | 成都品果科技有限公司 | A kind of detection algorithm of animal face key point |
| US20180061054A1 (en) * | 2016-08-29 | 2018-03-01 | CephX Technologies Ltd. | Automated Cephalometric Analysis Using Machine Learning |
| US10007988B2 (en) * | 2014-02-28 | 2018-06-26 | The Board Of Regents Of The Nevada System Of Higher Education On Behalf Of The University Of Nevada, Las Vegas | Systems and methods for approximating the soft tissue profile of the skull of an unknown subject |
-
2018
- 2018-07-16 CN CN201810774478.8A patent/CN109146845A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120182294A1 (en) * | 2009-07-30 | 2012-07-19 | Universidad De Granada | Forensic identification system using craniofacial superimposition based on soft computing |
| CN103824049A (en) * | 2014-02-17 | 2014-05-28 | 北京旷视科技有限公司 | Cascaded neural network-based face key point detection method |
| US10007988B2 (en) * | 2014-02-28 | 2018-06-26 | The Board Of Regents Of The Nevada System Of Higher Education On Behalf Of The University Of Nevada, Las Vegas | Systems and methods for approximating the soft tissue profile of the skull of an unknown subject |
| CN105868769A (en) * | 2015-01-23 | 2016-08-17 | 阿里巴巴集团控股有限公司 | Method and device for positioning face key points in image |
| US20180061054A1 (en) * | 2016-08-29 | 2018-03-01 | CephX Technologies Ltd. | Automated Cephalometric Analysis Using Machine Learning |
| CN106485215A (en) * | 2016-09-29 | 2017-03-08 | 西交利物浦大学 | Face occlusion detection method based on depth convolutional neural networks |
| CN106599830A (en) * | 2016-12-09 | 2017-04-26 | 中国科学院自动化研究所 | Method and apparatus for positioning face key points |
| CN106845398A (en) * | 2017-01-19 | 2017-06-13 | 北京小米移动软件有限公司 | Face key independent positioning method and device |
| CN107403201A (en) * | 2017-08-11 | 2017-11-28 | 强深智能医疗科技（昆山）有限公司 | Tumour radiotherapy target area and jeopardize that organ is intelligent, automation delineation method |
| CN107609525A (en) * | 2017-09-19 | 2018-01-19 | 吉林大学 | Remote Sensing Target detection method based on Pruning strategy structure convolutional neural networks |
| CN107704817A (en) * | 2017-09-28 | 2018-02-16 | 成都品果科技有限公司 | A kind of detection algorithm of animal face key point |
Non-Patent Citations (5)
| Title |
|---|
| HANSANG LEE 等: "Cephalometric Landmark Detection in Dental X-ray Images Using Convolutional Neural Networks", 《PROC.SPIE 10134,MEDICAL IMAGING 2017:COMPUTER-AIDED DIAGNOSIS》 * |
| SERCAN O.ARIK 等: "Fully automated quantitative cephalometry using convolutional neural networks", 《JOURNAL OF MEDICAL IMAGING》 * |
| SIMON ANDERMATT 等: "Multi-dimensional Gated Recurrent Units for Automated Anatomical Landmark Localization", 《ARXIV》 * |
| 陈嫒嫒 等: "基于关键点的服装检索", 《计算机应用》 * |
| 陈锐 等: "基于级联卷积神经网络的人脸关键点定位", 《四川理工学院学报（自然科学版）》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110276289A (en) * | 2019-06-17 | 2019-09-24 | 厦门美图之家科技有限公司 | Generate the method and human face characteristic point method for tracing of Matching Model |
| CN110276289B (en) * | 2019-06-17 | 2021-09-07 | 厦门美图之家科技有限公司 | Method for generating matching model and face characteristic point tracking method |
| CN111599432A (en) * | 2020-05-29 | 2020-08-28 | 上海优医基医疗影像设备有限公司 | Three-dimensional craniofacial image feature point mark analysis system and method |
| CN111599432B (en) * | 2020-05-29 | 2024-04-02 | 上海优医基医疗影像设备有限公司 | Three-dimensional craniofacial image feature point marking analysis system and method |
| CN111933253A (en) * | 2020-07-14 | 2020-11-13 | 北京邮电大学 | Neural network-based marking point marking method and device for bone structure image |
| CN111933253B (en) * | 2020-07-14 | 2022-09-23 | 北京邮电大学 | Neural network-based marking point marking method and device for bone structure image |
| CN113241155A (en) * | 2021-03-31 | 2021-08-10 | 正雅齿科科技(上海)有限公司 | Method and system for acquiring mark points in lateral skull tablet |
| CN113241155B (en) * | 2021-03-31 | 2024-02-09 | 正雅齿科科技(上海)有限公司 | Method and system for acquiring mark points in skull side position slice |
| CN113158960A (en) * | 2021-05-06 | 2021-07-23 | 吴国军 | Medical image recognition model construction and recognition method and device |
| ES2932181A1 (en) * | 2021-07-02 | 2023-01-16 | Panacea Cooperative Res S Coop | Forensic identification procedure by automatic comparison of the 3D model of the skull and photograph(s) of the face (Machine-translation by Google Translate, not legally binding) |
| WO2023275420A1 (en) * | 2021-07-02 | 2023-01-05 | Panacea Cooperative Research S. Coop. | Method for forensic identification by automatically comparing a 3d model of a skull with one or more photos of a face |
| CN114820517A (en) * | 2022-04-26 | 2022-07-29 | 杭州隐捷适生物科技有限公司 | System and method for automatically detecting key points of lateral skull tablets based on deep learning |
| CN115797730A (en) * | 2023-01-29 | 2023-03-14 | 有方（合肥）医疗科技有限公司 | Model training method and device, and head shadow measurement key point positioning method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109146845A (en) | Head image sign point detecting method based on convolutional neural networks | |
| CN108319972B (en) | End-to-end difference network learning method for image semantic segmentation | |
| JP6999812B2 (en) | Bone age evaluation and height prediction model establishment method, its system and its prediction method | |
| CN106780460B (en) | A kind of Lung neoplasm automatic checkout system for chest CT images | |
| Zilly et al. | Boosting convolutional filters with entropy sampling for optic cup and disc image segmentation from fundus images | |
| CN108921089A (en) | Method for detecting lane lines, device and system and storage medium | |
| WO2018107939A1 (en) | Edge completeness-based optimal identification method for image segmentation | |
| CN107506770A (en) | Diabetic retinopathy eye-ground photography standard picture generation method | |
| CN108364280A (en) | Structural cracks automation describes and width accurately measures method and apparatus | |
| CN109636808A (en) | A kind of lobe of the lung dividing method based on full convolutional neural networks | |
| CN107016409A (en) | A kind of image classification method and system based on salient region of image | |
| CN103413120A (en) | Tracking method based on integral and partial recognition of object | |
| CN106096491B (en) | Automatic identification method for microaneurysms in fundus color photographic image | |
| CN107871322A (en) | Iris segmentation method and apparatus | |
| CN111612817A (en) | Target tracking method based on depth feature adaptive fusion and context information | |
| CN106683104A (en) | Prostate magnetic resonance image segmentation method based on integrated depth convolution neural network | |
| CN107292346B (en) | A kind of MR image hippocampus partitioning algorithm based on Local Subspace study | |
| CN108830149A (en) | A kind of detection method and terminal device of target bacteria | |
| CN107123130A (en) | Kernel correlation filtering target tracking method based on superpixel and hybrid hash | |
| CN105117705A (en) | Iris image quality cascade type evaluation method | |
| CN113393454A (en) | Method and device for segmenting pathological target examples in biopsy tissues | |
| CN108596920A (en) | A kind of Target Segmentation method and device based on coloured image | |
| CN104598914A (en) | Skin color detecting method and device | |
| CN109978895A (en) | A kind of tongue body image partition method and device | |
| CN109063630A (en) | A kind of fast vehicle detection method based on separable convolution technique and frame difference compensation policy |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190104 |