CN109035262A - A kind of multimode eye imaging specular light removing method - Google Patents
A kind of multimode eye imaging specular light removing method Download PDFInfo
- Publication number
- CN109035262A CN109035262A CN201810905552.5A CN201810905552A CN109035262A CN 109035262 A CN109035262 A CN 109035262A CN 201810905552 A CN201810905552 A CN 201810905552A CN 109035262 A CN109035262 A CN 109035262A
- Authority
- CN
- China
- Prior art keywords
- image
- bianry
- zero
- reference picture
- processed
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—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
- G06T7/155—Segmentation; Edge detection involving morphological operators
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30041—Eye; Retina; Ophthalmic
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Eye Examination Apparatus (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of multimode eyes, and specular light removing method is imaged, including obtains a reference picture bg, and reference picture bg is to be shot in darkroom or simulation dark room conditions without target image；Reference picture bg is converted into gray level image gr；By gray level image gr binaryzation, bianry image bn is obtained；It whether there is the pixel of non-zero in detection bianry image bn: if there are the pixels of non-zero in bianry image bn, enabling corrosion frequency n=0；If the pixel of non-zero is not present in bianry image bn, the eye fundus image of shooting is not processed, is terminated；Morphological erosion operation is carried out to bianry image bn, obtains bianry image er；Image src to be processed subtracts reference picture bg and weakens the product of coefficient k to get the image arrived after weakening hot spot.The light spot image that the present invention can be formed mirror-reflection is weakened or is reduced, and enhancing doctor carries out pathological analysis and diagnosis to eye fundus image.
Description
Technical field
The present invention relates to a kind of fundus imaging hot spot technology for eliminating fields, anti-more particularly to a kind of multimode eye imaging mirror surface
Penetrate light removing method.
Background technique
Traditional fundus camera uses white light or green light to the imaging of retina, and the wavelength of imaging is single, and camera is simultaneously
The information for receiving all spectrum in wide spectrum leads to the specific wavelength that some specific tissues or lesion locations to eyeground are sensitive
Information be submerged and can not embody, be difficult to carry out different levels, protrusion not similar shapes for the biological tissue of eyeground complexity
The fundus imaging of state and pathological characters.
The imaging of multimode eye function is different using penetration capacity of the different wave length spectrum to tissue and tissue is to different spectrum
Absorbability it is different, to obtain the image of eyeground different tissues level and institutional framework.Utilize the imaging point of multimode eye function
Analysis system can obtain a series of spectrum picture on the eyeground of the light irradiation lower tester of different wave lengths, to deserved in 300ms
Morphological feature or pathological characters that see the different levels on eyeground, that reflection emphasis is different.Meanwhile it can be obtained in multispectral imaging
Oxygen-containing hemoglobin and deoxyhemoglobin are obtained in the image of the non-equal absorption bands of specific equal absorption bands and narrowband, to this
The double-wavelength images functional information of available retinal vessel containing oxygen condition after being analyzed.It is to a large amount of ophthalmology disease
Early diagnosis, clinical diagnosis and treatment can provide important references.
However, illumination light may form mirror on camera lens in the imaging process of multimode eye function Image analysis system
Face reflection, and then hot spot is formed in the picture, as shown in Figure 1, this meeting severe jamming doctor carries out pathological analysis to eyeground and examines
It is disconnected.
Summary of the invention
The present invention is directed at least solve the technical problems existing in the prior art, a kind of multimode eye is especially innovatively proposed
Specular light removing method is imaged.
In order to realize above-mentioned purpose of the invention, the invention discloses a kind of multimode eyes, and specular light elimination side is imaged
Method, comprising the following steps:
S1, obtains a reference picture bg, and reference picture bg is to be shot in darkroom or simulation dark room conditions without mesh
Logo image；
Reference picture bg is converted to gray level image gr by S2；
Gray level image gr binaryzation is obtained bianry image bn by S3；
S4 detects the pixel that whether there is non-zero in bianry image bn:
If there are the pixels of non-zero in bianry image bn, corrosion frequency n=0 is enabled, executes step S5；
If the pixel of non-zero is not present in bianry image bn, the eye fundus image of shooting is not processed, is terminated；
S5 carries out a morphological erosion operation to bianry image bn, obtains bianry image er；
S6 detects the pixel that whether there is non-zero in bianry image er:
If there are the pixel of non-zero, n=n+1 in bianry image er, and enable bn=er, return step S5,
If the pixel of non-zero is not present in bianry image er, S7 is thened follow the steps；
S7 carries out the operation of n times morphological dilations to bianry image bn, obtains image dl；
S8 obtains the minimum circumscribed rectangle of non-zero pixels in image dl, obtains rectangular area roi；
S9, the image for extracting the region roi in image dl is image mask, and the image for extracting the region roi in image bg is figure
As bgr；
Image mask and image bgr are done and operation, obtain image bmi, calculate all non-zero pixels in image bmi by S10
Mean value vbi；
Image mask and image bgr are done NAND operation, obtain image bmo, and calculate all non-zeros in image bmo by S11
The mean value vbo of pixel；
S12 extracts the figure in the region roi in image src to be processed using the eye fundus image of shooting as image src to be processed
As being image srcr；
Image mask and image srcr are done and operation, obtain image smi, calculate all non-zero pictures in image smi by S13
The mean value vsi of element；
Image mask and image srcr are done NAND operation, obtain image smo, calculate all non-zeros in image smo by S14
The mean value vso of pixel；
S15 is calculated and is weakened coefficient
S16, image src to be processed subtract reference picture bg and weaken the product of coefficient k to get the figure arrived after weakening hot spot
Picture.
In the preferred embodiment of the present invention, step S2 are as follows:
Centered on the geometric center of reference picture bg, the 1/m of reference picture bg wide is width, and the m is greater than or equal to 1
Positive number, the 1/n of reference picture bg high are the image of high rectangular area as image bgc；The n is greater than or equal to 1 positive number,
And it is not 1 that m, n are simultaneously；Light leakage is removed, the aperture for preventing light leakage from generating is interfered generation is calculated, and can also reduce calculating
Amount；Image bgc is converted into gray level image gr；
Step S9 are as follows:
The image for extracting the region roi in image dl is image mask, and the image for extracting the region roi in image bgc is image
bgr；
Step S12 are as follows: using the eye fundus image of shooting as image src to be processed, with the geometric center of image src to be processed
Centered on, the 1/m of image src wide to be processed is width, and the 1/n of image src high to be processed is the image conduct of high rectangular area
Image srcc, the image for extracting the region roi in image srcc is image srcr.
In the preferred embodiment of the present invention, m=n=2.
In the preferred embodiment of the present invention, in step s 2, using OTSU algorithm by gray level image gr two-value
Change.
In the preferred embodiment of the present invention, step S8 are as follows: to the minimum of non-zero pixels in the image dl of acquisition
Boundary rectangle expands x times, and the x is positive number, obtains rectangular area roi.
In the preferred embodiment of the present invention, x=0.5.
In the preferred embodiment of the present invention, the reference picture bg in step S1 and the eye fundus image in step S4
For the image of same camera shooting.
In conclusion by adopting the above-described technical solution, the beneficial effects of the present invention are: the present invention can be anti-to mirror surface
It penetrates the light spot image to be formed to be weakened or reduced, enhancing doctor carries out pathological analysis and diagnosis to eye fundus image.
Detailed description of the invention
Fig. 1 is that the eyeground that the present invention is shot has light spot image schematic diagram.
Fig. 2 is schematic process flow diagram of the invention.
Fig. 3 is reference picture bg schematic diagram in the embodiment of the present invention one.
Fig. 4 is the central area location drawing picture schematic diagram chosen in the embodiment of the present invention one.
Fig. 5 is image bgc schematic diagram in the embodiment of the present invention one.
Fig. 6 is gray level image gr schematic diagram in the embodiment of the present invention one.
Fig. 7 is bianry image bn schematic diagram in the embodiment of the present invention one.
Fig. 8 is image er schematic diagram in the embodiment of the present invention one.
Fig. 9 is image er1 schematic diagram in the embodiment of the present invention one.
Figure 10 is image er2 schematic diagram in the embodiment of the present invention one.
Figure 11 is image er3 schematic diagram in the embodiment of the present invention one.
Figure 12 is image dl schematic diagram in the embodiment of the present invention one.
Figure 13 is rio area schematic in the embodiment of the present invention one.
Figure 14 is image mask schematic diagram in the embodiment of the present invention one.
Figure 15 is image bgr schematic diagram in the embodiment of the present invention one.
Figure 16 is image bmi schematic diagram in the embodiment of the present invention one.
Figure 17 is image bmo schematic diagram in the embodiment of the present invention one.
Figure 18 is image src schematic diagram to be processed in the embodiment of the present invention one.
Figure 19 is image srcc schematic diagram in the embodiment of the present invention one.
Figure 20 is image srcr schematic diagram in the embodiment of the present invention one.
Figure 21 is image smi schematic diagram in the embodiment of the present invention one.
Figure 22 is image smo schematic diagram in the embodiment of the present invention one.
Figure 23 is to weaken the image schematic diagram after hot spot in the embodiment of the present invention one.
Figure 24 is two reference picture bg schematic diagram of the embodiment of the present invention.
Figure 25 is gray level image gr schematic diagram in the embodiment of the present invention two.
Figure 26 is bianry image bn schematic diagram in the embodiment of the present invention two.
Figure 27 is image er schematic diagram in the embodiment of the present invention two.
Figure 28 is image er1 schematic diagram in the embodiment of the present invention two.
Figure 29 is image er2 schematic diagram in the embodiment of the present invention two.
Figure 30 is image er3 schematic diagram in the embodiment of the present invention two.
Figure 31 is image dl schematic diagram in the embodiment of the present invention two.
Figure 32 is rio area schematic in the embodiment of the present invention two.
Figure 33 is image mask schematic diagram in the embodiment of the present invention two.
Figure 34 is image bgr schematic diagram in the embodiment of the present invention two.
Figure 35 is image bmi schematic diagram in the embodiment of the present invention two.
Figure 36 is image bmo schematic diagram in the embodiment of the present invention two.
Figure 37 is image src selection rio area schematic to be processed in the embodiment of the present invention two.
Figure 38 is image srcr schematic diagram in the embodiment of the present invention two.
Figure 39 is image smi schematic diagram in the embodiment of the present invention two.
Figure 40 is image smo schematic diagram in the embodiment of the present invention two.
Figure 41 is that the embodiment of the present invention two weakens the image schematic diagram after hot spot.
In Figure 42 (a)~(m) be 12 kinds of wavelength of the invention shoot the eye fundus image come it is treated by the present method after show
It is intended to.
Specific embodiment
The embodiment of the present invention is described below in detail, examples of the embodiments are shown in the accompanying drawings, wherein from beginning to end
Same or similar label indicates same or similar element or element with the same or similar functions.Below with reference to attached
The embodiment of figure description is exemplary, and for explaining only the invention, and is not considered as limiting the invention.
The invention discloses a kind of multimode eyes, and specular light removing method is imaged, as shown in Figure 2, comprising the following steps:
The first step, obtains a reference picture bg, and reference picture bg is shot in darkroom or simulation dark room conditions
Without target image；Execute second step.
Reference picture bg is converted to gray level image gr by second step；Execute third step.
Gray level image gr binaryzation is obtained bianry image bn by third step；Execute the 4th step.In the present embodiment, sharp
With OTSU algorithm by gray level image gr binaryzation.
4th step detects the pixel that whether there is non-zero in bianry image bn:
If there are the pixels of non-zero in bianry image bn, corrosion frequency n=0 is enabled, executes the 5th step；
It is without any processing to the eye fundus image of shooting if the pixel of non-zero is not present in bianry image bn, terminate.
5th step carries out a morphological erosion operation to bianry image bn, obtains bianry image er；Execute the 6th step.
6th step detects the pixel that whether there is non-zero in bianry image er:
If there are the pixel of non-zero, n=n+1 in bianry image er, and enable bn=er, the 5th step is returned；
If the pixel of non-zero is not present in bianry image er, the 7th step is executed.；
7th step carries out the operation of n times morphological dilations to bianry image bn, obtains image dl；Execute the 8th step.
8th step obtains the minimum circumscribed rectangle of non-zero pixels in image dl, obtains rectangular area roi, extract the region
Interior image is image mask；Execute the 9th step.
9th step, extracting the image in image bg in the region roi is image bgr；Execute the tenth step.In present embodiment
In, the region roi in image bg, be in image bg as image mask size, and with image mask in image dl position
Same region.
Image mask and image bgr are done and operation, obtain image bmi, calculate all non-zeros in image bmi by the tenth step
The mean value vbi of pixel；Execute the 11st step.
Image mask and image bgr are done NAND operation, obtain image bmo, and calculate institute in image bmo by the 11st step
There is the mean value vbo of non-zero pixels；Execute the 12nd step.
12nd step extracts the region roi in image src to be processed using the eye fundus image of shooting as image src to be processed
Interior image is image srcr；Execute the 13rd step.In the present embodiment, the region roi in image src to be processed, is figure
As in src as image mask size, and with image mask in image dl the same region in position.
Image mask and image srcr are done and operation, obtain image smi, calculate all non-in image smi by the 13rd step
The mean value vsi of zero pixel；Execute the 14th step.
Image mask and image srcr are done NAND operation, obtain image smo, calculated in image smo and own by the 14th step
The mean value vso of non-zero pixels；Execute the 15th step.
15th step calculates and weakens coefficientExecute the 16th step.
16 step of younger brother, image src to be processed subtract reference picture bg and weaken the product of coefficient k to get weakening hot spot is arrived
Image afterwards terminates.In the present embodiment, first product, then subtract.
In the present embodiment, specifically the first step~the 7th step (S1~S7) is illustrated:
Illustrate 1:
S1, obtains a reference picture bg, and reference picture bg is to be shot in darkroom or simulation dark room conditions without mesh
Logo image；Execute step S2.
Reference picture bg is converted to gray level image gr by S2；Execute step S3.
Gray level image gr binaryzation is obtained bianry image bn by S3；Execute step S4.
There are the pixels of non-zero in S4, bianry image bn, then enable corrosion frequency n=0, execute step S5.
S5 carries out a morphological erosion operation to bianry image bn, obtains bianry image er；Execute step S6.
There are the pixels of non-zero in S6, bianry image er, then n=1, executes step S61.
S61 carries out a morphological erosion operation to bianry image er, obtains bianry image er1；Execute step S62.
There are the pixels of non-zero in S62, bianry image er1, then n=2, executes step S62.
S63 carries out a morphological erosion operation to bianry image er1, obtains bianry image er2；Execute step S63.
There are the pixels of non-zero in S64, bianry image er2, then n=3, executes step S65.
S65 carries out a morphological erosion operation to bianry image er2, obtains bianry image er3；Execute step S66.
There are the pixels of non-zero in S66, bianry image er3, then n=4, executes step S67.
S67 carries out a morphological erosion operation to bianry image er3, obtains bianry image er4；Execute step S68.
There are the pixels of non-zero in S68, bianry image er4, then n=5, executes step S69.
S69 carries out a morphological erosion operation to bianry image er5, obtains bianry image er5；It thens follow the steps
S70。
The pixel of non-zero is not present in S70, bianry image er5, thens follow the steps S7.
S7 carries out 5 morphological dilations operations to bianry image er4, obtains image dl.
Illustrate 2:
S1, obtains a reference picture bg, and reference picture bg is to be shot in darkroom or simulation dark room conditions without mesh
Logo image；Execute step S2.
Reference picture bg is converted to gray level image gr by S2；Execute step S3.
Gray level image gr binaryzation is obtained bianry image bn by S3；Execute step S4.
There are the pixels of non-zero in S4, bianry image bn, then enable corrosion frequency n=0, execute step S5.
S5 carries out a morphological erosion operation to bianry image bn, obtains bianry image er；Execute step S6.
There are the pixels of non-zero in S6, bianry image er, then n=1, executes step S61.
S61 carries out a morphological erosion operation to bianry image er, obtains bianry image er1；Execute step S62.
There are the pixels of non-zero in S62, bianry image er1, then n=2, executes step S62.
S63 carries out a morphological erosion operation to bianry image er1, obtains bianry image er2；Execute step S63.
There are the pixels of non-zero in S64, bianry image er2, then n=3, executes step S65.
S65 carries out a morphological erosion operation to bianry image er2, obtains bianry image er3；Execute step S66.
There are the pixels of non-zero in S66, bianry image er3, then n=4, executes step S67.
S67 carries out a morphological erosion operation to bianry image er3, obtains bianry image er4；Execute step S68.
The pixel of non-zero is not present in S68, bianry image er4, thens follow the steps S7.
S7 carries out 4 morphological dilations operations to bianry image er3, obtains image dl.
Illustrate 3:
S1, obtains a reference picture bg, and reference picture bg is to be shot in darkroom or simulation dark room conditions without mesh
Logo image；Execute step S2.
Reference picture bg is converted to gray level image gr by S2；Execute step S3.
Gray level image gr binaryzation is obtained bianry image bn by S3；Execute step S4.
There are the pixels of non-zero in S4, bianry image bn, then enable corrosion frequency n=0, execute step S5.
S5 carries out a morphological erosion operation to bianry image bn, obtains bianry image er；Execute step S6.
The pixel of non-zero is not present in S6, bianry image er, thens follow the steps S7；
S7 carries out 0 morphological dilations operation to bianry image bn, obtains image dl.In the step s 7 to bianry image
Bn carries out 0 morphological dilations operation, is to bianry image bn without morphological dilations operation, image bn is exactly image
dl。
In the preferred embodiment of the present invention, second step are as follows: centered on the geometric center of reference picture bg, ginseng
The 1/m of image bg wide is examined as width, the m is greater than or equal to 1 positive number, and the 1/n of reference picture bg high is high rectangular area
Image is as image bgc；The n is greater than or equal to 1 positive number, and it is not 1 that m, n are simultaneously；Image bgc is converted into grayscale image
As gr, third step is executed.In the present embodiment, m and n all can be but be not limited to 2, can adjust figure according to the actual situation
As the size of bgc, guarantee aperture not within the scope of image bgc.
9th step are as follows: the image for extracting the region roi in image bgc is image bgr, executes the tenth step.In present embodiment
In, the region roi in image bgc, be in image bgc as image mask size, and with image mask in image dl position
Set same region.
12nd step are as follows: using the eye fundus image of shooting as image src to be processed, in the geometry of image src to be processed
Centered on the heart, the 1/m of image src wide to be processed is width, and the 1/n of image src high to be processed is that the image of high rectangular area is made
For image srcc, the image for extracting the region roi in image srcc is image srcr, executes the 13rd step.In the present embodiment,
The region roi in image srcc to be processed is in image srcc as image mask size, and with image mask in image dl
The same region in middle position.
In the preferred embodiment of the present invention, the 8th step are as follows: to the minimum of non-zero pixels in the image dl of acquisition
Boundary rectangle expands x times, and the x is positive number, obtains rectangular area roi.In the present embodiment, expand multiple x can be but
It is not limited to 0.5, can be configured according to the actual situation.
In the preferred embodiment of the present invention, the eye fundus image in the reference picture bg and the 4th step of first step
For the image of same camera shooting.
The following are the examples one of this method:
1, reference picture bg is obtained, reference picture bg is in darkroom or to simulate one shot in darkroom without target figure
Picture；As shown in Figure 3.
2, centered on the geometric center of reference picture bg, the 1/2 of reference picture bg wide is wide, and the 1/ of reference picture bg high
2 be the image of high rectangular area as image bgc；As shown in Figure 4 and Figure 5.
3, image bgc is converted into gray level image gr；As shown in Figure 6.
4, using OTSU algorithm by gray level image gr binaryzation, obtained bianry image bn；As shown in Figure 7.
5, since there are the pixels of non-zero in bianry image bn, corrosion frequency n=0 is enabled.
6, a morphological erosion operation is carried out to bianry image bn, obtains bianry image er；As shown in Figure 8.
7, since there are the pixels of non-zero in bianry image er, corrode frequency n=1.
8, a morphological erosion operation is carried out to bianry image er, obtains bianry image er1；As shown in Figure 9.
9, since there are the pixels of non-zero in bianry image er1, corrode frequency n=2.
10, a morphological erosion operation is carried out to bianry image er1, obtains bianry image er2；As shown in Figure 10.
11, since there are the pixels of non-zero in bianry image er2, corrode frequency n=3.
12, a morphological erosion operation is carried out to bianry image er2, obtains bianry image er3；As shown in figure 11.
13, due to the pixel all 0 in bianry image er3, it is swollen that 3 morphology are carried out to bianry image er2
Swollen operation obtains image dl；As shown in figure 12.
14, the minimum circumscribed rectangle of non-zero pixels in image dl is expanded 0.5 times, obtains rectangular area roi；Such as Figure 13 institute
Show.
15, the image for extracting the region roi in image dl is image mask, as shown in figure 14；Extract image bg in image
The position dl and the identical image of size are image bgr；As shown in figure 15.
16, by image mask and image bgr is done and operation, image bmi is obtained, as shown in figure 16；And it calculates in image bmi
The mean value vbi=89.72 of all non-zero pixels.
17, image mask and image bgr are done into NAND operation, obtains image bmo, as shown in figure 17；And calculate image bmo
In all non-zero pixels mean value vbo=19.89；
18, using the eye fundus image of shooting as image src to be processed, as shown in figure 18, with the geometry of image src to be processed
Centered on center, the 1/2 of image src wide to be processed is width, and the 1/2 of image src high to be processed is the image of high rectangular area
As image srcc, as shown in figure 19, then extracting image identical with the position image dl and size in image srcc is image
srcr；As shown in figure 20.
19, by image mask and image srcr is done and operation, image smi is obtained, as shown in figure 21；And calculate image smi
In all non-zero pixels mean value vsi=199.66.
20, image mask and image srcr are done into NAND operation, obtains image smo, as shown in figure 22；And calculate image
The mean value vso=145.67 of all non-zero pixels in smo.
21, it calculates and weakens coefficient
22, image src to be processed subtracts the product of bg and k to get the image arrived after weakening hot spot, as shown in figure 23.
The following are the examples two of this method:
1, reference picture bg is obtained, reference picture bg is in darkroom or to simulate one shot in darkroom without target figure
Picture；As shown in figure 24.
2, image bg is converted into gray level image gr；As shown in figure 25.
3, using OTSU algorithm by gray level image gr binaryzation, obtained bianry image bn；As shown in figure 26.
4, since there are the pixels of non-zero in bianry image bn, corrosion frequency n=0 is enabled.
5, a morphological erosion operation is carried out to bianry image bn, obtains bianry image er；As shown in figure 27.
6, since there are the pixels of non-zero in bianry image er, corrode frequency n=1.
7, a morphological erosion operation is carried out to bianry image er, obtains bianry image er1；As shown in figure 28.
8, since there are the pixels of non-zero in bianry image er1, corrode frequency n=2.
9, a morphological erosion operation is carried out to bianry image er1, obtains bianry image er2；As shown in figure 29.
10, since there are the pixels of non-zero in bianry image er2, corrode frequency n=3.
11, a morphological erosion operation is carried out to bianry image er2, obtains bianry image er3；As shown in figure 30.
12, due to the pixel all 0 in bianry image er3, it is swollen that 3 morphology are carried out to bianry image er2
Swollen operation obtains image dl；As shown in figure 31.
13, the minimum circumscribed rectangle of non-zero pixels in image dl is expanded 0.5 times, rectangular area roi is obtained, such as Figure 32 institute
Show；Extracting the image in the region is image mask, as shown in figure 33.
14, it extracts in image bg as image mask size, and position is same in image dl with image mask
Image in region is image bgr；As shown in figure 34.
15, by image mask and image bgr is done and operation, image bmi is obtained, as shown in figure 35；And it calculates in image bmi
The mean value vbi=89.72 of all non-zero pixels.
16, image mask and image bgr are done into NAND operation, obtains image bmo, as shown in figure 36；And calculate image bmo
In all non-zero pixels mean value vbo=19.89；
17, using the eye fundus image of shooting as image src to be processed, extract in image src to be processed with image mask ruler
It is very little the same, and be image srcr with image of the image mask in image dl in the same region in position；Such as Figure 37 and Figure 38
It is shown.
18, by image mask and image srcr is done and operation, image smi is obtained, as shown in figure 39；And calculate image smi
In all non-zero pixels mean value vsi=199.66.
19, image mask and image srcr are done into NAND operation, obtains image smo, as shown in figure 40；And calculate image
The mean value vso=145.67 of all non-zero pixels in smo.
20, it calculates and weakens coefficient
21, image src to be processed subtracts the product of bg and k to get the image arrived after weakening hot spot, as shown in figure 41.
As shown in figure 42, (a)~(m) be 12 kinds of wavelength shoot come eye fundus image it is treated by the present method after figure
Picture.
Although an embodiment of the present invention has been shown and described, it will be understood by those skilled in the art that: not
A variety of change, modification, replacement and modification can be carried out to these embodiments in the case where being detached from the principle of the present invention and objective, this
The range of invention is defined by the claims and their equivalents.
Claims (7)
1. specular light removing method is imaged in a kind of multimode eye, which comprises the following steps:
S1, obtains a reference picture bg, and reference picture bg is to be shot in darkroom or simulation dark room conditions without target figure
Picture；
Reference picture bg is converted to gray level image gr by S2；
Gray level image gr binaryzation is obtained bianry image bn by S3；
S4 detects the pixel that whether there is non-zero in bianry image bn:
If there are the pixels of non-zero in bianry image bn, corrosion frequency n=0 is enabled, executes step S5；
If the pixel of non-zero is not present in bianry image bn, the eye fundus image of shooting is not processed, is terminated；
S5 carries out a morphological erosion operation to bianry image bn, obtains bianry image er；
S6 detects the pixel that whether there is non-zero in bianry image er:
If there are the pixel of non-zero, n=n+1 in bianry image er, and enable bn=er, return step S5,
If the pixel of non-zero is not present in bianry image er, S7 is thened follow the steps；
S7 carries out the operation of n times morphological dilations to bianry image bn, obtains image dl；
S8 obtains the minimum circumscribed rectangle of non-zero pixels in image dl, obtains rectangular area roi；
S9, the image for extracting the region roi in image dl is image mask, and the image for extracting the region roi in image bg is image
bgr；
Image mask and image bgr are done and operation, obtain image bmi by S10, calculate the equal of all non-zero pixels in image bmi
Value vbi；
Image mask and image bgr are done NAND operation, obtain image bmo, and calculate all non-zero pixels in image bmo by S11
Mean value vbo；
S12, using the eye fundus image of shooting as image src to be processed, the image for extracting the region roi in image src to be processed is
Image srcr；
Image mask and image srcr are done and operation, obtain image smi, calculate all non-zero pixels in image smi by S13
Mean value vsi；
Image mask and image srcr are done NAND operation, obtain image smo, calculate all non-zero pixels in image smo by S14
Mean value vso；
S15 is calculated and is weakened coefficient
S16, image src to be processed subtract reference picture bg and weaken the product of coefficient k to get the image arrived after weakening hot spot.
2. specular light removing method is imaged in multimode eye according to claim 1, which is characterized in that step S2 are as follows:
Centered on the geometric center of reference picture bg, the 1/m of reference picture bg wide is width, and the m is being greater than or equal to 1 just
Number, the 1/n of reference picture bg high are the image of high rectangular area as image bgc；The n is greater than or equal to 1 positive number, and
M, n be simultaneously be not 1；Image bgc is converted into gray level image gr；
Step S9 are as follows:
The image for extracting the region roi in image dl is image mask, and the image for extracting the region roi in image bgc is image bgr；
Step S12 are as follows: using the eye fundus image of shooting as image src to be processed, during the geometric center with image src to be processed is
The heart, the 1/m of image src wide to be processed are width, and the 1/n of image src high to be processed is the image of high rectangular area as image
Srcc, the image for extracting the region roi in image srcc is image srcr.
3. specular light removing method is imaged in multimode eye according to claim 2, which is characterized in that m=n=2.
4. specular light removing method is imaged in multimode eye according to claim 1, which is characterized in that in step s 2,
Using OTSU algorithm by gray level image gr binaryzation.
5. specular light removing method is imaged in multimode eye according to claim 1, which is characterized in that step S8 are as follows: right
The minimum circumscribed rectangle of non-zero pixels expands x times in the image dl of acquisition, and the x is positive number, obtains rectangular area roi.
6. specular light removing method is imaged in multimode eye according to claim 5, which is characterized in that x=0.5.
7. specular light removing method is imaged in multimode eye according to claim 1, which is characterized in that the ginseng in step S1
Examine the image that the eye fundus image in image bg and step S4 is the shooting of same camera.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810905552.5A CN109035262B (en) | 2018-08-10 | 2018-08-10 | Multi-mode-eye imaging specular reflection light elimination method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810905552.5A CN109035262B (en) | 2018-08-10 | 2018-08-10 | Multi-mode-eye imaging specular reflection light elimination method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109035262A true CN109035262A (en) | 2018-12-18 |
| CN109035262B CN109035262B (en) | 2021-08-17 |
Family
ID=64632594
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810905552.5A Active CN109035262B (en) | 2018-08-10 | 2018-08-10 | Multi-mode-eye imaging specular reflection light elimination method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109035262B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102590218A (en) * | 2012-01-16 | 2012-07-18 | 安徽中科智能高技术有限责任公司 | Device and method for detecting micro defects on bright and clean surface of metal part based on machine vision |
| CN102723054A (en) * | 2012-06-18 | 2012-10-10 | 西安电子科技大学 | Online calibration system and online calibration method for ununiformity of LED (light-emitting diode) display screen |
| CN103077517A (en) * | 2012-12-31 | 2013-05-01 | 华中科技大学 | Method for partitioning space target under non-uniform lighting condition |
| US20160089016A1 (en) * | 2014-09-30 | 2016-03-31 | Nidek Co., Ltd. | Fundus photographing apparatus with wavefront compensation and method for photographing fundus image |
| CN107451998A (en) * | 2017-08-08 | 2017-12-08 | 北京大恒普信医疗技术有限公司 | A kind of eye fundus image method of quality control |
-
2018
- 2018-08-10 CN CN201810905552.5A patent/CN109035262B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102590218A (en) * | 2012-01-16 | 2012-07-18 | 安徽中科智能高技术有限责任公司 | Device and method for detecting micro defects on bright and clean surface of metal part based on machine vision |
| CN102723054A (en) * | 2012-06-18 | 2012-10-10 | 西安电子科技大学 | Online calibration system and online calibration method for ununiformity of LED (light-emitting diode) display screen |
| CN103077517A (en) * | 2012-12-31 | 2013-05-01 | 华中科技大学 | Method for partitioning space target under non-uniform lighting condition |
| US20160089016A1 (en) * | 2014-09-30 | 2016-03-31 | Nidek Co., Ltd. | Fundus photographing apparatus with wavefront compensation and method for photographing fundus image |
| CN107451998A (en) * | 2017-08-08 | 2017-12-08 | 北京大恒普信医疗技术有限公司 | A kind of eye fundus image method of quality control |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109035262B (en) | 2021-08-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gagnon et al. | Procedure to detect anatomical structures in optical fundus images | |
| Spencer et al. | An image-processing strategy for the segmentation and quantification of microaneurysms in fluorescein angiograms of the ocular fundus | |
| Medhi et al. | An effective fovea detection and automatic assessment of diabetic maculopathy in color fundus images | |
| Youssef et al. | New feature-based detection of blood vessels and exudates in color fundus images | |
| US20180353066A1 (en) | Image processing apparatus and image processing method | |
| Richa et al. | Fundus image mosaicking for information augmentation in computer-assisted slit-lamp imaging | |
| Rekhi et al. | Automated detection and grading of diabetic macular edema from digital colour fundus images | |
| Naz et al. | Glaucoma detection in color fundus images using cup to disc ratio | |
| Naveen et al. | Diabetic retinopathy detection using image processing | |
| JP2008073280A (en) | Eye-fundus image processor | |
| Lu et al. | Automatic macula detection from retinal images by a line operator | |
| Semerád et al. | Retinal vascular characteristics | |
| Walter et al. | Automatic analysis of color fundus photographs and its application to the diagnosis of diabetic retinopathy | |
| Goldbaum et al. | Image understanding for automated retinal diagnosis | |
| Escorcia-Gutierrez et al. | Interactive optic disk segmentation via discrete convexity shape knowledge using high-order functionals | |
| CN109035262A (en) | A kind of multimode eye imaging specular light removing method | |
| Perez-Rovira et al. | Robust optic disc location via combination of weak detectors | |
| CN115868920A (en) | Quantitative evaluation method and device for conjunctival hyperemia and storage medium | |
| CN108230287B (en) | Method and device for detecting crystal region of anterior segment image | |
| Amrutha et al. | An efficient ridge detection method for retinopathy of prematurity severity analysis | |
| WO2011108995A1 (en) | Automatic analysis of images of the anterior chamber of an eye | |
| Sahoo et al. | A novel optic disc and blood vessel detection algorithm | |
| Joshi et al. | Fundus image analysis for detection of fovea: A review | |
| Shami et al. | Better detection of retinal abnormalities by accurate detection of blood vessels in retina | |
| Xu et al. | Simultaneous automatic detection of optic disc and fovea on fundus photographs |
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 |