CN106548471B - The medical microscopic images clarity evaluation method of coarse-fine focusing - Google Patents
The medical microscopic images clarity evaluation method of coarse-fine focusing Download PDFInfo
- Publication number
- CN106548471B CN106548471B CN201610922544.2A CN201610922544A CN106548471B CN 106548471 B CN106548471 B CN 106548471B CN 201610922544 A CN201610922544 A CN 201610922544A CN 106548471 B CN106548471 B CN 106548471B
- Authority
- CN
- China
- Prior art keywords
- image
- value
- color channel
- color
- pixel
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10056—Microscopic 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
Abstract
The invention discloses a kind of medical microscopic images clarity evaluation methods of coarse-fine focusing, and the RGB color image of input is first converted to Lab color space;Then pass through statistics luminance graph ILGray variance obtain grey scale change value, utilize Color Channel figure IaWith Color Channel figure IbColor notable figure is obtained, and obtains the partial gradient value of color notable figure;Finally, grey scale change value is multiplied with local change of gradient value, final image definition is obtained.Method proposed by the present invention comprehensively considers the global grey scale change and local change of gradient of image, not only there is wider steep area but also sensitivity with higher, big step-length rough focusing can be used for simultaneously using only a kind of sharpness evaluation function and small step-length is finely focused.
Description
Technical field
The invention belongs to technical field of image processing, more particularly, to a kind of medical microscopic images clarity evaluation side
Method.
Background technique
Medical microscopic images are widely used in the medical researches such as blood cell analysis, chromosome analysis, urine sediment analysis and face
In bed medical diagnosis.By focusing acquisition automatically, clearly micro-image is subsequent Medical Image Processing, target identification and pathology
The premise of analysis.Microscope is amplified the sightless observation object of naked eyes by the effect of optical system, from micromorphology research
With the characteristic of object of cognition, therefore, microscope focus automatically be very accurate complexity process.
Existing microscope auto focusing method generallys use the passive type type of focusing based on image feedback, i.e., using clear
Clear degree evaluation function carries out clarity evaluation to the image that different defocus station acquisitions arrive, and is found with this with best sharpness
Focal position.Therefore, the key focused automatically is the design of Image Definition.One ideal medicine micrograph
Image sharpness evaluation function should also be able to realize simultaneously fast other than should have unbiasedness, unimodality and anti-noise ability
Fast big step-length rough focusing and small step-length are finely focused.
The automatic focusing of microscope is generally divided into rough focusing and finely two stages of focusing.It is micro- in the rough focusing stage
The position of mirror objective table is often far from focal position, for quick close focal position, it usually needs uses big step-length rough focusing.
In order to guarantee to reach focal position, sharpness evaluation function at this time must have sufficiently wide steep area to avoid because of step
Long excessive focusing failure caused by crossing steep area.In the finely focusing stage, generally requires and obtained accurately using small step-length
Focal position, sharpness evaluation function at this time, which should have sufficiently high sensitivity, could obtain most accurate focal position.
Currently, there are two main classes for most widely used sharpness evaluation function: gradient function and statistical function.Based on gradient
The sharpness evaluation function of function, such as the sum of grey scale difference absolute value (SMD), Brenner function, gradient of image and gray scale energy letter
The functions such as number (Tenengrad), Laplce's energy (EOL), their sensitivity is good, but steep area is narrow, is not suitable for big
Range rough focusing;The calculation amount of the statistical functions such as gray variance is small, possesses wider steep area, and the monotonicity in shoulder
It is good, however, its poor sensitivity, it is burnt not to be suitable for small range accurate adjustment.In order to solve this problem, the image of some coarse-fine combinations is automatic
Focus method uses different sharpness evaluation functions in thick, thin two focusing stages respectively.However, the property of these focus methods
The influence of focusing strategy can be will receive, such as the selection of step-length, the search strategy in rough focusing stage.Therefore, at present still without one
Kind sharpness evaluation function can be achieved at the same time quickly big step-length rough focusing and small step-length and finely focus, that is, both have sufficiently wide
Steep area possess sufficiently high sensitivity again.
Summary of the invention
Aiming at the above defects or improvement requirements of the prior art, the present invention provides a kind of medicine micrographs of coarse-fine focusing
Image sharpness evaluation method had had it both sufficiently wide its object is to design a kind of function of thoroughly evaluating image definition
Steep area possess sufficiently high sensitivity again, no matter the rough focusing stage or can show in the fine focusing stage compared with
Good performance.
Technical solution proposed by the present invention is as follows:
A kind of medical microscopic images clarity evaluation method of coarse-fine focusing, which is characterized in that the method includes following
Several steps:
Step 1: the original image of input rgb format, and it is converted into Lab color space, respectively obtain luminance graph IL
With two Color Channel figures: Color Channel figure IaWith Color Channel figure Ib;
Step 2: passing through the grey scale change value F for obtaining original image with minor functionVar:
Wherein,For ILGray average, (x, y) be pixel position;
Step 3: obtaining the partial gradient value of original image, detailed process is as follows:
(3.1) respectively to Color Channel figure IaWith Color Channel figure IbIn the gray scale of each pixel carry out mean value and normalizing
Change operation, then operating result linear superposition is obtained into color notable figure S;
(3.2) each pixel calculates partial gradient value in color notable figure S resulting to above step, specifically, statistics
Current pixel transverse direction, longitudinal and the pixel grey scale absolute value of the difference on two tilted directions product;
G (x, y)=[| S (x+2, y)-S (x, y) | × | S (x, y+2)-S (x, y) |] ×
[| S (x+1, y+1)-S (x, y) | × | S (x-1, y+1)-S (x, y) |]
(3.3) only statistics is greater than partial gradient value of the sum of the functional value of thresholding T as input picture:
FGrad=∑ [G (x, y) > T]
In this step, it is preferable that T is set as 10;
Step 4: by variation of image grayscale value FVarWith local gradient value FGradIt is multiplied, obtains final image definition F.
Preferably, the step (3.1) specifically includes:
(3.1.1) carries out averaging operation using following formula:
Wherein,WithRespectively Color Channel figure IaWith Color Channel figure IbGray average, | | be absolute value sign,
I′aWith I 'bRespectively Color Channel figure IaWith Color Channel figure IbImage after removing averaging operation;
(3.1.2) is respectively to I 'aWith I 'bOperation is normalized, obtains normalized image NaWith normalized image Nb。
Na(x, y)=N (I 'a(x, y))
Nb(x, y)=N (I 'b(x, y))
Wherein, N () is normalized function, and formula is as follows:
Wherein, max () and min () is respectively maximum value and minimum value function;
(3.1.3) passes through following formula for normalized image NaWith normalized image NbLinear superposition obtains color notable figure
S:
(x, y) is the position of pixel in image.
What the present invention can reach has the beneficial effect that:
Medical microscopic images thick, thin two different focusing stages to sharpness evaluation function different requirement again, i.e.,
In the big step-length rough focusing stage need that there is wider steep area, small step-length finely focus the stage need it is with higher sensitive
Degree.Method proposed by the present invention has comprehensively considered the global grey scale change and local change of gradient of image, and it is a kind of clear to be used only
Degree evaluation function can be used for big step-length rough focusing simultaneously and small step-length is finely focused.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, below to specific implementation of the invention
Mode is further elaborated.
Step 1: the original image of input rgb format, and it is converted into Lab color space, respectively obtain luminance graph IL
With two Color Channel figures: Color Channel figure IaWith Color Channel figure Ib;
Step 2: the grey scale change value of original image is obtained, specifically, by calculating luminance graph I with minor functionLGray scale
Variance FVar:
Wherein,For ILGray average, (x, y) be pixel position.
Step 3: obtaining the partial gradient value of original image, detailed process is as follows:
(3.1) respectively to Color Channel figure IaWith Color Channel figure IbIn the gray scale of each pixel carry out mean value and normalizing
Change operation, then operating result linear superposition is obtained into color notable figure S.
It is the background interference in order to eliminate the large area in medical microscopic images that (3.1.1), which removes averaging operation, and formula is such as
Under:
Wherein,WithRespectively Color Channel figure IaWith Color Channel figure IbGray average, | | be absolute value sign,
I′aWith I 'bRespectively Color Channel figure IaWith Color Channel figure IbImage after removing averaging operation.When being carried on the back in medical microscopic images
When scene area area is much larger than target area area, in the color channel image after removing averaging operation, background area be will be close to
Zero, and target area is much larger than zero because having a significant color difference with background area.Therefore, averaging operation is gone to be conducive to disappear
Interference except background area to focusing.
(3.1.2) is respectively to I 'aWith I 'bOperation is normalized, obtains normalized image NaWith normalized image Nb。
Na(x, y)=N (I 'a(x, y))
Nb(x, y)=N (I 'b(x, y))
Wherein, N () is normalized function, and formula is as follows:
Wherein, max () and min () is respectively maximum value and minimum value function.Image N after normalizationaAnd NbPixel
Gray scale interval will expand to [0,255], increase the color contrast of target area and background area, to further eliminate back
Scape interference.
(3.1.3) passes through following formula for normalized image NaWith normalized image NbLinear superposition obtains color notable figure
S:
(3.2) partial gradient value is calculated to pixel each in the resulting color notable figure of above step, specifically, statistics is worked as
Preceding pixel transverse direction, longitudinal and the pixel grey scale absolute value of the difference on two tilted directions product;
G (x, y)=[| S (x+2, y)-S (x, y) | × | S (x, y+2)-S (x, y) |] ×
[| S (x+1, y+1)-S (x, y) | × | S (x-1, y+1)-S (x, y) |]
Wherein, | S (x+2, y)-S (x, y) | and | S (x, y+2)-S (x, y) | it is respectively used to extract horizontal direction and Vertical Square
To gradient value, | S (x+1, y+1)-S (x, y) | and | S (x-1, y+1)-S (x, y) | be respectively used to extract oblique and anti-slanting
Gradient value.Therefore, G (x, y) combines the gradient value of four direction, can more comprehensively reflect the partial gradient variation of each pixel.
(3.3) in order to reduce the interference of ambient noise, only the sum of the functional value of statistics greater than thresholding T is as input picture
Partial gradient value:
FGrad=∑ [G (x, y) > T]
In this step, it is preferable that T is set as 10.
Step 4: by variation of image grayscale value FVarWith local gradient value FGradIt is multiplied, obtains final image definition F:
F=FVar×FGrad
When from focal position farther out, the partial gradient variation of image is unobvious, and slight fluctuations state is presented;And global ash
Degree variation is more obvious, is in apparent monotonicity.The change rate of F is by F at this timeVarIt is leading, therefore, the precipitous sector width of F close to
FVarPrecipitous sector width.When Range Focusing position is more and more closer, the detailed information such as edge, texture of image are gradually enriched, office
Portion's change of gradient is much larger than global grey scale change, at this point, the change rate of F is mainly by FGradIt determines, and the change rate of F is greater than FGrad,
Therefore the sensitivity of F is more preferable.
The medical microscopic images clarity evaluation method of coarse-fine focusing of the invention first turns the RGB color image of input
It is changed to Lab color space;Then grey scale change value is obtained by the gray variance of statistics luminance graph L, using Color Channel figure a and
Color Channel figure b obtains notable figure, and obtains the partial gradient value of notable figure;Finally, by grey scale change value and local gradient value
It is multiplied, obtains final image definition.Method proposed by the present invention comprehensively considers the global grey scale change and part ladder of image
Degree variation, not only has wider steep area but also sensitivity with higher, and a kind of sharpness evaluation function is used only can be same
When finely focus for big step-length rough focusing and small step-length.
Claims (2)
1. a kind of medical microscopic images clarity evaluation method of coarse-fine focusing, which is characterized in that the method includes following several
A step:
Step 1: the original image of input rgb format, and it is converted into Lab color space, respectively obtain luminance graph ILWith two
A Color Channel figure: Color Channel figure IaWith Color Channel figure Ib;
Step 2: passing through the grey scale change value F for obtaining original image with minor functionVar:
Wherein,For ILGray average, (x, y) be pixel position;
Step 3: obtaining the partial gradient value of original image, detailed process is as follows:
(3.1) respectively to Color Channel figure IaWith Color Channel figure IbIn each pixel gray scale carry out mean value and normalization behaviour
Make, then operating result linear superposition is obtained into color notable figure S;
(3.2) each pixel calculates partial gradient value in color notable figure S resulting to above step, and specifically, statistics is current
Pixel transverse direction, longitudinal and the pixel grey scale absolute value of the difference on two tilted directions product;
G (x, y)=[| S (x+2, y)-S (x, y) | × | S (x, y+2)-S (x, y) |] × [| S (x+1, y+1)-S (x, y) | × | S
(x-1,y+1)-S(x,y)|]
(3.3) only statistics is greater than partial gradient value of the sum of the functional value of thresholding T as input picture:
FGrad=∑ [G (x, y) > T]
In this step, T is set as 10;
Step 4: by variation of image grayscale value FVarWith local gradient value FGradIt is multiplied, obtains final image definition F.
2. the medical microscopic images clarity evaluation method of coarse-fine focusing as described in claim 1, which is characterized in that described
Step (3.1) specifically includes:
(3.1.1) carries out averaging operation using following formula:
Wherein,WithRespectively Color Channel figure IaWith Color Channel figure IbGray average, | | be absolute value sign, I 'aWith
I′bRespectively Color Channel figure IaWith Color Channel figure IbImage after removing averaging operation;
(3.1.2) is respectively to I 'aWith I 'bOperation is normalized, obtains normalized image NaWith normalized image Nb;
Na(x, y)=N (I 'a(x,y))
Nb(x, y)=N (I 'b(x,y))
Wherein, N () is normalized function, and formula is as follows:
Wherein, max () and min () is respectively maximum value and minimum value function;
(3.1.3) passes through following formula for normalized image NaWith normalized image NbLinear superposition obtains color notable figure S:
(x, y) is the position of pixel in image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610922544.2A CN106548471B (en) | 2016-10-18 | 2016-10-18 | The medical microscopic images clarity evaluation method of coarse-fine focusing |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610922544.2A CN106548471B (en) | 2016-10-18 | 2016-10-18 | The medical microscopic images clarity evaluation method of coarse-fine focusing |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106548471A CN106548471A (en) | 2017-03-29 |
CN106548471B true CN106548471B (en) | 2019-04-05 |
Family
ID=58392359
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610922544.2A Active CN106548471B (en) | 2016-10-18 | 2016-10-18 | The medical microscopic images clarity evaluation method of coarse-fine focusing |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106548471B (en) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106991672B (en) * | 2017-03-30 | 2019-12-13 | 梧州学院 | Objective lens positioning method based on light intensity |
CN107767418A (en) * | 2017-10-25 | 2018-03-06 | 梧州学院 | A kind of low power microcobjective identification and localization method |
CN109361849B (en) * | 2018-09-30 | 2021-03-05 | 桂林优利特医疗电子有限公司 | Automatic focusing method |
CN111062916B (en) * | 2019-12-05 | 2023-12-29 | 苏州大学 | Definition evaluation method and device for microscopic image |
CN114785959B (en) * | 2022-06-16 | 2022-10-14 | 江苏美克医学技术有限公司 | Automatic focusing method and device for fluorescence microscope, storage medium and electronic equipment |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101902366A (en) * | 2009-05-27 | 2010-12-01 | 北京启明星辰信息技术股份有限公司 | Method and system for detecting abnormal service behaviors |
CN102496157A (en) * | 2011-11-22 | 2012-06-13 | 上海电力学院 | Image detection method based on Gaussian multi-scale transform and color complexity |
CN103985108A (en) * | 2014-06-03 | 2014-08-13 | 北京航空航天大学 | Method for multi-focus image fusion through boundary detection and multi-scale morphology definition measurement |
-
2016
- 2016-10-18 CN CN201610922544.2A patent/CN106548471B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101902366A (en) * | 2009-05-27 | 2010-12-01 | 北京启明星辰信息技术股份有限公司 | Method and system for detecting abnormal service behaviors |
CN102496157A (en) * | 2011-11-22 | 2012-06-13 | 上海电力学院 | Image detection method based on Gaussian multi-scale transform and color complexity |
CN103985108A (en) * | 2014-06-03 | 2014-08-13 | 北京航空航天大学 | Method for multi-focus image fusion through boundary detection and multi-scale morphology definition measurement |
Non-Patent Citations (1)
Title |
---|
Gergely Windisch and Miklós Kozlovszky.Evaluating Sharpness Metrics for HE Stained.《CINTI 2014 •15th IEEE International Symposium on Computational Intelligence and Informatics》.2014, |
Also Published As
Publication number | Publication date |
---|---|
CN106548471A (en) | 2017-03-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106548471B (en) | The medical microscopic images clarity evaluation method of coarse-fine focusing | |
CN104637064B (en) | A kind of defocus blur image definition detection method based on edge strength weight | |
CN106683080B (en) | A kind of retinal fundus images preprocess method | |
CN101221118A (en) | System and method for intelligent recognizing and counting sputum smear micro-image tubercle bacillus | |
US9934571B2 (en) | Image processing device, program, image processing method, computer-readable medium, and image processing system | |
CN100399183C (en) | Self-adaptive automatic focusing method used in digital camera | |
CN107133929B (en) | The low quality file and picture binary coding method minimized based on background estimating and energy | |
CN105868745B (en) | Weather recognition methods based on dynamic scene perception | |
CN107832801B (en) | Model construction method for cell image classification | |
JP4663602B2 (en) | Automatic focusing device, microscope and automatic focusing method | |
CN107545550B (en) | Cell image color cast correction method | |
CN110531484A (en) | A kind of microscope Atomatic focusing method that focus process model can be set | |
CN112037185B (en) | Chromosome splitting phase image screening method and device and terminal equipment | |
Costa et al. | A sputum smear microscopy image database for automatic bacilli detection in conventional microscopy | |
CN115047610B (en) | Chromosome karyotype analysis device and method for automatically fitting microscopic focusing plane | |
CN111105346A (en) | Full-scanning microscopic image splicing method based on peak value search and gray template registration | |
CN108665436B (en) | Multi-focus image fusion method and system based on gray mean reference | |
CN108364296B (en) | Cell population space distribution construction method based on multilayer holographic reconstruction and focusing strategy | |
Wang et al. | Automatic dissection position selection for cleavage-stage embryo biopsy | |
CN113763401B (en) | Quick multi-point automatic focusing method, system and application equipment thereof | |
CN111199536A (en) | Focus evaluation method and device | |
Wollmann et al. | Multi-channel deep transfer learning for nuclei segmentation in glioblastoma cell tissue images | |
Hao et al. | Improving the performances of autofocus based on adaptive retina-like sampling model | |
CN106846348B (en) | The method of glasses is automatically removed in facial image | |
Gherardi et al. | Illumination field estimation through background detection in optical microscopy |
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 |