CN107292835B - Method and device for automatically vectorizing retinal blood vessels of fundus image - Google Patents
Method and device for automatically vectorizing retinal blood vessels of fundus image Download PDFInfo
- Publication number
- CN107292835B CN107292835B CN201710397099.7A CN201710397099A CN107292835B CN 107292835 B CN107292835 B CN 107292835B CN 201710397099 A CN201710397099 A CN 201710397099A CN 107292835 B CN107292835 B CN 107292835B
- Authority
- CN
- China
- Prior art keywords
- blood vessel
- image
- channel
- image data
- extraction
- 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.)
- Expired - Fee Related
Links
- 210000001210 retinal vessel Anatomy 0.000 title claims abstract description 35
- 238000000034 method Methods 0.000 title claims abstract description 31
- 210000004204 blood vessel Anatomy 0.000 claims abstract description 180
- 238000000605 extraction Methods 0.000 claims abstract description 112
- 230000000694 effects Effects 0.000 claims abstract description 49
- 238000012545 processing Methods 0.000 claims abstract description 47
- 238000000926 separation method Methods 0.000 claims abstract description 32
- 238000003708 edge detection Methods 0.000 claims description 15
- 238000001914 filtration Methods 0.000 claims description 15
- 230000009467 reduction Effects 0.000 claims description 9
- 238000005457 optimization Methods 0.000 claims description 8
- 238000007670 refining Methods 0.000 claims description 7
- 230000000877 morphologic effect Effects 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 6
- 230000003044 adaptive effect Effects 0.000 claims description 4
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 238000004364 calculation method Methods 0.000 description 11
- 238000010586 diagram Methods 0.000 description 10
- 238000001514 detection method Methods 0.000 description 8
- 230000006870 function Effects 0.000 description 8
- 230000008859 change Effects 0.000 description 7
- 239000013598 vector Substances 0.000 description 7
- 201000010099 disease Diseases 0.000 description 5
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 5
- 206010020772 Hypertension Diseases 0.000 description 4
- 238000005259 measurement Methods 0.000 description 4
- 210000001525 retina Anatomy 0.000 description 4
- 206010003210 Arteriosclerosis Diseases 0.000 description 3
- 208000024172 Cardiovascular disease Diseases 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 3
- 208000011775 arteriosclerosis disease Diseases 0.000 description 3
- 238000005286 illumination Methods 0.000 description 3
- 230000001575 pathological effect Effects 0.000 description 3
- 208000017442 Retinal disease Diseases 0.000 description 2
- 206010038923 Retinopathy Diseases 0.000 description 2
- 206010012601 diabetes mellitus Diseases 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000011835 investigation Methods 0.000 description 2
- 230000004256 retinal image Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 208000006011 Stroke Diseases 0.000 description 1
- 210000001367 artery Anatomy 0.000 description 1
- 230000003143 atherosclerotic effect Effects 0.000 description 1
- 238000003759 clinical diagnosis Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000013075 data extraction Methods 0.000 description 1
- 230000000593 degrading effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000013399 early diagnosis Methods 0.000 description 1
- 238000005111 flow chemistry technique Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 208000019622 heart disease Diseases 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000004089 microcirculation Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000002207 retinal effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 210000003462 vein Anatomy 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
- H04N19/94—Vector quantisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color 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
- G06T2207/30101—Blood vessel; Artery; Vein; Vascular
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Eye Examination Apparatus (AREA)
- Image Processing (AREA)
Abstract
The invention relates to a method and a device for automatically vectorizing retinal blood vessels of fundus images, wherein the method comprises the following steps: performing data separation processing on a G channel image of a fundus image to obtain first image data; determining a blood vessel extraction result of a G channel according to the first image data; performing data separation processing on an R channel image of the fundus image to obtain second image data; determining a blood vessel extraction result of an R channel according to the second image data; weighting the blood vessel extraction result of the G channel and the blood vessel extraction result of the R channel to obtain a first comprehensive extraction effect graph; and determining vectorization data of retinal blood vessels of the fundus image according to the first comprehensive extraction effect map. The embodiment of the invention has good vessel vectorization effect and is beneficial to measuring and calculating the subsequent vessel diameter and the like.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a method and a device for automatically vectorizing retinal blood vessels of an eye fundus image.
Background
Clinical experience shows that the thickness, the shape, the continuity and the dynamic-to-static pulse width ratio (AVR) of blood vessels in fundus images are closely related to the disease course of stroke, hypertension, diabetes, arteriosclerosis and other diseases, for example, arterial narrowing is related to hypertension, and venous narrowing has a high possibility of stroke and coronary atherosclerotic heart disease. How to effectively perform calculation processing such as separation, reduction, continuity, deformation judgment and the like on retinal blood vessels, and further to be used for investigation and early warning of various diseases related to retinopathy is always a key concern in fundus image analysis.
With the rapid increase of fundus image data, doctors only rely on manual observation and experience diagnosis, the subjectivity is strong, the efficiency is low, for example, the arteriovenous width ratio in fundus images can be used for measuring the width change of retinal blood vessels, but because the relative change is small, in clinical detection, even an ophthalmologist with abundant experience can be difficult to distinguish the change, and the experience of the doctors is abundant and poor. Therefore, the method has very important clinical significance in automatically separating blood vessels from fundus images, analyzing continuity and trend of the blood vessels, automatically measuring AVR and the like by using an image processing and computer aided method, is helpful for a doctor to evaluate and diagnose the pathological degree of a patient according to the detection result of the fundus images, and further realizes screening, prevention and early treatment of diseases such as hypertension, diabetes, arteriosclerosis and the like.
In recent years, scholars at home and abroad carry out intensive research on aspects of blood vessel separation, blood vessel vectorization and the like of fundus images, related technologies mainly comprise mathematical morphology, blood vessel tracking, matched filtering, deformation model-based and recently popular machine learning schemes, but due to physical reasons of the fundus images, such as uneven tissue density, rich layers, high noise and high blood vessel tissue density, and phenomena such as uneven illumination, artifacts and the like existing during photographing, the edge of the image is difficult to detect in a self-adaptive manner according to the image characteristics, and edge details obtained by a single edge detection algorithm are not provided with much noise, namely the effect of blood vessel vectorization is unsatisfactory, and the measurement and calculation of the follow-up blood vessel diameter and the like are not facilitated.
Disclosure of Invention
The invention provides a method and a device for automatically vectorizing retinal blood vessels of fundus images, which have good effect of vectorizing the blood vessels and are beneficial to measuring and calculating the subsequent blood vessel diameters and the like.
A first aspect provides a method for automatic vectorization of retinal blood vessels in a fundus image, the method comprising: performing data separation processing on a G channel image of a fundus image to obtain first image data; determining a blood vessel extraction result of a G channel according to the first image data; performing data separation processing on an R channel image of the fundus image to obtain second image data; determining a blood vessel extraction result of an R channel according to the second image data; weighting the blood vessel extraction result of the G channel and the blood vessel extraction result of the R channel to obtain a first comprehensive extraction effect graph; and determining vectorization data of retinal blood vessels of the fundus image according to the first comprehensive extraction effect map.
A second aspect provides an apparatus for automatic vectorization of retinal blood vessels in a fundus image, the apparatus comprising: the first separation module is used for carrying out data separation processing on a G channel image of a fundus image to obtain first image data; the first extraction module is used for determining a blood vessel extraction result of a G channel according to the first image data obtained by the first separation module; the second separation module is used for carrying out data separation processing on the R channel image of the fundus image to obtain second image data; the second extraction module is used for determining a blood vessel extraction result of the R channel according to the second image data obtained by the second separation module; the comprehensive processing module is used for weighting the blood vessel extraction result of the G channel determined by the first extraction module and the blood vessel extraction result of the R channel determined by the second extraction module to obtain a first comprehensive extraction effect graph; and the vectorization processing module is used for determining the vectorization data of the retinal vessel of the fundus image according to the first comprehensive extraction effect image obtained by the comprehensive processing module.
In the embodiment of the invention, on the basis of processing the G channel data, the result extraction of the R channel data is added, a certain weight proportion is given, the result extraction is combined with the G channel extraction result for calculation, and the effect of extracting the blood vessel can be optimized from different dimensions, so that the vectorization effect of the blood vessel is good, and the measurement and calculation of the subsequent blood vessel diameter and the like are facilitated.
Drawings
Fig. 1 is a flowchart of a method for automatic vectorization of retinal blood vessels of a fundus image according to an embodiment of the present invention;
fig. 2 is a diagram illustrating a G channel data extraction effect of a retinal image according to an embodiment of the present invention;
fig. 3 is a diagram illustrating the effect of complex Morlet mother wavelet transform provided by the embodiment of the present invention;
FIG. 4 is a schematic diagram of an active growth edge detection gradient pixel search according to an embodiment of the present invention;
fig. 5 is a schematic diagram of a template of a Sobel operator in 8 directions according to an embodiment of the present invention;
fig. 6 is a schematic block diagram of the overall processing flow of a method for automatic vectorization of retinal blood vessels of a fundus image according to an embodiment of the present invention;
fig. 7 is a structural diagram of an apparatus for automatically vectorizing retinal blood vessels of a fundus image according to an embodiment of the present invention.
Detailed Description
The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
For the convenience of understanding of the embodiments of the present invention, the following description will be further explained with reference to specific embodiments, which are not to be construed as limiting the embodiments of the present invention.
Fig. 1 is a flowchart of a method for automatic vectorization of retinal blood vessels of a fundus image according to an embodiment of the present invention, and with reference to fig. 1, the method includes:
in step 101, data separation processing is performed on a G-channel image of a fundus image, and first image data is obtained.
And step 102, determining a blood vessel extraction result of a G channel according to the first image data.
In one example, an adaptive histogram is adopted to perform equalization optimization on an image corresponding to channel image data to obtain third image data, wherein the channel image data is the first image data; preprocessing the third image data based on a complex Morlet mother wavelet formula to obtain a thick blood vessel value and a thin blood vessel value, segmenting the blood vessel by means of Top-Hat variation and double-loop filtering, and further optimizing a tiny blood vessel part in the image by adopting multi-scale Gaussian matching filtering to obtain a first blood vessel extraction result; and aiming at the first blood vessel extraction result, performing blood vessel edge optimization by adopting an active growth edge detection algorithm, and performing growth reduction on the blood vessel with poor extraction effect to obtain a second blood vessel extraction result.
In step 103, data separation processing is performed on the R-channel image of the fundus image, and second image data is obtained.
And 104, determining a blood vessel extraction result of the R channel according to the second image data.
In one example, the blood vessel extraction result of the R channel may be determined in the same manner as in step 102, which is not described herein.
And 105, weighting the blood vessel extraction result of the G channel and the blood vessel extraction result of the R channel to obtain a first comprehensive extraction effect graph.
And 106, determining vectorization data of retinal blood vessels of the fundus image according to the first comprehensive extraction effect map.
In one example, according to the first comprehensive extraction effect map, removing thin blood vessels smaller than a set pixel in combination with a blood vessel morphological structure; refining the edges of the blood vessels by adopting a Sobel algorithm to obtain the center lines of the blood vessels; based on the blood vessel structure, filtering the intersection points and the bifurcation points of the blood vessels on the central line of the blood vessels to obtain a second comprehensive extraction effect graph; and determining vectorization data of retinal blood vessels of the fundus image according to the second comprehensive extraction effect map.
The method adopts a Sobel algorithm to carry out thinning processing on the edges of the blood vessels based on 8 direction templates to obtain the center lines of the blood vessels.
In the embodiment of the present invention, the blood vessel edge may be located based on the blood vessel center line, and the coordinates of the blood vessel edge and the coordinates of the blood vessel center line may be recorded and stored, where the coordinates of the blood vessel edge and the coordinates of the blood vessel center line constitute vectorized data of a retinal blood vessel of the fundus image.
In the embodiment of the invention, on the basis of processing the G channel data, the result extraction of the R channel data is added, a certain weight proportion is given, the result extraction is combined with the G channel extraction result for calculation, and the effect of extracting the blood vessel can be optimized from different dimensions, so that the vectorization effect of the blood vessel is good, and the measurement and calculation of the subsequent blood vessel diameter and the like are facilitated.
In the embodiment of the invention, a combined algorithm can be adopted to perform automatic vectorization flow processing on the fundus blood vessels aiming at the retina pictures shot by the fundus camera.
In one example, based on a complex Morlet mother wavelet detection algorithm, in combination with an actively growing edge connection algorithm and a Sobel edge detection refinement algorithm, a new blood vessel vectorization combination calculation mode is adopted, and through the improved combination technical method, the blood vessel separation effect can be better optimized, the blood vessel image is subjected to structure reduction and boundary coordinate positioning processing, and reliable premises are provided for subsequent classification of blood vessel artery and vein, calculation of the diameter ratio and feature prediction of cardiovascular diseases.
The retina blood vessel network can directly reflect the influence of cardiovascular diseases such as hypertension, arteriosclerosis and the like on the morphological structure of the blood vessel network, and is an important part of cardiovascular disease microcirculation examination, so that the computer image technology is adopted to pre-judge and diagnose pathological characteristics of the fundus picture, and the retina blood vessel network has very important significance for clinical application. The basis of pathological feature prediction and judgment of the fundus picture is to effectively separate, extract and calculate vectors of fundus blood vessels; the embodiment of the invention can effectively remove the fundus noise, better separate the fundus blood vessels, further convert the images into vector data and facilitate the subsequent calculation processing of disease characteristics.
The technical scheme adopted by the embodiment of the invention mainly relates to 3 core algorithms, wherein the method comprises the steps of carrying out blood vessel detection based on a complex Morlet mother wavelet algorithm, carrying out fracture reduction treatment on blood vessels based on an active growth algorithm, and carrying out blood vessel center line extraction based on a Sobel edge detection thinning algorithm. The 3 algorithms are described below.
Complex Morlet mother wavelet algorithm:
practical experience shows that the multiscale analysis of continuous wavelet transform is more suitable for fundus blood vessel detection with different widths, directions and linear structures, because the cross section of a blood vessel can be modeled by adopting a Gaussian function, in the multiscale wavelet transform, a low-resolution filter detects a coarse blood vessel, and a high-resolution filter detects a fine blood vessel, but the fundus image generally has the edges of non-blood vessel structures such as bright spots, uneven chromaticity and the like, and complex Morlet mother wavelets are selected for blood vessel extraction in order to overcome the influence of the complex Morlet mother wavelets on the Gaussian kernel function.
The complex Morlet mother wavelet formula is:
wherein f isbRepresents the bandwidth, fcThe method comprises the steps of representing the central frequency of a wavelet function, taking a real part of a complex Morlet wavelet as a Gaussian linear filter, taking an imaginary part of the complex Morlet wavelet as an edge filter, decomposing an eye fundus image by applying continuous complex wavelets in different scales and directions to obtain corresponding wavelet coefficients, performing wavelet transformation once on the basis of rotation at an angle of 15 degrees each time, and finally taking the maximum value of the real part of a wavelet transformation system as the characteristic of each pixel so as to perform edge extraction on a thin blood vessel and a thick blood vessel respectively when the scale values are different.
Edge connection algorithm based on active growth:
the method comprises the steps of performing primary detection by using a Canny algorithm with a good effect on weak edge detection based on an actively-grown edge connection algorithm in combination with an opening operation and a closing operation in mathematical morphology, then detecting the end point position and direction of an edge line, and growing and degrading the end point by using the property that a missed edge is possibly positioned in the edge extension along the end point direction.
The active growth algorithm adopts the structural element mode as follows:
growing dstij＝max(src(x+x′,y+y′))
Degenerate dstij＝min(src(x+x′,y+y′))
Wherein dstijPixel values representing the coordinate (i, j) points, x, y representing the source pixels, and x ', y' representing the pixels in the structural elements.
Sobel edge detection and refinement algorithm:
the edge detection processing of the image can be simply understood as extracting the outline of the area in the image, the area in the image is divided according to pixels, the gray level of the pixels in each area is approximately the same, the boundary between the areas is called as the edge, and the result of the image edge detection visually looks like the skeleton of the image.
The Sobel edge detection algorithm is based on first order differentiation, gradient is a measure of function change, and an image can be regarded as a sampling point array of an image gray continuous function, so that the obvious change of the image gray value can be detected by using a discrete approximation function of the gradient.
For the image function f (x, y), its gradient at point (x, y) is a vector defined as:
the direction of the gradient is in the direction of the maximum rate of change of the function f (x, y), and the magnitude of the gradient is G [ f (x, y) ]]Expressed, the algorithm formula is:
the Sobel differential operator is an omnidirectional differential operator under an odd-sized template, and a 3x3 convolution template can be expressed as follows:
the Sobel operator introduces a weighted local average factor, so that the Sobel operator has a certain smoothing effect on random noise in the image, and because the Sobel operator is like the difference between two lines or two columns, partial false edges are removed, actual edges are smoothed, and elements on two sides of the edges are enhanced.
The results show that the embodiment of the invention has higher precision in the aspects of blood vessel separation, blood vessel reduction, blood vessel continuity detection and the like by combining a plurality of algorithms, reduces and reconstructs the edges of fine blood vessels as truly as possible, vector results are closer to the hand-drawn results of experts, and the calculation time is far shorter than that of an artificial method, so that the invention has higher practical value for the early diagnosis and investigation system of retinopathy based on the blood vessel detection results, and can provide greater help for clinical diagnosis of ophthalmologists by applying the system to the measurement related to the width of retinal blood vessels or the ratio of arteriovenous vessels subsequently.
A more specific embodiment is provided below to describe a method for automatic vectorization of retinal blood vessels in a fundus image according to the present invention.
As shown in fig. 2, the graph is a graph of the effect of extracting G channel data of a retinal image, and is used for preprocessing a color fundus image, wherein the speed is greatly improved due to the fact that the contrast of blood vessels of the G channel is obvious relative to other channels, and the speed is greatly improved due to the image processing based on a gray-scale image.
As shown in fig. 3, the complex Morlet mother wavelet transform effect graph is a continuous wavelet transform based on multi-scale analysis, and combines a gaussian linear filter and an edge filter, so as to eliminate the influence of bright spots, uneven chromaticity and the like, and be used for extracting the fundus image blood vessels.
As shown in fig. 4, a schematic diagram of searching for gradient pixels for active growing edge detection is shown, which compares the gradient values of the front and the back along the gradient direction to search for the local maximum of the pixel point for vessel edge restoration and to reduce the peripheral noise to a certain extent.
As shown in fig. 5, the schematic diagram is an 8-directional Sobel operator template, which is an improvement of a conventional Sobel algorithm, and on the basis of original 0-degree and 90-degree templates, direction templates of 45 degrees, 135 degrees, 180 degrees, 225 degrees, 270 degrees and 315 degrees are added, so that multi-directional edges of an image can be detected, the problem that too few templates can only detect edges in horizontal and vertical directions is avoided, and the blood vessel edge information is more complete, and the edge refinement method is used for edge refinement of a fundus extraction blood vessel section to finally generate a blood vessel center line.
As shown in fig. 6, a schematic block diagram of the overall processing flow of the method for automatic vectorization of retinal blood vessels of a fundus image according to the embodiment of the present invention includes receiving and processing an image and outputting a vector result. The specific implementation steps comprise:
And step 604, preprocessing the fundus image based on a complex Morlet mother wavelet formula to obtain a thick and thin blood vessel value.
Before step 604, adaptive histogram equalization may be used to optimize the problems of uneven illumination, over-concentrated gray scale distribution, and low contrast in the image.
And step 605, segmenting the blood vessel by using a Top-Hat change and double loop filtering mode, and further optimizing a tiny blood vessel part in the image by adopting multi-scale Gaussian matching filtering.
And 606, aiming at the blood vessel extraction result, optimizing the blood vessel edge by adopting an active growth edge detection algorithm, and growing and restoring the blood vessel with poor extraction effect.
In step 607, weighting setting is performed on the blood vessel extraction result of the G, R channel to obtain a final extraction effect map.
Compared with other vessel vectorization combination modes, the vessel vectorization combination method based on the combination of the complex Morlet wavelet algorithm, the active growth edge connection algorithm and the Sobel edge refinement algorithm is more comprehensive in vessel feature description and has stronger image adaptability and denoising performance aiming at different illumination conditions, so that the vessel extraction effect based on the scheme is better, the vessel skeleton is clearer, the influence of noise such as bright spots and uneven chromaticity on vessel extraction is reduced, and the structural vector reduction can be performed on weakened vessels to a certain degree.
The embodiment of the invention adopts R, G dual-channel blood vessel extraction result weighting. When the blood vessel extraction is carried out on the fundus image, the conventional method only carries out result processing on G channel data, and the method increases the result extraction of the R channel data on the basis of processing the G channel data, gives a certain weight proportion, combines the result extraction with the G channel extraction result for calculation, and can optimize the blood vessel extraction effect from different dimensions.
Fig. 7 is a structural diagram of an apparatus for automatically vectorizing retinal blood vessels of a fundus image according to an embodiment of the present invention, where the apparatus includes:
a first separation module 701, configured to perform data separation processing on a G-channel image of a fundus image to obtain first image data;
a first extraction module 702, configured to determine a blood vessel extraction result of a G channel according to the first image data obtained by the first separation module 701;
a second separation module 703, configured to perform data separation processing on the R-channel image of the fundus image to obtain second image data;
a second extraction module 704, configured to determine a blood vessel extraction result of the R channel according to the second image data obtained by the second separation module 703;
a comprehensive processing module 705, configured to perform weighting processing on the blood vessel extraction result of the G channel determined by the first extraction module 702 and the blood vessel extraction result of the R channel determined by the second extraction module 704 to obtain a first comprehensive extraction effect graph;
a vectorization processing module 706, configured to determine vectorization data of retinal blood vessels of the fundus image according to the first comprehensive extraction effect map obtained by the comprehensive processing module 705.
Optionally, the first extraction module 702 and the second extraction module 704 are specifically configured to perform equalization optimization on an image corresponding to channel image data by using an adaptive histogram to obtain third image data, where the channel image data is the first image data or the second image data; preprocessing the third image data based on a complex Morlet mother wavelet formula to obtain a thick blood vessel value and a thin blood vessel value, segmenting the blood vessel by means of Top-Hat variation and double-loop filtering, and further optimizing a tiny blood vessel part in the image by adopting multi-scale Gaussian matching filtering to obtain a first blood vessel extraction result; and aiming at the first blood vessel extraction result, performing blood vessel edge optimization by adopting an active growth edge detection algorithm, and performing growth reduction on the blood vessel with poor extraction effect to obtain a second blood vessel extraction result.
Optionally, the vectorization processing module 706 is specifically configured to, according to the first comprehensive extraction effect map, remove a thin blood vessel smaller than a set pixel by combining a blood vessel morphological structure; refining the edges of the blood vessels by adopting a Sobel algorithm to obtain the center lines of the blood vessels; based on the blood vessel structure, filtering the intersection points and the bifurcation points of the blood vessels on the central line of the blood vessels to obtain a second comprehensive extraction effect graph; and determining vectorization data of retinal blood vessels of the fundus image according to the second comprehensive extraction effect map.
Optionally, the vectorization processing module 706 is specifically configured to perform refinement processing on the edges of the blood vessel based on 8 direction templates by using a Sobel algorithm to obtain a blood vessel center line.
Optionally, the vectorization processing module 706 is specifically configured to locate a blood vessel edge based on the blood vessel center line, and record and store coordinates of the blood vessel edge and coordinates of the blood vessel center line, where the coordinates of the blood vessel edge and the coordinates of the blood vessel center line form vectorization data of a retinal blood vessel of the fundus image.
Those of skill would further appreciate that the various illustrative components and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative components and steps have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied in hardware, a software module executed by a processor, or a combination of the two. A software module may reside in Random Access Memory (RAM), memory, Read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
The above-mentioned embodiments are intended to illustrate the objects, technical solutions and advantages of the present invention in further detail, and it should be understood that the above-mentioned embodiments are merely exemplary embodiments of the present invention, and are not intended to limit the scope of the present invention, and any modifications, equivalent substitutions, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (8)
1. A method for automatic vectorization of retinal blood vessels in a fundus image, the method comprising:
performing data separation processing on a G channel image of a fundus image to obtain first image data;
determining a blood vessel extraction result of a G channel according to the first image data;
performing data separation processing on an R channel image of the fundus image to obtain second image data;
determining a blood vessel extraction result of an R channel according to the second image data;
weighting the blood vessel extraction result of the G channel and the blood vessel extraction result of the R channel to obtain a first comprehensive extraction effect graph;
determining the vectorization data of the retinal blood vessels of the fundus image according to the first comprehensive extraction effect image, wherein the vectorization data comprises the steps of removing the thin blood vessels smaller than set pixels by combining a blood vessel morphological structure according to the first comprehensive extraction effect image; refining the edges of the blood vessels by adopting a Sobel algorithm to obtain the center lines of the blood vessels; based on the blood vessel structure, filtering the intersection points and the bifurcation points of the blood vessels on the central line of the blood vessels to obtain a second comprehensive extraction effect graph; and determining vectorization data of retinal blood vessels of the fundus image according to the second comprehensive extraction effect map.
2. The method of claim 1, wherein said determining a vessel extraction result for a G channel from said first image data and said determining a vessel extraction result for an R channel from said second image data comprises:
carrying out equalization optimization on an image corresponding to channel image data by adopting a self-adaptive histogram to obtain third image data, wherein the channel image data is the first image data or the second image data;
preprocessing the third image data based on a complex Morlet mother wavelet formula to obtain a thick blood vessel value and a thin blood vessel value, segmenting the blood vessel by means of Top-Hat variation and double-loop filtering, and further optimizing a tiny blood vessel part in the image by adopting multi-scale Gaussian matching filtering to obtain a first blood vessel extraction result;
and aiming at the first blood vessel extraction result, performing blood vessel edge optimization by adopting an active growth edge detection algorithm, and performing growth reduction on the blood vessel with poor extraction effect to obtain a second blood vessel extraction result.
3. The method of claim 1, wherein the refining the vessel edge by using the Sobel algorithm to obtain the vessel centerline comprises:
and refining the edges of the blood vessels by using a Sobel algorithm based on 8 direction templates to obtain the center lines of the blood vessels.
4. The method according to claim 1, wherein the determining vectorized data of retinal blood vessels of the fundus image from the second integrated extraction effect map comprises:
and positioning the blood vessel edge based on the blood vessel central line, and recording and storing the coordinates of the blood vessel edge and the coordinates of the blood vessel central line, wherein the coordinates of the blood vessel edge and the coordinates of the blood vessel central line form vectorized data of the retinal blood vessel of the fundus image.
5. An apparatus for automatic vectorization of retinal blood vessels in a fundus image, the apparatus comprising:
the first separation module is used for carrying out data separation processing on a G channel image of a fundus image to obtain first image data;
the first extraction module is used for determining a blood vessel extraction result of a G channel according to the first image data obtained by the first separation module;
the second separation module is used for carrying out data separation processing on the R channel image of the fundus image to obtain second image data;
the second extraction module is used for determining a blood vessel extraction result of the R channel according to the second image data obtained by the second separation module;
the comprehensive processing module is used for weighting the blood vessel extraction result of the G channel determined by the first extraction module and the blood vessel extraction result of the R channel determined by the second extraction module to obtain a first comprehensive extraction effect graph;
the vectorization processing module is used for determining the vectorization data of the retinal blood vessels of the fundus image according to the first comprehensive extraction effect image obtained by the comprehensive processing module, and is specifically used for removing the thin blood vessels smaller than the set pixels by combining the morphological structure of the blood vessels according to the first comprehensive extraction effect image; refining the edges of the blood vessels by adopting a Sobel algorithm to obtain the center lines of the blood vessels; based on the blood vessel structure, filtering the intersection points and the bifurcation points of the blood vessels on the central line of the blood vessels to obtain a second comprehensive extraction effect graph; and determining vectorization data of retinal blood vessels of the fundus image according to the second comprehensive extraction effect map.
6. The apparatus according to claim 5, wherein the first extraction module and/or the second extraction module are specifically configured to perform equalization optimization on an image corresponding to channel image data using an adaptive histogram to obtain third image data, where the channel image data is the first image data or the second image data; preprocessing the third image data based on a complex Morlet mother wavelet formula to obtain a thick blood vessel value and a thin blood vessel value, segmenting the blood vessel by means of Top-Hat variation and double-loop filtering, and further optimizing a tiny blood vessel part in the image by adopting multi-scale Gaussian matching filtering to obtain a first blood vessel extraction result; and aiming at the first blood vessel extraction result, performing blood vessel edge optimization by adopting an active growth edge detection algorithm, and performing growth reduction on the blood vessel with poor extraction effect to obtain a second blood vessel extraction result.
7. The apparatus according to claim 5, wherein the vectorization processing module is specifically configured to perform refinement processing on the vessel edges based on 8 direction templates by using a Sobel algorithm to obtain a vessel centerline.
8. The apparatus according to claim 5, wherein the vectorization processing module is specifically configured to locate a blood vessel edge based on the blood vessel center line, and record and store coordinates of the blood vessel edge and coordinates of the blood vessel center line, where the coordinates of the blood vessel edge and the coordinates of the blood vessel center line form vectorization data of a retinal blood vessel of the fundus image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710397099.7A CN107292835B (en) | 2017-05-31 | 2017-05-31 | Method and device for automatically vectorizing retinal blood vessels of fundus image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710397099.7A CN107292835B (en) | 2017-05-31 | 2017-05-31 | Method and device for automatically vectorizing retinal blood vessels of fundus image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107292835A CN107292835A (en) | 2017-10-24 |
| CN107292835B true CN107292835B (en) | 2020-03-13 |
Family
ID=60095346
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710397099.7A Expired - Fee Related CN107292835B (en) | 2017-05-31 | 2017-05-31 | Method and device for automatically vectorizing retinal blood vessels of fundus image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107292835B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108399647B (en) * | 2018-03-05 | 2021-10-26 | 四川和生视界医药技术开发有限公司 | Editing method and editing device for retinal blood vessel edge line |
| CN110543802A (en) * | 2018-05-29 | 2019-12-06 | 北京大恒普信医疗技术有限公司 | Method and device for identifying left eye and right eye in fundus image |
| CN109166124B (en) * | 2018-11-20 | 2021-12-14 | 中南大学 | Retinal blood vessel morphology quantification method based on connected region |
| CN109919882B (en) * | 2019-01-18 | 2023-07-21 | 平安科技（深圳）有限公司 | Image optimization method based on fundus color photograph image and related equipment |
| CN109919915B (en) * | 2019-02-18 | 2021-03-23 | 广州视源电子科技股份有限公司 | Retina fundus image abnormal region detection method and device based on deep learning |
| CN109829942B (en) * | 2019-02-21 | 2023-04-28 | 韶关学院 | Automatic quantification method for retinal vessel diameter of fundus image |
| CN113643353B (en) * | 2020-09-04 | 2024-02-06 | 深圳硅基智能科技有限公司 | Measurement method for enhancing resolution of vascular caliber of fundus image |
| CN112734784B (en) * | 2021-01-28 | 2024-07-19 | 依未科技(北京)有限公司 | Method, device, medium and equipment for determining fundus blood vessel boundary with high precision |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103810492A (en) * | 2014-01-17 | 2014-05-21 | 北京大恒图像视觉有限公司 | Ambiguity analytical method of eye fundus image |
| CN103870838A (en) * | 2014-03-05 | 2014-06-18 | 南京航空航天大学 | Eye fundus image characteristics extraction method for diabetic retinopathy |
| CN106372593A (en) * | 2016-08-30 | 2017-02-01 | 上海交通大学 | Optic disc area position method based on blood vessel convergence |
-
2017
- 2017-05-31 CN CN201710397099.7A patent/CN107292835B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103810492A (en) * | 2014-01-17 | 2014-05-21 | 北京大恒图像视觉有限公司 | Ambiguity analytical method of eye fundus image |
| CN103870838A (en) * | 2014-03-05 | 2014-06-18 | 南京航空航天大学 | Eye fundus image characteristics extraction method for diabetic retinopathy |
| CN106372593A (en) * | 2016-08-30 | 2017-02-01 | 上海交通大学 | Optic disc area position method based on blood vessel convergence |
Non-Patent Citations (4)
| Title |
|---|
| Yin Y 等.Retinal vessel segmentation using a probabilistic tracking method.《Pattern Recognition》.2012,第45卷(第4期), * |
| Zhao TQ 等.Retinal vessels segmentation based on level set and region growing.《Pattern Recognition》.2014,第47卷(第7期), * |
| 周琳.眼底图像中血管分割技术研究.《中国优秀硕士学位论文全文数据库-信息科技辑》.2011,(第11期), * |
| 殷本俊 等.基于Morlet小波变换的视网膜血管分割.《计算机辅助设计与图形学学报》.2015,第27卷(第7期), * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107292835A (en) | 2017-10-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107292835B (en) | Method and device for automatically vectorizing retinal blood vessels of fundus image | |
| WO2021169128A1 (en) | Method and apparatus for recognizing and quantifying fundus retina vessel, and device and storage medium | |
| CN109190690B (en) | Method for detecting and identifying cerebral microhemorrhage points based on SWI image of machine learning | |
| CN110415216B (en) | CNV automatic detection method based on SD-OCT and OCTA retina images | |
| CN111815663A (en) | Hepatic vessel segmentation system based on Hessian matrix and gray scale method | |
| CN111882566A (en) | Blood vessel segmentation method, device, equipment and storage medium of retina image | |
| Khademi et al. | Whole volume brain extraction for multi-centre, multi-disease FLAIR MRI datasets | |
| CN114937024A (en) | Image evaluation method and device and computer equipment | |
| CN108665474B (en) | B-COSFIRE-based retinal vessel segmentation method for fundus image | |
| Uribe-Valencia et al. | Automated Optic Disc region location from fundus images: Using local multi-level thresholding, best channel selection, and an Intensity Profile Model | |
| US20130222767A1 (en) | Methods and systems for detecting peripapillary atrophy | |
| Kavitha et al. | Blood vessel, optical disk and damage area-based features for diabetic detection from retinal images | |
| CN110930346B (en) | Automatic detection method and storage device for eyeground image microangioma | |
| Tsai et al. | Adaptive segmentation of vessels from coronary angiograms using multi-scale filtering | |
| Desiani et al. | A robust techniques of enhancement and segmentation blood vessels in retinal image using deep learning | |
| CN108230306A (en) | Eyeground color picture blood vessel and arteriovenous recognition methods | |
| Liu et al. | Retinal vessel segmentation using densely connected convolution neural network with colorful fundus images | |
| Lazar et al. | A novel approach for the automatic detection of microaneurysms in retinal images | |
| JP5740403B2 (en) | System and method for detecting retinal abnormalities | |
| CN114862799B (en) | Full-automatic brain volume segmentation method for FLAIR-MRI sequence | |
| CN115841472A (en) | Method, device, equipment and storage medium for identifying high-density characteristics of middle cerebral artery | |
| Martins | Automatic microaneurysm detection and characterization through digital color fundus images | |
| CN112365489A (en) | Ultrasonic image blood vessel bifurcation detection method | |
| CN113191998B (en) | AIF curve extraction system, method, device and medium based on artery segmentation | |
| CN118505518B (en) | Ophthalmic auxiliary imaging method and system |
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 | ||
| CB02 | Change of applicant information |
Address after: 116000 room 19, block B, block 32A, Torch Road, Dalian high tech Industrial Park, Liaoning, China, 1907-1 Applicant after: Ruida Sheng medical technology (Dalian) Co., Ltd. Address before: 116024 creative Incubation Park, 720 Huangpu Road, hi tech park, Dalian, Liaoning 4F Applicant before: Redasen Technology (Dalian) Co, Ltd. |
|
| CB02 | Change of applicant information | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200313 Termination date: 20210531 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |