CN106815853A - To the dividing method and device of retinal vessel in eye fundus image - Google Patents
To the dividing method and device of retinal vessel in eye fundus image Download PDFInfo
- Publication number
- CN106815853A CN106815853A CN201611158462.1A CN201611158462A CN106815853A CN 106815853 A CN106815853 A CN 106815853A CN 201611158462 A CN201611158462 A CN 201611158462A CN 106815853 A CN106815853 A CN 106815853A
- Authority
- CN
- China
- Prior art keywords
- retinal vessel
- fundus image
- eye fundus
- vessel
- retinal
- 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
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic 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
- G06T2207/20036—Morphological image processing
-
- 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
- Eye Examination Apparatus (AREA)
Abstract
The invention discloses a kind of dividing method and device to retinal vessel in eye fundus image.Wherein, the method includes:Obtain eye fundus image;Eye fundus image is processed based on extra large plucked instrument matrix, obtains the first retinal vessel figure;Binary conversion treatment is carried out to the first retinal vessel figure, the second retinal vessel figure is obtained;Blood vessel to being interrupted in the second retinal vessel figure is reconstructed, and obtains the 3rd retinal vessel figure.The present invention solves the technical problem that complexity is high and segmentation result is inaccurate when splitting to retinal vessel in eye fundus image in the prior art.
Description
Technical field
The present invention relates to computer internet field, in particular to a kind of to retinal vessel in eye fundus image
Dividing method and device.
Background technology
In the mankind from the extraneous various information for receiving, the information that there are about more than 80% comes from vision.So eyes are
The most important sense organ of human body.Eyeground optic nerve is the part that brain stretches out, and is the important organs of vision, fundus oculi disease
Often result in hypopsia or permanent loss.In addition, Fundus oculi changes caused by eyeground circulation disorder and systemic disease
Can to some extent show and levy in retina and choroid.Therefore treatment is carried out to fundus photograph can be to some general diseases
Early diagnosis or Index for diagnosis function are provided.
Traditional funduscopy method is checked by the observation analysis of fundus photograph and doctor, due to being shot, punching
The influence of many factors such as the resolution capability of human eye is extended to, it is difficult to accurate, objective, comprehensive analysis is made to eye fundus image, closely
Nian Lai, advanced computer image processing technology has been used for the treatment of eye fundus image and analysis, is rapid, accurate, objective
Analysis eye fundus image provides a scientific method for modernization.Eye ground blood vessel as eye fundus image key character,
It is that every analysis work obtains premise and basis, therefore research to blood vessel segmentation is extremely important, can be complete
The profile of eye ground blood vessel is extracted, will be the follow-up key factor to eye fundus image Treatment Analysis quality.
The position of blood vessel can be positioned to the segmentation of retinal vessel to reduce the mistaken diagnosis of lesion, it is also possible to pass through
The segmentation of blood vessel come detect vascular tree and set up its geometrical relationship come aid in discus nervi optici and depression inspection, it is also possible to pass through
The blood vessel for splitting obtains the various parameters of blood vessel, and checks the exception of blood vessel.Hypertension and diabetes are presented in retina
A upper most important sign is exactly the lopsided exception of blood vessel, such as:Arteriovenous impression, copper wire shape or filamentary silver shape artery, and
Change of flexibility and Branch Angle etc..The aberrant angiogenesis of manual analysis retina are relatively good methods, but, in face of big rule
When the population analysis of mould are observed, limitation just occurs, lack accuracy, uniformity and repeatability, it is therefore desirable to by retina
Out, be presented to the subband of the sight person of readding has vascular morphology measurement data (such as to blood vessel segmentation:Flexibility, Branch Angle etc.) and it is different
Often detection is marked (such as:Arteriovenous impression, copper wire shape or filamentary silver shape artery, outburst area etc. of deformity) retinal images,
Final judgement is made to help see the person of readding.See the form that the person of readding only needs to just be able to observe that by user interface retina
With the result of abnormality detection, and then identify retina lesion sign.The efficiency of screening and accurate can so be greatly improved
Degree.And exactly blood vessel accurately can be partitioned into from eye fundus image to the most important condition of retinal vessel analysis.
With developing rapidly for computer, there is scholar's research to go out the method that optical fundus blood vessel is split automatically, optical fundus blood vessel
The method of segmentation is broadly divided into the method and non-supervisory method of supervision.The dividing method of supervision depends on the classification based training collection to carry out
Training, typically can have relatively good segmentation result than non-supervisory.Non-supervisory dividing method has based on morphologic segmentation side
Method, the algorithm based on the dividing method followed the trail of and based on snake models.Although the method for supervision can obtain one it is relatively good
Vessel segmentation, but this method needs a relatively good training set to be trained, such as in the side with neutral net
When method carries out blood vessel segmentation, it is necessary to need a training set for possessing various change pictures to be trained, if training degree is not enough,
Result just occurs deviation, causes the limitation for using.In non-supervisory dividing method, based on morphologic method such as region
Growth, morphological reconstruction scheduling algorithm, possess a Time Calculation complexity very high, can reach as many as 13 minutes;The side of tracking
Although method can accurately obtain the width of blood vessel when blood vessel is split, it is easy to fall into the crosspoint of blood vessel and branch
Enter deadlock, so as to obtain inaccurate blood vessel segmentation;Method based on skeleton pattern in computation complexity and time complexity all
Higher, general sliced time is all in more than ten minutes.Therefore, the method for above-mentioned optical fundus blood vessel segmentation is all having in actual applications
There are unworthiness and limitation.
For when splitting to retinal vessel in eye fundus image in the prior art, complexity is high and segmentation result is inaccurate
Problem, effective solution is not yet proposed at present.
The content of the invention
A kind of dividing method and device to retinal vessel in eye fundus image is the embodiment of the invention provides, at least to solve
The technical problem that complexity is high and segmentation result is inaccurate when certainly splitting to retinal vessel in eye fundus image in the prior art.
A kind of one side according to embodiments of the present invention, there is provided segmentation side to retinal vessel in eye fundus image
Method, including:Obtain eye fundus image;Eye fundus image is processed based on extra large plucked instrument matrix, obtains the first retinal vessel figure;To
One retinal vessel figure carries out binary conversion treatment, obtains the second retinal vessel figure;To what is interrupted in the second retinal vessel figure
Blood vessel is reconstructed, and obtains the 3rd retinal vessel figure.
Another aspect according to embodiments of the present invention, additionally provides a kind of segmentation dress to retinal vessel in eye fundus image
Put, including:Acquisition module, for obtaining eye fundus image;First processing module, for being carried out to eye fundus image based on extra large plucked instrument matrix
Treatment, obtains the first retinal vessel figure;Second processing module, for carrying out binary conversion treatment to the first retinal vessel figure,
Obtain the second retinal vessel figure;Reconstructed module, for being reconstructed to the blood vessel interrupted in the second retinal vessel figure, obtains
3rd retinal vessel figure.
In embodiments of the present invention, by obtaining eye fundus image, it is primarily based on extra large plucked instrument matrix and eye fundus image is processed,
The first retinal vessel figure is obtained, binary conversion treatment then is carried out to the first retinal vessel figure, obtain the second retinal vessel
Figure, is finally reconstructed to the blood vessel interrupted in the second retinal vessel figure, obtains the 3rd retinal vessel figure, has reached from eye
The purpose of retinal vessel is partitioned into base map picture, retina accurately can be partitioned into from eye fundus image using extra large plucked instrument matrix
Blood vessel, the blood vessel for interrupting is reconstructed after the second retinal vessel figure is obtained can make retinal vessel more accurate, have
Beneficial to morphometry subsequently to blood vessel, it is easy to follow-up analysis and research to disease, and then solve in the prior art to eye
The technical problem that complexity is high and segmentation result is inaccurate when retinal vessel is split in base map picture.
Brief description of the drawings
Accompanying drawing described herein is used for providing a further understanding of the present invention, constitutes the part of the application, this hair
Bright schematic description and description does not constitute inappropriate limitation of the present invention for explaining the present invention.In the accompanying drawings:
Fig. 1 is a kind of according to embodiments of the present invention 1 flow chart to the dividing method of retinal vessel in eye fundus image;
Fig. 2 is according to embodiments of the present invention 1 blood vessel pixel map;
Fig. 3 is the orientation node figure of according to embodiments of the present invention 1 Weighted Coefficients;
Fig. 4 is a kind of according to embodiments of the present invention 2 structure chart to the segmenting device of retinal vessel in eye fundus image;
Fig. 5 is a kind of according to embodiments of the present invention 2 optional segmenting device to retinal vessel in eye fundus image
Structure chart;
Fig. 6 is a kind of according to embodiments of the present invention 2 optional segmenting device to retinal vessel in eye fundus image
Structure chart;
Fig. 7 is a kind of according to embodiments of the present invention 2 optional segmenting device to retinal vessel in eye fundus image
Structure chart;And
Fig. 8 is a kind of according to embodiments of the present invention 2 optional segmenting device to retinal vessel in eye fundus image
Structure chart.
Specific embodiment
In order that those skilled in the art more fully understand the present invention program, below in conjunction with the embodiment of the present invention
Accompanying drawing, is clearly and completely described to the technical scheme in the embodiment of the present invention, it is clear that described embodiment is only
The embodiment of a part of the invention, rather than whole embodiments.Based on the embodiment in the present invention, ordinary skill people
The every other embodiment that member is obtained under the premise of creative work is not made, should all belong to the model of present invention protection
Enclose.
It should be noted that term " first ", " in description and claims of this specification and above-mentioned accompanying drawing
Two " it is etc. for distinguishing similar object, without for describing specific order or precedence.It should be appreciated that so using
Data can exchange in the appropriate case, so as to embodiments of the invention described herein can with except illustrating herein or
Order beyond those of description is implemented.Additionally, term " comprising " and " having " and their any deformation, it is intended that cover
Lid is non-exclusive to be included, for example, the process, method, system, product or the equipment that contain series of steps or unit are not necessarily limited to
Those steps or unit clearly listed, but may include not list clearly or for these processes, method, product
Or other intrinsic steps of equipment or unit.
Embodiment 1
According to embodiments of the present invention, there is provided a kind of method to the dividing method of retinal vessel in eye fundus image is implemented
Example, it is necessary to explanation, can be in the such as one group calculating of computer executable instructions the step of the flow of accompanying drawing is illustrated
Performed in machine system, and, although logical order is shown in flow charts, but in some cases, can be being different from
Order herein performs shown or described step.
Fig. 1 is the dividing method to retinal vessel in eye fundus image according to embodiments of the present invention, as shown in figure 1, should
Method comprises the following steps:
Step S102, obtains eye fundus image.
Specifically, eye fundus image can be obtained by way of taking pictures.
Step S104, is processed eye fundus image based on extra large plucked instrument matrix, obtains the first retinal vessel figure.
Specifically, extra large plucked instrument matrix i.e. Hessian matrixes, function of many variables f (x real-valued for1,x2,...,xn),
If function f (x1,x2,...,xn) second-order partial differential coefficient all exist, then define Hessian matrixes be:
Wherein, DiRepresent to i-th differential operator of variable,That is f (x1,x2,...,xn)
Hessian matrixes are:
Hessian matrixes are actually the second dervative under multivariable situation, and it describes shade of gray in all directions
Change.In n characteristic value of Hessian matrixes, the corresponding characteristic vector of characteristic value of amplitude maximum represents P point curvature
Maximum direction, the minimum corresponding characteristic vector of characteristic value of same amplitude represents the minimum direction of curvature.
Eye fundus image is processed by using extra large plucked instrument matrix, can obtain showing the first retinal blood that blood vessel is present
Guan Tu.
Step S106, binary conversion treatment is carried out to the first retinal vessel figure, obtains the second retinal vessel figure.
Specifically, binary conversion treatment is carried out by the first retinal vessel figure, can be by the first colored retinal blood
Pipe figure is converted into the second retinal vessel figure of black and white.
Step S108, the blood vessel to being interrupted in the second retinal vessel figure is reconstructed, and obtains the 3rd retinal vessel figure.
Specifically, the form of retinal vessel contains very valuable information, but the second view obtained above
The discontinuous situation of blood vessel is there may be in film vessel graph, if a continuous blood vessel is divided into many sections, the shape of blood vessel
State measurement will be inaccurate, it is therefore desirable to find out the blood vessel of interruption, and the part interrupted is reconnected.
In the above embodiment of the present invention, by obtaining eye fundus image, be primarily based on extra large plucked instrument matrix is carried out to eye fundus image
Treatment, obtains the first retinal vessel figure, then carries out binary conversion treatment to the first retinal vessel figure, obtains the second retina
Vessel graph, is finally reconstructed to the blood vessel interrupted in the second retinal vessel figure, obtains the 3rd retinal vessel figure, reaches
The purpose of retinal vessel is partitioned into from eye fundus image, can be accurately partitioned into from eye fundus image using extra large plucked instrument matrix and regarded
Retinal vasculature, the blood vessel for interrupting is reconstructed after the second retinal vessel figure is obtained can make retinal vessel more smart
Really, be conducive to the follow-up morphometry to blood vessel, be easy to follow-up analysis and research to disease, and then solve in the prior art
The technical problem that complexity is high and segmentation result is inaccurate when splitting to retinal vessel in eye fundus image.
In a kind of optional embodiment, before step S104, including step S202:Eye fundus image is pre-processed;
Wherein, pretreatment includes:Obtain the green channel images of eye fundus image and green channel images are filtered and strengthen right
Than the treatment of degree.
In a kind of optional embodiment, in step S202 green channel images are filtered and strengthened with the place of contrast
Reason, including:Green channel images are filtered with treatment using image procossing Mean Filtering Algorithm;And use self adaptation Nogata
Figure equalization algorithm to the green channel images for processing after filtering strengthen the treatment of contrast.
Specifically, removal noise can be filtered to green channel images using image procossing Mean Filtering Algorithm, its
In, filter operator uses three times view disk radius.Can be using adaptive histogram equalization to the green that processes after filtering
Channel image carries out contrast treatment, strengthens contrast, wherein, adaptive histogram equalization (AHE) is for lifting figure
A kind of computer image processing technology of the contrast of picture.Common histogram equalization algorithm is used for the pixel of entire image
Identical histogram is converted, for pixel Distribution value compares image in a balanced way, the effect of common histogram equalization algorithm
Very well, however, the image weighed for pixel value skewness, that is, including part substantially darker than other regions or bright
Image, such as eye fundus image, hence it is evident that the contrast of part darker than other regions or bright cannot effectively strengthen, and general
Logical histogram equalization algorithm is different, and then AHE algorithms redistribute brightness to change by calculating the local histogram of image
Picture contrast.Therefore, adaptive histogram equalization algorithm is more suitable for the present invention, to improve the local contrast of eye fundus image
Spend and obtain more image details.
In a kind of optional embodiment, step S106 includes step S302, using local entropy Threshold Segmentation Algorithm to the
One retinal vessel figure carries out binary conversion treatment.
Specifically, according to the superiority of Entropic thresholding algorithm, using local entropy Threshold Segmentation Algorithm to the first retina
Vessel graph carries out binary conversion treatment, can be very good to isolate background and target.
The principles and methods of local entropy Threshold Segmentation Algorithm:Assuming that gray values of the f (x, y) for image midpoint (x, y) place, f (x,
y)>0, for the image of a secondary M × N sizes, define HfIt is the entropy of the image, i.e.,:
In formulaIt is intensity profile.If M × N is the local window of image, claim Hf
It is the local entropy of image.
Local entropy can react the dispersion degree of gradation of image, and in the place that local entropy is maximum, gradation of image is relatively equal
Even, in the small place of local entropy, gradation of image discreteness is larger, and the big position of value differences is target in window under normal circumstances
With the edge of background, so can be the relatively uniform Target Segmentation of gray scale out according to local entropy.Because local entropy is window
The common contribution of interior many pixels, for the insensitive for noise of single-point, so local entropy also has the effect of filtering.By local entropy
Operand is taken the logarithm than larger in formula definition, the speed of service is slow, define 0<pij<<1, can with Taylor expansion cast out high math power (or
Person's Equivalent Infinitesimal) obtain approximate formula:
The local entropy of image is calculated by above formula, image maximum local entropy position is determined, by maximum local entropy position
The local entropy for calculating image and the similitude of the local entropy split, cut zone is determined according to similitude.When segmentation terminates, i.e.,
Can obtain the binary object image of image.
In a kind of optional embodiment, after step S106, including step S402, the second retinal vessel figure is carried out
Post processing;Wherein, post processing includes:Determine the lesion region of the second retinal vessel figure, and making an uproar of removing that lesion region causes
Sound.
Specifically, because eye fundus image might have interior change, including lesion, perfect blood vessel right and wrong can be partitioned into
Often difficult, it is therefore desirable to the second retinal vessel figure is post-processed.
In a kind of optional embodiment, the lesion region of the second retinal vessel figure is determined in step S402, including:
Step S502, connected region is noted to the second retinal vessel icon.
Step S504, determines isolated area, wherein, isolated area is connected region of the area less than predetermined threshold value.
Step S506, eccentricity, circularity and/or the linearity according to isolated area determine lesion region.
Specifically, when determining the lesion region of the second retinal vessel figure, it is necessary first to the second retinal vessel icon
Note connected region, according to predetermined threshold value, determines connected region of the area less than predetermined threshold value, area can be less than into predetermined threshold value
Connected region be defined as isolated area, then calculate some geometrical properties of isolated area, including but not limited to isolated area
Area, girth, Euler's numbers, eccentricity, circularity and the linearity etc..
Wherein, Euler's numbers subtract the number of connected region inner void for the number of connected region;Eccentricity is isolated region
The oval eccentricity of the major axis in domain and short axle composition, computing formula is:A represents the major axis of isolated area,
B represents the short axle of isolated area;Circularity is the amount of isolated area and the similarity degree of circle, according to round girth and round face
Long-pending computing formula can be obtained, and the computing formula of circularity is:AsRepresent the area of isolated area, LsRepresent isolated region
The girth in domain;The linearity can be obtained according to the girth of isolated area and area ratio, specific computing formula is:
The noise that lesion causes typically is rendered as circular or elliptical shape, and shape of blood vessel is generally line style, therefore
After calculating eccentricity, circularity and the linearity of isolated area, can be with eccentricity threshold value set in advance, circularity threshold
Value and linearity threshold value are compared, it is necessary to illustrate, can select or many in eccentricity, circularity and the linearity
Individual factor is compared, and by comparing determination lesion region, then removes the noise that lesion region causes, and makes the second retinal blood
Pipe figure is more accurate.
In a kind of optional embodiment, step S108 includes:
Step S602, determines the point of interruption of the blood vessel of interruption in the second retinal vessel figure.
Step S604, the corresponding point of destination of the point of interruption is determined using dijkstra's algorithm, wherein, point of destination is vascular skeleton
On node, vascular skeleton is the maximum connected region of area in the second retinal vessel figure.
Step S606, disconnecting point and the corresponding point of destination of the point of interruption.
Specifically, during the point of interruption of the blood vessel interrupted during the second retinal vessel figure is determined in step S602, can be first
Connected component labeling is carried out to the second retinal vessel figure, maximum connected region is chosen as vascular skeleton, to other connections
Each pixel in region carries out 3 × 3 field deep search traversal, if pixel only one of which pixel value in 3 × 3 fields is
1 tie point, then will arrive view disk distance by the tie point labeled as the end points of connected region in the end points of each connected region
Used as the point of interruption, as shown in Fig. 2 unblanketed grid represents blood vessel pixel in Fig. 2, and pixel value is 1 to nearest point, can be seen
To the blood vessel pixel positioned at center, only one of which pixel value is 1 tie point in 3 × 3 fields.
Specifically, determining that the point of interruption is corresponding using dijkstra's algorithm i.e. Dijkstra's algorithm in step S604
During point of destination, pixel of the first retinal vessel figure medium vessels centerline pixels higher than vessel boundary is assumed initially that, choose connection
Maximum pixel is worth pixel as the source node of figure in 7 × 7 fields of the point of interruption in region, and other pixels form its of figure
His node, node is made up of with the boundary cost of node pixel cost and direction cost, wherein pixel cost namely two sections
Image pixel intensities between point are poor, and direction cost refers to the change size followed the trail of by source node to direction during the node.Utilize
Dijkstra's algorithm, the search of shortest path is carried out to source node, when the node for searching is the point above vascular skeleton,
Terminate search, the as point of destination of the node above vascular skeleton for searching.
Wherein, dijkstra's algorithm is typical Shortest Path Searching Algorithm, all to other for calculating a node
The shortest path of node.Be mainly characterized by centered on starting point outer layers extension, until expanding to terminal untill.Algorithm idea
It is a digraph for Weighted Coefficients to make G=(V, E), it is two groups that the vertex set V in figure is divided to, and first group most short to have obtained
(active node in S, often tries to achieve a shortest path to the vertex set S in path later when initial, just adds its corresponding summit
Enter in set S, in whole summits are all added to S);Second group of vertex set U for being not determining shortest path, is adding
During, the shortest path length on each summit any summit in being not more than from source node V to U in always keeping from source node V to S
Shortest path length.Algorithm steps are as follows:
A. when initial, S only includes source point, i.e. the distance of S={ v }, v is 0.U includes other summits in addition to v, i.e.,:U=
{ remaining summit }, if summit u has side in v and U,<u,v>Normally there are weights, if u is not v goes out side abutment points,<u,v>Power
It is ∞ to be worth.
B. one is chosen from U apart from v minimum summit k, (the selected distance is exactly the most short of v to k in k additions S
Path length).
C. it is the new intermediate point for considering with k, the distance on each summit in modification U;If the distance from source point v to summit u (is passed through
Summit k) than original distance (it is short without summit k), then change the distance value of summit u, the summit k of amended distance value away from
Power on plus side.
D. repeat step b and c is in all summits are included in S.
As shown in figure 3, be a digraph for Weighted Coefficients in Fig. 3, when initial, S={ A }, first the node B of shortest path
Add, obtain S={ A, B }, be subsequently adding shortest path node C, obtain S={ A, B, C }, be eventually adding D nodes, obtain S=
{ A, B, C, D }, you can obtain the shortest path A- of A to D>B->C->D.
Specifically, in step S602, after the point of interruption and point of destination is obtained, connection endpoint and point of destination, you can complete blood
Pipe is reconnected.
Embodiment 2
According to embodiments of the present invention, there is provided a kind of product to the segmenting device of retinal vessel in eye fundus image is implemented
Example, Fig. 4 is the segmenting device to retinal vessel in eye fundus image according to embodiments of the present invention, as shown in figure 4, the device bag
Include acquisition module 101, first processing module 103, Second processing module 105 and reconstructed module 107.
Wherein, acquisition module 101, for obtaining eye fundus image;First processing module 103, for based on extra large plucked instrument matrix pair
Eye fundus image is processed, and obtains the first retinal vessel figure;Second processing module 105, for the first retinal vessel figure
Binary conversion treatment is carried out, the second retinal vessel figure is obtained;Reconstructed module 107, for being interrupted in the second retinal vessel figure
Blood vessel be reconstructed, obtain the 3rd retinal vessel figure.
In the above-described embodiments, eye fundus image is obtained by acquisition module 101, is based on by first processing module 103 first
Extra large plucked instrument matrix is processed eye fundus image, obtains the first retinal vessel figure, is then regarded by Second processing module 105 pair first
Retinal vasculature figure carries out binary conversion treatment, obtains the second retinal vessel figure, finally by 107 pairs of the second retinal bloods of reconstructed module
The blood vessel interrupted in pipe figure is reconstructed, and obtains the 3rd retinal vessel figure, has reached and retina is partitioned into from eye fundus image
The purpose of blood vessel, accurately retinal vessel can be partitioned into using extra large plucked instrument matrix from eye fundus image, obtaining the second view
The blood vessel for interrupting is reconstructed after film vessel graph can make retinal vessel more accurate, be conducive to the follow-up form to blood vessel
Measurement, is easy to follow-up analysis and research to disease, and then solve in the prior art to retinal vessel in eye fundus image point
The technical problem that complexity is high when cutting and segmentation result is inaccurate.
Herein it should be noted that above-mentioned acquisition module 101, first processing module 103, Second processing module 105 and weight
Structure module 107 corresponds to the example that the step S102 in embodiment 1 is realized to step S108, above-mentioned module with corresponding step
It is identical with application scenarios, but it is not limited to the disclosure of that of above-described embodiment 1.It should be noted that above-mentioned module is used as device
A part of can be performed in the such as one group computer system of computer executable instructions.
In a kind of optional embodiment, as shown in figure 5, device also includes:Pretreatment module 201, at first
Before reason module 103 is processed eye fundus image based on extra large plucked instrument matrix, eye fundus image is pre-processed;Wherein, pre-process
Including:The green channel images for obtaining eye fundus image and the treatment that green channel images are filtered and strengthened with contrast.
In a kind of optional embodiment, pretreatment module 201 is filtered and strengthens contrast to green channel images
Treatment, including:Green channel images are filtered with treatment using image procossing Mean Filtering Algorithm;And use self adaptation
Algorithm of histogram equalization to the green channel images for processing after filtering strengthen the treatment of contrast.
In a kind of optional embodiment, 105 pairs of the first retinal vessel figures of Second processing module carry out binary conversion treatment,
Including:Binary conversion treatment is carried out to the first retinal vessel figure using local entropy Threshold Segmentation Algorithm.
In a kind of optional embodiment, as shown in fig. 6, device also includes:Post-processing module 301, at second
Reason module is obtained after the second retinal vessel figure, and the second retinal vessel figure is post-processed;Wherein, post processing includes:
Determine the lesion region of the second retinal vessel figure, and remove the noise that lesion region causes.
In a kind of optional embodiment, as shown in fig. 7, post-processing module 301 determines including labeling module 401, first
The determining module 405 of module 403 and second.Wherein, labeling module 401 is used to note connected region to the second retinal vessel icon;
First determining module 403 is used to determine isolated area, wherein, isolated area is connected region of the area less than predetermined threshold value;The
Two determining modules 405 determine lesion region for the eccentricity according to isolated area, circularity and/or the linearity.
In a kind of optional embodiment, as shown in figure 8, reconstructed module 107 includes that the 3rd determining module the 501, the 4th is true
Cover half block 503 and link block 505.Wherein the 3rd determining module 501 is used for the blood for determining to be interrupted in the second retinal vessel figure
The point of interruption of pipe;4th determining module 503 is used to determine the corresponding point of destination of the point of interruption using dijkstra's algorithm, wherein, mesh
Point be the node on vascular skeleton, vascular skeleton is the maximum connected region of area in the second retinal vessel figure;Connection mode
Block 505 is used for disconnecting point and the corresponding point of destination of the point of interruption.
The embodiments of the present invention are for illustration only, and the quality of embodiment is not represented.
In the above embodiment of the present invention, the description to each embodiment all emphasizes particularly on different fields, and does not have in certain embodiment
The part of detailed description, may refer to the associated description of other embodiment.
In several embodiments provided herein, it should be understood that disclosed technology contents, can be by other
Mode is realized.Wherein, device embodiment described above is only schematical, such as division of described unit, Ke Yiwei
A kind of division of logic function, can there is other dividing mode when actually realizing, such as multiple units or component can combine or
Person is desirably integrated into another system, or some features can be ignored, or does not perform.Another, shown or discussed is mutual
Between coupling or direct-coupling or communication connection can be the INDIRECT COUPLING or communication link of unit or module by some interfaces
Connect, can be electrical or other forms.
The unit that is illustrated as separating component can be or may not be it is physically separate, it is aobvious as unit
The part for showing can be or may not be physical location, you can with positioned at a place, or can also be distributed to multiple
On unit.Some or all of unit therein can be according to the actual needs selected to realize the purpose of this embodiment scheme.
In addition, during each functional unit in each embodiment of the invention can be integrated in a processing unit, it is also possible to
It is that unit is individually physically present, it is also possible to which two or more units are integrated in a unit.Above-mentioned integrated list
Unit can both be realized in the form of hardware, it would however also be possible to employ the form of SFU software functional unit is realized.
If the integrated unit is to realize in the form of SFU software functional unit and as independent production marketing or use
When, can store in a computer read/write memory medium.Based on such understanding, technical scheme is substantially
The part for being contributed to prior art in other words or all or part of the technical scheme can be in the form of software products
Embody, the computer software product is stored in a storage medium, including some instructions are used to so that a computer
Equipment (can be personal computer, server or network equipment etc.) perform each embodiment methods described of the invention whole or
Part steps.And foregoing storage medium includes:USB flash disk, read-only storage (ROM, Read-Only Memory), arbitrary access are deposited
Reservoir (RAM, Random Access Memory), mobile hard disk, magnetic disc or CD etc. are various can be with store program codes
Medium.
The above is only the preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, under the premise without departing from the principles of the invention, some improvements and modifications can also be made, these improvements and modifications also should
It is considered as protection scope of the present invention.
Claims (10)
1. a kind of dividing method to retinal vessel in eye fundus image, it is characterised in that including:
Obtain eye fundus image;
The eye fundus image is processed based on extra large plucked instrument matrix, obtains the first retinal vessel figure;
Binary conversion treatment is carried out to the first retinal vessel figure, the second retinal vessel figure is obtained;
Blood vessel to being interrupted in the second retinal vessel figure is reconstructed, and obtains the 3rd retinal vessel figure.
2. method according to claim 1, it is characterised in that treatment is carried out to the eye fundus image based on extra large plucked instrument matrix
Before, including:The eye fundus image is pre-processed;
Wherein, the pretreatment includes:Obtain the green channel images of the eye fundus image and to the green channel images
It is filtered and strengthens the treatment of contrast.
3. method according to claim 2, it is characterised in that the green channel images are filtered and strengthened with contrast
The treatment of degree, including:
Treatment is filtered to the green channel images using image procossing Mean Filtering Algorithm;
The green channel images by the filtering process are carried out with enhancing using adaptive histogram equalization algorithm right
Than the treatment of degree.
4. method according to claim 1, it is characterised in that carried out at binaryzation to the first retinal vessel figure
Reason, including:Binary conversion treatment is carried out to the first retinal vessel figure using local entropy Threshold Segmentation Algorithm.
5. method according to claim 1, it is characterised in that after obtaining the second retinal vessel figure, including:To described
Second retinal vessel figure is post-processed;
Wherein, the post processing includes:Determine the lesion region of the second retinal vessel figure, and remove the lesion region
The noise for causing.
6. method according to claim 5, it is characterised in that determine the lesion region of the second retinal vessel figure,
Including:
Connected region is noted to the second retinal vessel icon;
Determine isolated area, wherein, the isolated area is the connected region of the area less than predetermined threshold value;
Eccentricity, circularity and/or the linearity according to the isolated area determine the lesion region.
7. method according to claim 1, it is characterised in that the blood vessel to being interrupted in the second retinal vessel figure enters
Line reconstruction, including:
Determine the point of interruption of the blood vessel of interruption in the second retinal vessel figure;
Determine the corresponding point of destination of the point of interruption using dijkstra's algorithm, wherein, the point of destination is on vascular skeleton
Node, the vascular skeleton is the connected region of area maximum in the second retinal vessel figure;
Connect the point of interruption and the corresponding point of destination of the point of interruption.
8. a kind of segmenting device to retinal vessel in eye fundus image, it is characterised in that including:
Acquisition module, for obtaining eye fundus image;
First processing module, for being processed the eye fundus image based on extra large plucked instrument matrix, obtains the first retinal vessel figure;
Second processing module, for carrying out binary conversion treatment to the first retinal vessel figure, obtains the second retinal vessel
Figure;
Reconstructed module, for being reconstructed to the blood vessel interrupted in the second retinal vessel figure, obtains the 3rd retinal blood
Guan Tu.
9. device according to claim 8, it is characterised in that described device also includes:Pretreatment module, for described
Before first processing module is processed the eye fundus image based on extra large plucked instrument matrix, the eye fundus image is pre-processed;
Wherein, the pretreatment includes:Obtain the green channel images of the eye fundus image and to the green channel images
It is filtered and strengthens the treatment of contrast.
10. device according to claim 8, it is characterised in that described device also includes:Post-processing module, in institute
State after Second processing module obtains the second retinal vessel figure, the second retinal vessel figure is post-processed;
Wherein, the post processing includes:Determine the lesion region of the second retinal vessel figure, and remove the lesion region
The noise for causing.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611158462.1A CN106815853B (en) | 2016-12-14 | 2016-12-14 | Method and device for segmenting retinal blood vessels in fundus image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611158462.1A CN106815853B (en) | 2016-12-14 | 2016-12-14 | Method and device for segmenting retinal blood vessels in fundus image |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106815853A true CN106815853A (en) | 2017-06-09 |
CN106815853B CN106815853B (en) | 2020-06-05 |
Family
ID=59109622
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611158462.1A Active CN106815853B (en) | 2016-12-14 | 2016-12-14 | Method and device for segmenting retinal blood vessels in fundus image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106815853B (en) |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108198185A (en) * | 2017-11-20 | 2018-06-22 | 海纳医信(北京)软件科技有限责任公司 | Dividing method and device, storage medium, the processor of eyeground lesion image |
CN108205807A (en) * | 2017-12-29 | 2018-06-26 | 四川和生视界医药技术开发有限公司 | Retinal vascular images edit methods and retinal vascular images editing device |
CN108230322A (en) * | 2018-01-28 | 2018-06-29 | 浙江大学 | A kind of eyeground feature detection device based on weak sample labeling |
CN108269257A (en) * | 2018-03-05 | 2018-07-10 | 四川和生视界医药技术开发有限公司 | The edit methods and editing device of retinal vessel |
CN108376048A (en) * | 2018-03-05 | 2018-08-07 | 四川和生视界医药技术开发有限公司 | The restorative procedure and prosthetic device of retinal vessel |
CN108416749A (en) * | 2018-02-28 | 2018-08-17 | 沈阳东软医疗系统有限公司 | A kind of ultrasonoscopy processing method, device and computer equipment |
CN108447552A (en) * | 2018-03-05 | 2018-08-24 | 四川和生视界医药技术开发有限公司 | The edit methods of retinal vessel and the editing device of retinal vessel |
CN108805871A (en) * | 2018-06-14 | 2018-11-13 | 艾瑞迈迪医疗科技(北京)有限公司 | Blood-vessel image processing method, device, computer equipment and storage medium |
CN109829931A (en) * | 2019-01-07 | 2019-05-31 | 三峡大学 | A kind of Segmentation Method of Retinal Blood Vessels increasing PCNN based on region |
CN110599417A (en) * | 2019-09-04 | 2019-12-20 | 中国人民解放军63677部队 | Retina image blood vessel extraction method based on Hessian matrix |
CN111493853A (en) * | 2020-04-24 | 2020-08-07 | 天津恒宇医疗科技有限公司 | Blood vessel parameter evaluation method and system for angiodermic diseases |
CN111612743A (en) * | 2020-04-24 | 2020-09-01 | 杭州电子科技大学 | Coronary artery central line extraction method based on CT image |
CN111861999A (en) * | 2020-06-24 | 2020-10-30 | 北京百度网讯科技有限公司 | Detection method and device for artery and vein cross compression sign, electronic equipment and readable storage medium |
CN113344842A (en) * | 2021-03-24 | 2021-09-03 | 同济大学 | Blood vessel labeling method of ultrasonic image |
CN113379741A (en) * | 2021-08-10 | 2021-09-10 | 湖南师范大学 | Retinal blood vessel segmentation method, device and storage medium based on blood vessel characteristics |
CN113822839A (en) * | 2020-06-18 | 2021-12-21 | 飞依诺科技(苏州)有限公司 | Medical image processing method and device, computer equipment and storage medium |
CN115511883A (en) * | 2022-11-10 | 2022-12-23 | 北京鹰瞳科技发展股份有限公司 | Method, apparatus and storage medium for determining curvature of retinal fundus blood vessel |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140015994A1 (en) * | 2012-07-12 | 2014-01-16 | Ramot At Tel-Aviv University Ltd. | Method and system for reducing chromatic aberration |
CN103871057A (en) * | 2014-03-11 | 2014-06-18 | 深圳市旭东数字医学影像技术有限公司 | Magnetic resonance image-based bone segmentation method and system thereof |
CN104574319A (en) * | 2015-01-22 | 2015-04-29 | 深圳大学 | Blood vessel reinforcing method and system of lung CT picture |
CN105989598A (en) * | 2015-02-13 | 2016-10-05 | 中国科学院沈阳自动化研究所 | Eye fundus image vessel segmentation method based on local enhancement active contour module |
CN106097378A (en) * | 2016-07-24 | 2016-11-09 | 江西理工大学 | A kind of level set retinal vascular images dividing method merging shape prior |
-
2016
- 2016-12-14 CN CN201611158462.1A patent/CN106815853B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140015994A1 (en) * | 2012-07-12 | 2014-01-16 | Ramot At Tel-Aviv University Ltd. | Method and system for reducing chromatic aberration |
CN103871057A (en) * | 2014-03-11 | 2014-06-18 | 深圳市旭东数字医学影像技术有限公司 | Magnetic resonance image-based bone segmentation method and system thereof |
CN104574319A (en) * | 2015-01-22 | 2015-04-29 | 深圳大学 | Blood vessel reinforcing method and system of lung CT picture |
CN105989598A (en) * | 2015-02-13 | 2016-10-05 | 中国科学院沈阳自动化研究所 | Eye fundus image vessel segmentation method based on local enhancement active contour module |
CN106097378A (en) * | 2016-07-24 | 2016-11-09 | 江西理工大学 | A kind of level set retinal vascular images dividing method merging shape prior |
Non-Patent Citations (2)
Title |
---|
梁礼明 等: "基于Hessian矩阵和水平集的视网膜血管分割", 《科学技术与工程》 * |
黎飞: "视网膜血管图像增强与检测", 《中国优秀硕士学位论文全文数据库.信息科技辑》 * |
Cited By (22)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108198185A (en) * | 2017-11-20 | 2018-06-22 | 海纳医信(北京)软件科技有限责任公司 | Dividing method and device, storage medium, the processor of eyeground lesion image |
CN108198185B (en) * | 2017-11-20 | 2020-10-16 | 海纳医信(北京)软件科技有限责任公司 | Segmentation method and device for fundus focus image, storage medium and processor |
CN108205807A (en) * | 2017-12-29 | 2018-06-26 | 四川和生视界医药技术开发有限公司 | Retinal vascular images edit methods and retinal vascular images editing device |
CN108230322A (en) * | 2018-01-28 | 2018-06-29 | 浙江大学 | A kind of eyeground feature detection device based on weak sample labeling |
CN108230322B (en) * | 2018-01-28 | 2021-11-09 | 浙江大学 | Eye ground characteristic detection device based on weak sample mark |
CN108416749A (en) * | 2018-02-28 | 2018-08-17 | 沈阳东软医疗系统有限公司 | A kind of ultrasonoscopy processing method, device and computer equipment |
CN108447552A (en) * | 2018-03-05 | 2018-08-24 | 四川和生视界医药技术开发有限公司 | The edit methods of retinal vessel and the editing device of retinal vessel |
CN108376048B (en) * | 2018-03-05 | 2021-05-28 | 四川和生视界医药技术开发有限公司 | Retinal blood vessel repair method and repair device |
CN108376048A (en) * | 2018-03-05 | 2018-08-07 | 四川和生视界医药技术开发有限公司 | The restorative procedure and prosthetic device of retinal vessel |
CN108269257A (en) * | 2018-03-05 | 2018-07-10 | 四川和生视界医药技术开发有限公司 | The edit methods and editing device of retinal vessel |
CN108805871A (en) * | 2018-06-14 | 2018-11-13 | 艾瑞迈迪医疗科技(北京)有限公司 | Blood-vessel image processing method, device, computer equipment and storage medium |
CN109829931A (en) * | 2019-01-07 | 2019-05-31 | 三峡大学 | A kind of Segmentation Method of Retinal Blood Vessels increasing PCNN based on region |
CN109829931B (en) * | 2019-01-07 | 2023-07-11 | 三峡大学 | Retinal vessel segmentation method based on region growing PCNN |
CN110599417A (en) * | 2019-09-04 | 2019-12-20 | 中国人民解放军63677部队 | Retina image blood vessel extraction method based on Hessian matrix |
CN111493853A (en) * | 2020-04-24 | 2020-08-07 | 天津恒宇医疗科技有限公司 | Blood vessel parameter evaluation method and system for angiodermic diseases |
CN111612743A (en) * | 2020-04-24 | 2020-09-01 | 杭州电子科技大学 | Coronary artery central line extraction method based on CT image |
CN113822839A (en) * | 2020-06-18 | 2021-12-21 | 飞依诺科技(苏州)有限公司 | Medical image processing method and device, computer equipment and storage medium |
CN113822839B (en) * | 2020-06-18 | 2024-01-23 | 飞依诺科技股份有限公司 | Medical image processing method, medical image processing device, computer equipment and storage medium |
CN111861999A (en) * | 2020-06-24 | 2020-10-30 | 北京百度网讯科技有限公司 | Detection method and device for artery and vein cross compression sign, electronic equipment and readable storage medium |
CN113344842A (en) * | 2021-03-24 | 2021-09-03 | 同济大学 | Blood vessel labeling method of ultrasonic image |
CN113379741A (en) * | 2021-08-10 | 2021-09-10 | 湖南师范大学 | Retinal blood vessel segmentation method, device and storage medium based on blood vessel characteristics |
CN115511883A (en) * | 2022-11-10 | 2022-12-23 | 北京鹰瞳科技发展股份有限公司 | Method, apparatus and storage medium for determining curvature of retinal fundus blood vessel |
Also Published As
Publication number | Publication date |
---|---|
CN106815853B (en) | 2020-06-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106815853A (en) | To the dividing method and device of retinal vessel in eye fundus image | |
Kumar et al. | An automated early diabetic retinopathy detection through improved blood vessel and optic disc segmentation | |
CN111340789A (en) | Method, device, equipment and storage medium for identifying and quantifying eye fundus retinal blood vessels | |
Trucco et al. | Modeling the tortuosity of retinal vessels: does caliber play a role? | |
CN110197493A (en) | Eye fundus image blood vessel segmentation method | |
Claro et al. | An hybrid feature space from texture information and transfer learning for glaucoma classification | |
CN110751636B (en) | Fundus image retinal arteriosclerosis detection method based on improved coding and decoding network | |
US20220284584A1 (en) | Computerised tomography image processing | |
CN107330449A (en) | A kind of BDR sign detection method and device | |
CN112017185B (en) | Focus segmentation method, device and storage medium | |
CN108764342B (en) | Semantic segmentation method for optic discs and optic cups in fundus image | |
CN111882566B (en) | Blood vessel segmentation method, device, equipment and storage medium for retina image | |
CN107704886A (en) | A kind of medical image hierarchy system and method based on depth convolutional neural networks | |
CN106846293A (en) | Image processing method and device | |
Singh et al. | Extraction of retinal blood vessels by using an extended matched filter based on second derivative of Gaussian | |
CN115578783B (en) | Device and method for identifying eye diseases based on eye images and related products | |
CN108427963A (en) | A kind of dermopathic classifying identification method of melanoma based on deep learning | |
CN111797900B (en) | Artery and vein classification method and device for OCT-A image | |
Rodrigues et al. | Retinal vessel segmentation using parallel grayscale skeletonization algorithm and mathematical morphology | |
CN114692698A (en) | One-dimensional electrocardiogram data classification method based on residual error network | |
CN106408558A (en) | Analysis method of hard exudates and high-reflection signals in diabetic retinopathy image | |
CN115409764A (en) | Multi-mode fundus blood vessel segmentation method and device based on domain self-adaptation | |
Yadav et al. | Automatic Cataract Severity Detection and Grading Using Deep Learning | |
Fondón et al. | Automatic optic cup segmentation algorithm for retinal fundus images based on random forest classifier | |
Joshi et al. | Automated artery-venous classification of retinal blood vessels based on structural mapping method |
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 |