CN108629769A - Eye fundus image optic disk localization method and system based on best fraternal similarity - Google Patents
Eye fundus image optic disk localization method and system based on best fraternal similarity Download PDFInfo
- Publication number
- CN108629769A CN108629769A CN201810409425.6A CN201810409425A CN108629769A CN 108629769 A CN108629769 A CN 108629769A CN 201810409425 A CN201810409425 A CN 201810409425A CN 108629769 A CN108629769 A CN 108629769A
- Authority
- CN
- China
- Prior art keywords
- image
- similarity
- optic disk
- matched
- point
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
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
- G06T7/0014—Biomedical image inspection using an image reference approach
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- 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
- G06T2207/30041—Eye; Retina; Ophthalmic
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Eye Examination Apparatus (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses eye fundus image optic disk localization methods and system based on best fraternal similarity, include the following steps:Step (1):The eye fundus image for choosing health extracts the image in minimum enclosed rectangle as template image according to the minimum enclosed rectangle for including optic disk region of healthy eye fundus image;Step (2):Template image and image to be matched are pre-processed;Step (3):Best fraternal similarity between calculation template image and image to be matched;Step (4):It is scanned in image to be matched, the region corresponding to best brother's similarity maximum value is final positioning result.New idea is proposed for the positioning work of eye fundus image optic disk and obtains preferable effect.
Description
Technical field
The present invention relates to the eye fundus image optic disk based on best fraternal similarity (Best-Buddies Similarity) is fixed
Position method and system.
Background technology
In recent years, the positioning and identification of the physiological structure based on eye fundus image are widely studied and applied, this is for timely
Prevent and a kind of blinding ophthalmology diseases such as diagnosis cataract have great importance and act on.But due to suffering from
The patient of ophthalmology disease is more and more, and the quantity of ophthalmologist, oculist cannot be satisfied a large amount of sick Man's Demands, therefore more and more meters
Experts ' Attention and the research Computer Automatic Recognition and diagnostic system based on medical image of calculation machine vision and medical domain.In eye
In base map picture, optic disk is one of most important physiological structure, therefore it is to automatically analyze place to carry out the positioning of efficiently and accurately to optic disk
Manage step extremely important in eye fundus image research work.
The method that domestic and international researcher proposes a variety of optic disk detection and localizations according to the characteristic of eyeground optic disk, these methods
3 classes can be roughly divided into:(1) the optic disk localization method based on vessel properties;(2) the optic disk positioning based on optic disk external appearance characteristic
Method;(3) the optic disk localization method of eye fundus image vessel information and optic disk feature is comprehensively utilized.
Using the special construction of eyeground optic disk region medium vessels and information is moved towards based on the algorithm of vessel properties, when in image
When causing the external appearance characteristic of optic disk not apparent enough there are disturbing factors such as lesion or noises, it still effectively can accurately position and regard
Disk, but this kind of method mostly will be based on stringent geometric templates, which results in such algorithms to have higher algorithm
Complexity.And constructive geometry template needs based on accurately dividing blood vessel structure, regardless of whether there is lesion in image
Region, it is complete and accurately blood vessel segmentation inherently one is relatively difficult and the higher work of complexity, which results in based on
The optic disk localization method of vessel properties usually not only needs longer processing time, can not detection and localization eyeground optic disk in real time,
And this keeps the work of automatic positioning optic disk more complicated.
Brightness, color, shape, size and the texture that algorithm based on optic disk external appearance characteristic has paid close attention to optic disk itself are special
Sign.But in the case of the brightness of the lesion region and optic disk occurred in the eye fundus image is approximate, simply by the brightness of optic disk,
The features such as shape probably accidentally detected lesion region so as to cause positioning optic disk failure, and since lesion region is to regarding
The influence of disk appearance causes the brightness of optic disk and integrality to be interfered and destroyed, this has also seriously affected the accurate of optic disk positioning
Degree.
Therefore it reduces lesion region in eye fundus image and positions the interference of work to optic disk, while avoiding to blood in eye fundus image
The problem of accurate segmentation of pipe structure reduces algorithm complexity, reduces workload, be those skilled in the art's urgent need to resolve.
Invention content
The problem of based on traditional eye fundus image optic disk localization method, the present invention are proposed based on best fraternal phase
Like the eye fundus image optic disk localization method and system of degree.Best brother's similitude not only considers the appearance of eyeground optic disk, but
By a kind of thought of new template matches, in conjunction with eye fundus image optic disk characteristics of gradient change to the best brother in algorithm
Method for measuring similarity is optimized and is improved, which is applied to the field of eye fundus image optic disk positioning for the first time, calculates mould
The RGB difference in appearance of plate image and image to be matched, differences in spatial location, and introduce the gradient measurement based on first derivative and carry out
Optimization, it is ensured that the accuracy of positioning so that the positioning work of eye fundus image optic disk has better robustness, avoids simultaneously
Segmentation to eye fundus image blood vessel structure has higher position success rate and lower algorithm complexity.
The present invention is based on the thought of template matches, the optic disk image of health is chosen as template image, by eye to be positioned
For base map picture as image to be matched, the best fraternal similarity between calculation template image and image to be matched searches for similarity
The maximum region of value be optic disk region.The invention positions work for eye fundus image optic disk and provides a kind of new thinking.
To achieve the goals above, the present invention adopts the following technical scheme that:
As the first aspect of the present invention, the eye fundus image optic disk localization method based on best fraternal similarity is provided;
Based on the eye fundus image optic disk localization method of best fraternal similarity, include the following steps:
Step (1):The eye fundus image for choosing health, according to the external square of minimum comprising optic disk region of healthy eye fundus image
Shape extracts the image in minimum enclosed rectangle as template image;
Step (2):Template image and image to be matched are pre-processed;
Step (3):Best fraternal similarity between calculation template image and image to be matched;
Step (4):It is scanned in image to be matched, the region corresponding to best brother's similarity maximum value is final
Positioning result.
In the step (1), image to be matched is the eye fundus image where optic disk to be positioned.
In the step (2), template image and image to be matched are pre-processed, it is big according to the size of image subblock
Small, i.e. the length and width of image is adjusted the size of template image and image to be matched, is made Prototype drawing as unit of pixel
The size of picture and image to be matched is the integral multiple of image subblock size;Then template image and template image are divided respectively
At the identical nonoverlapping image subblock of several sizes, each image subblock is then respectively seen as a point, to
Form the point set being made of image subblock, the respectively corresponding point set of template image and the corresponding point set of image to be matched;
It is corresponding based on the corresponding point set of template image and image to be matched formed in step (2) in the step (3)
Point set, the distance between point that the point point corresponding with image to be matched that the corresponding point of calculation template image is concentrated is concentrated value, is looked for
Go out best brother's point of two point sets to (Best-Buddies Pairs), best brother is calculated according to the number of best brother's point pair
Younger brother's similarity;
Best brother's point is to being defined as follows:Two point sets are defined first,WithWhereinIndicate the feature point set of template image,Indicate the feature point set of candidate region in image to be matched,
U, V respectively represent the number for concentrating characteristic point, wherein ri,sj∈Rd.There are such a points to { ri∈R,sj∈ S }, work as ri
Nearest neighbor point to set S is sjAnd sjNearest neighbor point to set R is also riWhen, then such point is to being referred to as best brother
Younger brother's point is to (Best-Buddies Pairs, BBP).
Judge best brother's point to Seg (ri,sj, R, S) mathematic(al) representation be:
Wherein, NN (ri, S) and=argmin d (ri, s), s ∈ S indicate riTo the nearest neighbor point method of discrimination of point set S, d
(ri, s) and indicate a kind of arbitrary distance metric, ∧ is indicated and operation, NN (ri, S) and=sjIndicate the r in point set SiArest neighbors
Point is sj, NN (ri, S) and=sj∧NN(sj, R) and=riIndicate riAnd sjNearest neighbor point each other, that is, best brother point pair, it is on the contrary
It is as the same, then Seg (i, j)=1, if not best brother's point pair, then Seg (i, j)=0.
By counting the sum of best brother's point pair between point set R and S, the then obtained point pair of normalized
As a result sum is the end value of best fraternal similitude, and best fraternal similarity measure values BBS (R, S) expression formula is as follows
Formula (2):
Each candidate region in template image and image to be matched is expressed as the point in the spaces xyRGB, calculates two
The value of best fraternal similarity measurement between a point set, it is necessary first to the distance between each pair of point value is obtained, by two points
The distance between measurement be divided into three parts, first, the difference of the color gray value between two points, second is that between two points
Differences in spatial location, third, image gradient difference.
The distance between the point that the point point corresponding with image to be matched that the corresponding point of template image is concentrated is concentrated value d (ri,
sj), following formula (3):
Wherein, ri (A)Indicate the color pixel values for the point that the corresponding point of template image is concentrated;Indicate image pair to be matched
The color pixel values for the point that the point answered is concentrated;ri (L)Indicate the spatial position for the point that the corresponding point of template image is concentrated;It indicates
The corresponding spatial position for putting the point concentrated of image to be matched;ri (G)Indicate the gradient for the point that the corresponding point of template image is concentrated
Value;Indicate the Grad for the point that the corresponding point of image to be matched is concentrated;β is expressed as weight coefficient.
In the step (4), it is based on step (3), searches for the highest region of value of similarity by column in image to be matched,
As final positioning result.
In the step (4), similarity image is constructed:The height of best fraternal similarity is shown by color gradient form
Low, the high part colours of similarity are deep, and the low part colours of similarity are shallow, observe by the naked eye and find face from image to be matched
The most deep region of color, as with the highest region of template image matching degree.
As the second aspect of the present invention, the eye fundus image optic disk positioning system based on best fraternal similarity is provided;
Based on the eye fundus image optic disk positioning system of best fraternal similarity, including:It memory, processor and is stored in
The computer instruction run on memory and on a processor when the computer instruction is run by processor, completes above-mentioned
Step described in one method.
As the third aspect of the present invention, a kind of computer readable storage medium is provided;
A kind of computer readable storage medium, thereon operation have computer instruction, the computer instruction to be transported by processor
When row, the step described in any of the above-described method is completed.
The beneficial effects of the invention are as follows:
(1) present invention utilizes the template matching methods based on best fraternal similitude, and best brother's similarity is a kind of
Strong robustness, and the higher similarity measurements quantity algorithm of operational efficiency can effectively overcome background in a jumble and block, such spy
Property be applied in the positioning work of eye fundus image optic disk, in conjunction with the feature in optic disk region, introduce the image gradient based on first derivative
The algorithm is optimized, lesion region and illumination variation etc. in eye fundus image can not only be overcome to do optic disk positioning work
It disturbs, computation complexity is higher caused by can also avoiding the accurate segmentation to eye fundus image medium vessels structure, cannot achieve in real time
The defect of optic disk positioning.The step (2) forms point set using image subblock, replaces the point formed with pixel in traditional algorithm
Collection.
(2) present invention is applied to eye fundus image and regards after being optimized to best fraternal method for measuring similarity
In the work of disk positioning.Eye fundus image optic disk localization method proposed by the present invention based on best fraternal similarity has been achieved for
Preferable result.The present invention will be compared in experimental result with the method for other positioning optic disks, and of the invention is positioned to
Power, position error all show performance very outstanding, the two are public in DRIVE and DIARETDB1 for position success rate
It is respectively 100% and 97.7% on data set altogether, positioning mean error is respectively 10.4 and 12.9, and unit is pixel.
Description of the drawings
The accompanying drawings which form a part of this application are used for providing further understanding of the present application, and the application's shows
Meaning property embodiment and its explanation do not constitute the improper restriction to the application for explaining the application.
Fig. 1 is the flow chart of optic disk localization method of the present invention.
Fig. 2 (a) and Fig. 2 (b) is the extraction of template image.
The step of Fig. 3 (a)-Fig. 3 (h) is present invention positioning optic disk.
Fig. 4 (a)-Fig. 4 (l) is some examples that the present invention carries out optic disk positioning in DRIVE data sets.
Fig. 5 (a)-Fig. 5 (t) is some examples that the present invention carries out optic disk positioning in DIARETDB1 data sets.
Specific implementation mode
It is noted that following detailed description is all illustrative, it is intended to provide further instruction to the application.Unless another
It indicates, all technical and scientific terms that the present invention uses have logical with the application person of an ordinary skill in the technical field
The identical meanings understood.
It should be noted that term used herein above is merely to describe specific implementation mode, and be not intended to restricted root
According to the illustrative embodiments of the application.As used herein, unless the context clearly indicates otherwise, otherwise singulative
It is also intended to include plural form, additionally, it should be understood that, when in the present specification using term "comprising" and/or " packet
Include " when, indicate existing characteristics, step, operation, device, component and/or combination thereof.
As shown in Figure 1, in a kind of eye fundus image based on best fraternal similarity (Best-Buddies Similarity)
In optic disk localization method:
(1) construction of template image:In the eye fundus image of health, according to doctor to the label in optic disk region, takes and include
The minimum enclosed rectangle in optic disk region is template image, and in Fig. 2 (a) and Fig. 2 (b), rectangle is template image.
(2) image preprocessing:After step (2), in order to ensure the point concentrated carries out matching and not influencing enough
The accuracy of experimental result, while ensureing that calculation amount is smaller, the size for the sub-block divided should be according to the big of selected template
Small to carry out adjustment appropriate, when template image is larger, it is 3 or even bigger that can make the size of sub-block;And when template image compared with
Hour, it can be directly using single pixel as a point, the size of this value can be adjusted according to specific application scenarios.It
Afterwards according to the size of sub-block, the size of original template image and image to be matched is readjusted, it is sub-block to make the size of original image
The integral multiple of size.
Assuming that image subblock size is 3*3, according to this size, by the length and width of template image and image to be matched
Degree is adjusted to the integral multiple of image subblock length and width.Such as:Template image is 523*478, is adjusted to 522*477.
(3) best fraternal similarity of the point between is calculated:Best brother's similarity depends between two point sets most preferably
The number of brother's point pair, and best brother point is to being defined as follows:Two point sets are defined first,WithWhereinIndicate the feature point set of template image,Indicate candidate in image to be matched
The feature point set in region, U, V respectively represent the number for concentrating characteristic point, wherein ri,sj∈Rd.There are such a points to { ri
∈R,sj∈ S }, work as riNearest neighbor point to set S is sjAnd sjNearest neighbor point to set R is also riWhen, then claiming in this way
Point to being best brother's point to (Best-Buddies Pairs, BBP).Judge that the mathematic(al) representation of best brother's point pair is:
Wherein, NN (ri, S) and=argmin d (ri, s), s ∈ S indicate riTo the nearest neighbor point method of discrimination of point set S, wherein
D (ri, s) and indicate a kind of arbitrary distance metric, ∧ is indicated and operation, NN (ri, S) and=sjIndicate the r in point set SiIt is nearest
Adjoint point is sj, NN (ri, S) and=sj∧NN(sj, R) and=riIndicate riAnd sjNearest neighbor point each other, that is, best brother point pair, instead
As the same, then Seg (i, j)=1, if not best brother's point pair, then Seg (i, j)=0.By counting between point set R and S
Best brother's point pair sum, as a result the then sum of the obtained point pair of normalized is best fraternal similitude
End value, and the best fraternal following formula of similarity measurement value expression (2):
Each candidate region in template and image to be matched is expressed as the point in the spaces xyRGB by us, calculates two
The value of best fraternal similarity measurement between a point set, it is necessary first to the distance between each pair of point value is obtained, in conjunction with eyeground
Image optic disk feature, introduces the image gradient based on first derivative, and the distance between two points measurement is divided into three parts,
When the difference of the color gray value between them, second is that the differences in spatial location between them, third, image gradient difference.Point
The following formula of specific distance measure (3) between:
Wherein subscript A illustrates that the color pixel values difference between characteristic point, subscript L indicate the space bit between characteristic point
Distance difference is set, is normalized between [0,1], β is expressed as weight coefficient, and subscript G represents the r of sub-blockiAnd sjGrad.
(4) according to the value of best fraternal similarity, similarity image is constructed, the maximum region of value for searching for similarity is
Final positioning result.It is to position successfully that the distance between center and optic disk center of institute localization region, which are less than 60 pixels,.
Fig. 3 (a), Fig. 3 (e) are image to be matched, and the rectangular area in Fig. 3 (b), Fig. 3 (f) is template image,
Fig. 3 (c), Fig. 3 (g) are similarity image, and the cross in Fig. 3 (d), Fig. 3 (h) is final positioning result.
Cross is the correct position of optic disk in Fig. 4 (a), Fig. 4 (c), Fig. 4 (e), Fig. 4 (g), Fig. 4 (i), Fig. 4 (k),
Cross in Fig. 4 (b), Fig. 4 (d), Fig. 4 (f), Fig. 4 (h), Fig. 4 (j), Fig. 4 (l) is experimental result of the present invention.
Fig. 5 (a), Fig. 5 (c), Fig. 5 (e), Fig. 5 (g), Fig. 5 (i), Fig. 5 (k), Fig. 5 (m), Fig. 5 (o), Fig. 5 (q), Fig. 5 (s)
Middle cross is the correct position of optic disk,
Fig. 5 (b), Fig. 5 (d), Fig. 5 (f), Fig. 5 (h), Fig. 5 (j), Fig. 5 (l), Fig. 5 (n), Fig. 5 (p), Fig. 5 (r), Fig. 5 (t)
Middle cross is experimental result of the present invention.
The foregoing is merely the preferred embodiments of the application, are not intended to limit this application, for the skill of this field
For art personnel, the application can have various modifications and variations.Within the spirit and principles of this application, any made by repair
Change, equivalent replacement, improvement etc., should be included within the protection domain of the application.
Claims (9)
1. the eye fundus image optic disk localization method based on best fraternal similarity, characterized in that include the following steps:
Step (1):The eye fundus image for choosing health, according to the minimum enclosed rectangle for including optic disk region of healthy eye fundus image,
The image in minimum enclosed rectangle is extracted as template image;
Step (2):Template image and image to be matched are pre-processed;
Step (3):Best fraternal similarity between calculation template image and image to be matched;
Step (4):It is scanned in image to be matched, the region corresponding to best brother's similarity maximum value is determined for final
Position result.
2. the eye fundus image optic disk localization method as described in claim 1 based on best fraternal similarity, characterized in that described
In step (1), image to be matched is the eye fundus image where optic disk to be positioned.
3. the eye fundus image optic disk localization method as described in claim 1 based on best fraternal similarity, characterized in that described
In step (2), template image and image to be matched are pre-processed, according to the size of image subblock, i.e. image
Length and width adjusts the size of template image and image to be matched as unit of pixel, makes template image and to be matched
The size of image is the integral multiple of image subblock size;Then template image and template image are divided into several rulers respectively
The identical nonoverlapping image subblock of very little size, is then respectively seen as a point by each image subblock, to be formed by image
The point set of sub-block composition, the respectively corresponding point set of template image and the corresponding point set of image to be matched.
4. the eye fundus image optic disk localization method as described in claim 1 based on best fraternal similarity, characterized in that described
In step (3), based on the corresponding point set of template image and the corresponding point set of image to be matched formed in step (2), mould is calculated
The distance between point that the point point corresponding with image to be matched that the corresponding point of plate image is concentrated is concentrated is worth, and finds out two point sets
Best brother's point pair calculates best fraternal similarity according to the number of best brother's point pair.
5. the eye fundus image optic disk localization method as described in claim 1 based on best fraternal similarity, characterized in that by mould
Each candidate region in plate image and image to be matched is expressed as the point in the spaces xyRGB, calculates between two point sets
The value of best brother's similarity measurement, it is necessary first to obtain the distance between each pair of point value, the distance between two points are spent
Amount is divided into three parts, when the difference of the color gray value between two points, second is that the differences in spatial location between two points,
Third, image gradient difference.
6. the eye fundus image optic disk localization method as described in claim 1 based on best fraternal similarity, characterized in that described
In step (4), it is based on step (3), searches for the highest region of value of similarity by column in image to be matched, as final determines
Position result.
7. the eye fundus image optic disk localization method as described in claim 1 based on best fraternal similarity, characterized in that described
In step (4), similarity image is constructed:Show that the height of best fraternal similarity, similarity are high by color gradient form
Part colours are deep, and the low part colours of similarity are shallow, observe by the naked eye and find the most deep region of color from image to be matched,
As with the highest region of template image matching degree.
8. the eye fundus image optic disk positioning system based on best fraternal similarity, characterized in that including:Memory, processor with
And the computer instruction run on a memory and on a processor is stored, it is complete when the computer instruction is run by processor
At the step described in the claims 1-7 either method.
9. a kind of computer readable storage medium, characterized in that operation has computer instruction, the computer instruction to be located thereon
When managing device operation, the step described in the claims 1-7 either method is completed.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810409425.6A CN108629769B (en) | 2018-05-02 | 2018-05-02 | Fundus image optic disk positioning method and system based on optimal brother similarity |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810409425.6A CN108629769B (en) | 2018-05-02 | 2018-05-02 | Fundus image optic disk positioning method and system based on optimal brother similarity |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108629769A true CN108629769A (en) | 2018-10-09 |
CN108629769B CN108629769B (en) | 2020-09-29 |
Family
ID=63695220
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810409425.6A Active CN108629769B (en) | 2018-05-02 | 2018-05-02 | Fundus image optic disk positioning method and system based on optimal brother similarity |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108629769B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109800787A (en) * | 2018-12-14 | 2019-05-24 | 西安交通大学 | Image template matching method based on relative characteristic range error measurement |
CN110459299A (en) * | 2019-07-10 | 2019-11-15 | 中山大学 | A kind of retina color fundus photograph image screening technique |
CN110503639A (en) * | 2019-08-15 | 2019-11-26 | 依未科技(北京)有限公司 | The method and apparatus for handling eye fundus image |
CN111709434A (en) * | 2020-06-28 | 2020-09-25 | 哈尔滨工业大学 | Robust multi-scale template matching method based on nearest neighbor feature point matching |
CN112712521A (en) * | 2021-01-18 | 2021-04-27 | 佛山科学技术学院 | Automatic fundus optic disk positioning method based on global gradient search |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102908120A (en) * | 2012-10-09 | 2013-02-06 | 北京大恒图像视觉有限公司 | Eye fundus image registration method, eye fundus image optic disk nerve and vessel measuring method and eye fundus image matching method |
US20160253550A1 (en) * | 2013-11-11 | 2016-09-01 | Beijing Techshino Technology Co., Ltd. | Eye location method and device |
-
2018
- 2018-05-02 CN CN201810409425.6A patent/CN108629769B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102908120A (en) * | 2012-10-09 | 2013-02-06 | 北京大恒图像视觉有限公司 | Eye fundus image registration method, eye fundus image optic disk nerve and vessel measuring method and eye fundus image matching method |
US20160253550A1 (en) * | 2013-11-11 | 2016-09-01 | Beijing Techshino Technology Co., Ltd. | Eye location method and device |
Non-Patent Citations (2)
Title |
---|
TALI DEKEL等: "Best-Buddies Similarity for Robust Template", 《2015 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR)》 * |
邹北骥等: "彩色眼底图像视盘自动定位与分割", 《光学精密工程》 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109800787A (en) * | 2018-12-14 | 2019-05-24 | 西安交通大学 | Image template matching method based on relative characteristic range error measurement |
CN109800787B (en) * | 2018-12-14 | 2020-12-29 | 西安交通大学 | Image template matching method based on relative feature distance error measurement |
CN110459299A (en) * | 2019-07-10 | 2019-11-15 | 中山大学 | A kind of retina color fundus photograph image screening technique |
CN110459299B (en) * | 2019-07-10 | 2023-06-27 | 中山大学 | Retina fundus color photograph image screening method |
CN110503639A (en) * | 2019-08-15 | 2019-11-26 | 依未科技(北京)有限公司 | The method and apparatus for handling eye fundus image |
CN111709434A (en) * | 2020-06-28 | 2020-09-25 | 哈尔滨工业大学 | Robust multi-scale template matching method based on nearest neighbor feature point matching |
CN111709434B (en) * | 2020-06-28 | 2022-10-04 | 哈尔滨工业大学 | Robust multi-scale template matching method based on nearest neighbor feature point matching |
CN112712521A (en) * | 2021-01-18 | 2021-04-27 | 佛山科学技术学院 | Automatic fundus optic disk positioning method based on global gradient search |
CN112712521B (en) * | 2021-01-18 | 2023-12-12 | 佛山科学技术学院 | Automatic positioning method of fundus optic disk based on global gradient search and storage medium thereof |
Also Published As
Publication number | Publication date |
---|---|
CN108629769B (en) | 2020-09-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108629769A (en) | Eye fundus image optic disk localization method and system based on best fraternal similarity | |
US7646904B2 (en) | Computer-aided classification of anomalies in anatomical structures | |
US9990712B2 (en) | Organ detection and segmentation | |
JP6587610B2 (en) | System for processing images acquired by medical imaging equipment and method for operating the system | |
US8559689B2 (en) | Medical image processing apparatus, method, and program | |
CN113379693B (en) | Capsule endoscope key focus image detection method based on video abstraction technology | |
US8175348B2 (en) | Segmenting colon wall via level set techniques | |
Zheng et al. | Multi-part modeling and segmentation of left atrium in C-arm CT for image-guided ablation of atrial fibrillation | |
CN108985210A (en) | A kind of Eye-controlling focus method and system based on human eye geometrical characteristic | |
El-Baz et al. | A new CAD system for early diagnosis of detected lung nodules | |
van Rikxoort et al. | Automatic segmentation of the liver in computed tomography scans with voxel classification and atlas matching | |
US20150125052A1 (en) | Drusen lesion image detection system | |
Ibragimov et al. | Segmentation of tongue muscles from super-resolution magnetic resonance images | |
He et al. | Fast automatic 3D liver segmentation based on a three‐level AdaBoost‐guided active shape model | |
WO2010035518A1 (en) | Medical image processing apparatus and program | |
Lee et al. | Hybrid airway segmentation using multi-scale tubular structure filters and texture analysis on 3D chest CT scans | |
Yang et al. | Medical Image Segmentation Using Descriptive Image Features. | |
CN110533667B (en) | Lung tumor CT image 3D segmentation method based on image pyramid fusion | |
Yao et al. | Employing topographical height map in colonic polyp measurement and false positive reduction | |
Schultheis et al. | Using deep learning segmentation for endotracheal tube position assessment | |
CN112862786B (en) | CTA image data processing method, device and storage medium | |
Cui et al. | Cobb Angle Measurement Method of Scoliosis Based on U-net Network | |
Merickel Jr et al. | Segmentation of the optic nerve head combining pixel classification and graph search | |
Niemeijer | Automatic detection of diabetic retinopathy in digital fundus photographs | |
CN112862785A (en) | CTA image data identification method, device and storage medium |
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 |