CN111291706A

CN111291706A - Retina image optic disc positioning method

Info

Publication number: CN111291706A
Application number: CN202010112482.5A
Authority: CN
Inventors: 陈维洋; 赵树
Original assignee: Qilu University of Technology
Current assignee: Qilu University of Technology
Priority date: 2020-02-24
Filing date: 2020-02-24
Publication date: 2020-06-16
Anticipated expiration: 2040-02-24
Also published as: CN111291706B

Abstract

The invention discloses a retina image optic disc positioning method, which comprises the following steps: s1, preprocessing the color retina image, including performing gray processing and mask processing on the retina image to obtain a retina gray image; s2, calculating the standard deviation of the texture features of the retina gray level image based on the retina gray level image obtained in the step S1, and calculating to obtain a binary image based on standard deviation feature segmentation processing; s3, performing morphological dilation operation on the segmentation image obtained in the step S2 to obtain a dilation image; s4, circle detection is carried out on the expansion image obtained in the step S3, and the positioning circle meeting the conditions is marked and stored; and S5, marking the mark and the stored positioning circle on the color image, wherein the center of the positioning circle is the positioned optical disk. The invention can accurately position the optic disk in the image according to the retina image processed by the texture characteristic standard deviation and the shape characteristic circle, and is beneficial to the research of medical retina images.

Description

Retina image optic disc positioning method

Technical Field

The invention relates to the technical field of retinal image processing, in particular to a retinal image optic disc positioning method.

Background

The fundus image is the only in vivo blood vessel image that can be displayed by direct viewing or by taking a picture, providing the clinician with a possible window to view retinal structures, including the optic disc (also known as the optic disc), the arteriovenous and the macula. Fundus image analysis can not only assist in the diagnosis and treatment of eye diseases such as glaucoma, cataracts, etc., but also assist clinicians in examining certain systemic vascular conditions such as diabetes, hypertension, etc. Therefore, ophthalmoscopy has become an important non-invasive examination in medical practice, and clinicians can screen for systemic diseases and diagnose retinal diseases by regular examination of retinal morphological changes.

The optic disc, is a clear reddish discoid structure with a diameter of about 1.5mm and a clear border from the macula to about 3mm of the nose, and is a part on the retina where nerve fibers and retinal blood vessels converge and penetrate out of the eyeball, so that the optic disc is the most dense area of the retinal blood vessel network, and the blood vessels extend from the optic disc to the whole retinal area. The optic disk is an important feature of fundus images, and morphological changes in size, shape, and color thereof are important indices for diagnosing various diseases.

In fundus image detection, disc positioning is one of the prerequisites for analyzing and diagnosing fundus images, and can be used for not only disc center position estimation and disc segmentation, but also for determining the positions of other retinal structures (such as macula), and even for parameter measurement.

The chinese patent application No. 2019105832331 discloses a retinal image blood vessel separation method and system based on gray variance and standard deviation, which provides a certain theoretical basis for optic disc positioning.

Disclosure of Invention

The invention aims to provide a retinal image optic disc positioning method based on texture features and shape features to ensure the accuracy of optic disc positioning.

The technical scheme adopted by the invention is as follows:

a retina image optic disc positioning method comprises the following steps:

s1, preprocessing the color retina image, including performing gray processing and mask processing on the retina image to obtain a retina gray image;

s2, calculating the standard deviation of the texture features of the retina gray level image based on the retina gray level image obtained in the step S1, and calculating to obtain a binary image based on standard deviation feature segmentation processing;

s3, performing morphological dilation operation on the segmentation image obtained in the step S2 to obtain a dilation image;

s4, circle detection is carried out on the expansion image obtained in the step S3, and the positioning circle meeting the conditions is marked and stored;

and S5, marking the mark and the stored positioning circle on the color image, wherein the center of the positioning circle is the positioned optical disk.

As a further optimization, in step S2 of the present invention, the acquiring step of the binary image includes:

s21, calculating a standard deviation characteristic value of each pixel on the retina gray level image through the gray level co-occurrence matrix, and obtaining a characteristic matrix based on the standard deviation of each pixel;

s22, sorting the standard deviations of all the pixels, setting a first threshold value according to the number of the pixels, and carrying out binarization segmentation on the feature matrix obtained in the step S21 to obtain the binary image in the step S2.

Specifically, the value of the first threshold is 2/3 of the non-zero-value pixel.

Specifically, the expression for feature matrix segmentation in the present invention is:

wherein: z is a sorting set of non-zero-value pixel points of the feature matrix, Y is a binary image, and K is a first threshold.

As a further optimization, in step S4 of the present invention, the step of performing circle detection on the image a4 includes:

s41, identifying the optic disc area according to the expansion image obtained in the step S3, and respectively recording the minimum value and the maximum value on the transverse coordinate and the minimum value and the maximum value on the longitudinal coordinate of the optic disc area, wherein the minimum value of the transverse coordinate of the optic disc area is m1, the maximum value of the transverse coordinate of the optic disc area is m2, the minimum value of the longitudinal coordinate of the optic disc area is n1, and the maximum value of the longitudinal coordinate of the optic disc area is n 2;

s42, defining a circle center and a radius, setting an initial value of the radius, setting a value range of an abscissa of the circle center as (m1, m2), and setting a value range of an ordinate of the circle center O as (n1, n 2);

s43, uniformly taking a set number of verification points on the circumference, executing the step S44 when all the verification points are located in the video area, otherwise, executing the step S45;

s44, recording the circle center position and the verification point to finish circle detection;

s45, the radius is decreased by 1, and step S43 is repeated.

Specifically, the initial value of the radius r is (50, 100), and the value of the verification point is (200, 360).

Specifically, the window size of the gray level co-occurrence matrix is 17x17, and the gray level is 8.

Specifically, the processing of the present invention for graying the color retina image includes the following steps:

the color retina image is grayed by a component method, a maximum value method, an average value method or a weighted average method.

Specifically, the mask is a two-dimensional matrix array or a multi-value image and is used for highlighting the region of interest and shielding a noise region.

The invention has the following advantages:

1. according to the method, the retinal image is segmented based on the standard deviation, blood vessels in the retinal area are effectively separated, impurities and noise are removed, after morphological dilation operation is carried out, circle detection is carried out on the segmented image through image shape characteristics, and the retinal area in the retinal image is positioned;

2. the circle detection of the invention is carried out by giving an initial value of radius, selecting a certain number of verification points on the circumference, judging whether the positioning circle is qualified by judging whether the detection points are all positioned in the optic disc area, and accurately finding out the optic disc area on the retina image;

3. according to the circle detection method, the traversal range of the circle detection can be reduced, the system calculation times are reduced, and the positioning speed is improved by positioning the coordinates of the optic disc area;

4. the value range of the initial value of the radius and the value range of the number of the verification points ensure accurate positioning of the optic disc, reduce the calculation burden of a system and improve the positioning speed.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

The invention is further described below with reference to the accompanying drawings:

FIG. 1 is a color retinal image embodiment;

FIG. 2 is a grayscale image of the color retinal image of FIG. 1;

FIG. 3 is another retinal image after standard deviation segmentation;

FIG. 4 is the image of FIG. 3 after morphological dilation;

FIG. 5 is a resulting image of the circle detection performed on FIG. 4;

FIG. 6 is an image of the circle of FIG. 5 labeled on the original color retinal image;

FIG. 7 is the image of FIG. 4 as a mask to extract the optic disc area on the grayscale image;

FIG. 8 is an image of another retinal image after standard deviation segmentation;

FIG. 9 is the image of FIG. 8 after morphological dilation and circle detection;

FIG. 10 is an image of the circle of FIG. 9 labeled on the original color retinal image;

FIG. 11 is an image of the disk region extracted on the gray scale image as a mask after morphological dilation of FIG. 8;

FIG. 12 is an image of the results of further circle detection on FIG. 4;

FIG. 13 is an image of the circle detection of FIG. 12 as noted on the original color retinal image;

fig. 14 is a mask used in step S1.

Detailed Description

The present invention is further described in the following with reference to the drawings and the specific embodiments so that those skilled in the art can better understand the present invention and can implement the present invention, but the embodiments are not to be construed as limiting the present invention, and the embodiments and the technical features of the embodiments can be combined with each other without conflict.

It is to be understood that the terms first, second, and the like in the description of the embodiments of the invention are used for distinguishing between the descriptions and not necessarily for describing a sequential or chronological order. The "plurality" in the embodiment of the present invention means two or more.

The term "and/or" in the embodiment of the present invention is only an association relationship describing an associated object, and indicates that three relationships may exist, for example, a and/or B may indicate: a exists alone, B exists alone, and A and B exist at the same time. In addition, the character "/" herein generally indicates that the former and latter associated objects are in an "or" relationship.

The embodiment provides a retinal image optic disc positioning method, which is applied to the field of medical assistance and used for accurately positioning the position of a optic disc in a retinal image, and comprises the following steps:

s1, preprocessing the color retinal image shown in fig. 1, including performing graying processing and masking processing on the retinal image, to obtain a retinal grayscale image, as shown in fig. 2;

for the color retinal image acquisition, in the development stage of the embodiment, the color retinal image acquisition is performed in the experimental stage for processing and verification in the experimental stage, and is acquired from the online DRIVE database, the website address of the online DRIVE database is https:// www.isi.uu.nl/Research/data bases/DRIVE/, the DREVE database is from the diabetic retinopathy screening project of the Netherlands, the online driving diabetic retinopathy screening project is acquired by a Canon CR5 non-mydriatic 45-degree visual Field (FOV) 3CCD camera, the image size is 565 x 584 pixels, and the color retinal image conversion is performed between the color retinal image during the non-circle marking detection of the figures 1 and 6 and the color retinal image during the non-circle marking detection of the figure 10. In the working stage of the present embodiment, the color retinal image is the retinal image of the patient actually taken during the examination of the patient.

Each color image is formed by combining RGB three-color channels, that is, various colors are formed based on three primary colors (red, green and blue), when the R value (G value) and the B value (G value) represent a gray color, and the R value (G value) and the B value (B value) represent gray values, the method for graying a color image generally includes a component method, a maximum value method, an average value method and a weighted average method, and the purpose is to equalize the R value, the G value and the B value of a pixel point.

The mask is to cover a filter membrane on the original image, the filter membrane can not only extract the region of interest, but also shield some image regions to reduce parameter operation, in the optical image processing, the mask is generally a film or a filter, and in the digital image processing, the mask is a two-dimensional matrix array, which can be a multi-value image. In the embodiment, the used mask is as shown in fig. 14, and is used for extracting the region of interest and reducing the number of image pixels, so that the program speed is simpler and faster, and the retina image in the DRIVE database carries the mask by itself.

S2, calculating the standard deviation of the texture characteristics based on the retina gray scale image obtained in the step S1,

wherein: the calculation formula of the standard deviation of the retina gray level image is as follows:

wherein: g is the gray value of the gray image at the pixel point p (i, j), Mean is the Mean value, and the calculation formula of Mean value is as follows:

the calculation for obtaining the binary image based on the standard deviation feature segmentation processing comprises the following steps: (ii) a

S21, calculating a standard deviation eigenvalue of each pixel on the retina gray image through a gray level co-occurrence matrix to obtain a characteristic matrix based on the standard deviation eigenvalue of each pixel, where the gray level co-occurrence matrix (GLCM) is a common method for describing texture features by studying spatial correlation characteristics of gray levels. The gray level co-occurrence matrix adopts a gray level co-occurrence matrix with the window size of 17x17 and the gray level of 8 levels;

s22, sorting the standard deviations of all pixels, and carrying out binarization segmentation on the feature matrix obtained in the step S21 by taking 2/3 of non-zero-value pixel points as a first threshold value to obtain the binary image in the step S2; the expression for the feature matrix segmentation is:

wherein: z is a sorting set of non-zero-value pixel points of the feature matrix, Y is a binary image, and K is a first threshold. Binarization, i.e. the gray value of the image is only two levels, and visually appears to have only a black and white effect, represented by 0 and 255, or represented by 0 and 1. In the binarization segmentation process, the number of image pixels is 584 × 565, zero-valued pixels after mask processing are removed, and after all the remaining pixels are sorted according to the standard deviation, the front 2/3 of the sequence is assigned as 0 and visually appears as a black area, and the rear 1/3 of the total pixels is assigned as 1 and visually appears as a white area, as shown in fig. 3 and 8.

S3, performing morphological dilation operation on the segmented image obtained in step S2 to obtain a dilated image, which may also be used as a mask of the optic disc region as shown in the figure, and may be used in other retinal image processing as shown in fig. 7 and 11;

as shown in fig. 3, 8, and 12, the images after the standard deviation feature segmentation have a display region of the blood vessels near the optic disc, and many black regions exist between the white regions, and the presence of these black regions affects the determination of the result when selecting a detection point in the subsequent circle detection. Morphological dilation is similar to the scrolling operation, assuming that there is an image a and a structural element B, B moves over a, where B defines that its center is the anchor point, and the maximum pixel value of a under B coverage is calculated to replace the pixel of the anchor point, where B as a structural element can be any shape.

S4, performing circle detection on the dilated image obtained in step S3, and marking and saving the qualified positioning circle, as shown in fig. 5 and 9; the circle detection step on the re-expansion image comprises the following steps:

s41, as shown in fig. 4, identifying the optic disc region according to the dilated image obtained in step S3, and recording the minimum value and the maximum value on the lateral coordinate and the minimum value and the maximum value on the longitudinal coordinate of the optic disc region, respectively, defining the minimum value of the lateral coordinate of the optic disc region as m1, the maximum value of the lateral coordinate as m2, the minimum value of the longitudinal coordinate of the optic disc region as n1, and the maximum value of the longitudinal coordinate as n 2;

s42, defining a circle center and a radius, setting an initial value of the radius, wherein the initial value of the radius is (50, 100), the range of the abscissa of the circle center is (m1, m2), and the range of the ordinate of the circle center O is (n1, n 2);

s43, uniformly taking a set number of verification points on the circumference, wherein the value range of the verification points is (200, 360), when the number of the selected verification points is large, the operation speed is influenced, when the number of the selected verification points is small, the accuracy of the positioning of the video disc is influenced, the number of the selected verification points is 360, when all the verification points are positioned in the video disc area, the step S44 is executed, and in the verification process, the rounding operation is carried out when most of the coordinates of the verification points are decimal due to insufficient size of the image. Otherwise, go to step S45;

s45, the radius is decreased by 1, and step S43 is repeated.

S5, mark the mark and the saved positioning circle on the color image, the center of the positioning circle is the positioned optic disc, as shown in fig. 6 and 10.

The image features mainly include color features, texture features, shape features and spatial relationship features of the image. The shape features are represented in two types, one is outline features, the other is region features, the outline features of the image mainly aim at the outer boundary of the object, and the region features of the image are related to the whole shape region. Therefore, the shape of a circle is used to represent the shape feature in the retinal image in the present embodiment.

As shown in fig. 5 and 12, as the processing of the same graph, the initial value of the radius of the circle detection in fig. 5 is selected to be 50, but as shown in fig. 6, the center of the circle is close to the edge of the optical disc, as shown in fig. 13, the center of the circle is close to the center of the optical disc, although the circle is located on the optical disc, the result of locating the center of the optical disc is more accurate, and other detection data are found, the radius is selected to be 100 during the circle detection, and the result is closer to the center of the optical disc, therefore, the more refined the procedure is, that is, the radius is selected to be a larger initial value, the more verification points are selected, and the more the center of the circle is closer to the center of the optical disc, the more accurate. However, the larger the selection of the initial radius value, the larger the number of radius traversals, the larger the system calculation amount, the larger the number of verification points to be selected, and the same the system calculation load, so the initial radius value is selected to be 100 because the size of the optic disc area is about 1/6 of the image size, the image size is close to 600X600 because the initial value of the preferred selection is 100, and the number of verification points is selected to be 360.

Through tests, the optic disc positioning accuracy of the optic disc positioning method in the embodiment in all retina pictures in the Drive database reaches 100%.

The above-mentioned embodiments are merely preferred embodiments for fully illustrating the present invention, and the scope of the present invention is not limited thereto. The equivalent substitution or change made by the technical personnel in the technical field on the basis of the invention is all within the protection scope of the invention. The protection scope of the invention is subject to the claims.

Claims

1. A retina image optic disc positioning method is characterized in that: the method comprises the following steps:

2. The method of claim 1, wherein: in step S2, the acquiring step of the binary image includes:

3. The method of claim 2, wherein: the value of the first threshold is 2/3 of non-zero value pixel points.

4. The method of claim 2, wherein: the expression for the feature matrix segmentation is:

5. The method of claim 1, wherein: in step S4, the step of performing circle detection on the image a4 includes:

s45, the radius is decreased by 1, and step S43 is repeated.

6. The method of claim 1, wherein: the initial value of the radius r is (50, 100), and the verification point takes a value of (200, 360).

7. The method of claim 1, wherein: the window size of the gray level co-occurrence matrix is 17x17, and the gray level is 8.

8. The method of claim 1, wherein: the color retina image is grayed by a component method, a maximum value method, an average value method or a weighted average method.

9. The method of claim 1, wherein: the mask is a two-dimensional matrix array or a multi-value image and is used for highlighting the region of interest and shielding the noise region.