CN106875413A - A kind of adhesion red blood cell automatic counting method based on high light spectrum image-forming - Google Patents
A kind of adhesion red blood cell automatic counting method based on high light spectrum image-forming Download PDFInfo
- Publication number
- CN106875413A CN106875413A CN201710076460.6A CN201710076460A CN106875413A CN 106875413 A CN106875413 A CN 106875413A CN 201710076460 A CN201710076460 A CN 201710076460A CN 106875413 A CN106875413 A CN 106875413A
- Authority
- CN
- China
- Prior art keywords
- image
- cytoplasm
- bianry image
- connected domain
- end member
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30242—Counting objects in image
Landscapes
- Image Processing (AREA)
- Investigating Or Analysing Biological Materials (AREA)
Abstract
The invention discloses a kind of adhesion red blood cell automatic counting method based on high light spectrum image-forming, comprise the following steps：Read in the hyperspectral image data of blood film and be compressed, the high-spectral data after compression is decomposed using continuous maximum angular convex cone method, obtain the abundance figure of blood film preset number end member；Binary conversion treatment is carried out to the abundance figure of each end member with reference to Da-Jin algorithm, and tiny noise spot is removed using Mathematical Morphology Method erosion operation；Holes filling is carried out to each end member bianry image, the size according to connected domain number and largest connected domain selects cytoplasm bianry image；Cytoplasm connected domain to cytoplasm bianry image is marked and area statistics, chooses the intermediate value of cytoplasm connected domain area as reference valueR；All cytoplasm connected domains are identified and counting.The present invention takes full advantage of spectrum and image information, solves the enumeration problem of adhesion red blood cell, improves the accuracy of the automatic count results of red blood cell.
Description
Technical field
The present invention relates to digital image processing techniques field, more particularly to a kind of adhesion red blood cell based on high light spectrum image-forming
Automatic counting method.
Background technology
Red blood cell carries oxygen transportation and immunologic function as a kind of most common haemocyte.Red blood cell count(RBC) is blood
The important indicator of routine inspection, has important reference value in terms of prevention from suffering from the diseases and diagnosis.At present, automatic blood cell point
Analyzer is effectively improved analysis speed, but its analysis result has false negative rate higher, and part sample still needs inspection
Member carries out microscope and rechecks to reduce rate of missed diagnosis and misdiagnosis rate again.Blood film microexamination is used as clinically judging haemocyte
The goldstandard of pathological change is still essential analysis means.And blood cell differential of the tradition based on analysis of digital microscopy images and knowledge
Other method can free reviewer from cumbersome time-consuming microscopy work, reduce the erroneous judgement that human factor causes, and carry
Discrimination high.But due to the interference of cellular morphology diversity, cytoadherence and some compositions, still fail to look for so far
To a kind of method for meeting clinical requirement to arbitrary cell image procossing precision.
Be combined for traditional optical imagery and spectral technique by micro- high light spectrum image-forming technology, is obtaining sample space information
While, also for each pixel in image provides dozens of to hundreds of narrow-band spectrum information, blood can not only be analyzed thin
The morphosis of born of the same parents, additionally it is possible to analyze cell component content relevant information, to realize that more accurate specimen discerning, biochemical parameter are carried
Take etc. provide may, be expected to solve the above-mentioned bottleneck problem based on image method.But how to make full use of acquired figure
Picture and spectral information, the accuracy for improving quantification and qualification is still the key of high light spectrum image-forming technology application.
The content of the invention
It is an object of the invention to provide a kind of adhesion red blood cell automatic counting method based on high light spectrum image-forming, the method
Spectrum and image information can be made full use of, adhesion red blood cell is efficiently identified, the accuracy rate of red blood cell count(RBC) is improved.
Realizing the concrete technical scheme of the object of the invention is：
A kind of adhesion red blood cell automatic counting method based on high light spectrum image-forming, the described method comprises the following steps：
(1) read in the hyperspectral image data of blood film and be compressed, using continuous maximum angular convex cone method to passing through
The hyperspectral image data for compressing treatment is decomposed, and obtains the abundance figure of blood film preset number end member；
(2) binary conversion treatment is carried out respectively to the abundance figure of the blood film preset number end member with reference to Da-Jin algorithm, and is adopted
Tiny noise spot is removed with Mathematical Morphology Method erosion operation, the bianry image of preset number end member abundance figure is obtained；
(3) bianry image to the preset number end member abundance figure carries out holes filling operation respectively, and according to process
The number of connected domain and the size in largest connected domain select cytoplasm bianry image in each end member bianry image after holes filling；
(4) the cytoplasm connected domain to the cytoplasm bianry image is marked and area statistics, chooses cytoplasm and connects
The intermediate value of logical domain area is used as reference value R；
(5) all cytoplasm connected domains are identified and counting.
Further, step (1) is specifically included：
Read in the hyperspectral image data Data (2x, 2y, λ) of blood film；
The image of each wave band in the hyperspectral image data is compressed using quadratic linear interpolation method, obtains pressure
Hyperspectral image data Data ' (x, y, λ) after contracting；
The hyperspectral image data Data ' (x, y, λ) after the compression is divided using continuous maximum angular convex cone method
Solution, obtains the abundance figure I of n end member of blood filmj(x, y) (j=1,2 ..., n).
Further, step (2) is specifically included：
Adaptively obtain the abundance figure I of n end member of the blood film respectively using Da-Jin algorithmj(x, y) (j=1,
2 ..., segmentation threshold T n)j(j=1,2 .., n)；
According to the segmentation threshold Tj(j=1,2 .., n) respectively to the abundance figure I of the n end memberj(x, y) (j=1,
2 .., n) carry out binary conversion treatment, and to carry out the removal of mathematical morphology erosion operation tiny using the structural element of 3 × 3 sizes
Noise spot, obtains the abundance figure I of the n end memberj(x, y) (j=1,2 .., n) corresponding bianry image Bj(x, y) (j=1,
2 .., n).
Further, step (3) is specifically included：
To the bianry image Bj(x, y) (j=1,2 .., n) carry out inversion operation respectively, obtain inverse bianry image Rj
(x, y) (j=1,2 .., n)；
Respectively by the inverse bianry image Rj(x, y) (j=1,2 .., n) in the pixel value in largest connected domain negate, its
The pixel value of his connected domain keeps constant, obtains image Rj' (x, y) (j=1,2 .., n), to the bianry image Bj(x, y) (j
=1,2 .., n) and corresponding described image Rj' (x, y) (j=1,2 .., n) carry out xor operation, after obtaining holes filling
Bianry image Bj' (x, y) (j=1,2 .., n)；
The bianry image B after described hole filling is obtained respectivelyj' (x, y) (and j=1,2 .., n) in connected domain number Nj
(j=1,2 .., n) and largest connected domain area Sj(j=1,2 .., n), and are judged, if N respectivelyjMore than 100 with
And SjMore than 1000 and less than 4000, then the bianry image B after described hole is filledj' (x, y) be cytoplasm bianry image Bc(x,
y)。
Further, step (4) is specifically included：
To the cytoplasm bianry image BcCytoplasm connected domain in (x, y) is marked, and is counted respectively according to mark
The size in each respective markers region, and using the intermediate value of size as reference value R.
Further, step (5) is specifically included：
Successively to the cytoplasm bianry image BcThe area of the cytoplasm connected domain of each mark is carried out in (x, y)
Judge；
If the area of the cytoplasm connected domain is less than 0.5 × R, do not counted；If more than 0.5 × R and less than 1.9
× R, is designated as individual cells；If being more than 1.9 × R, it is fitted using Least Chimb shape, if fitting convex-edge shape area is more than 2.5
× R, then be designated as two ACs, is otherwise designated as individual cells.
The beneficial effect of technical scheme that the present invention is provided is：Proposed by the invention is a kind of viscous based on high light spectrum image-forming
Connect red blood cell automatic counting method, by hyperspectral image data decompose the abundance figure of the main end member of acquisition, different ends
Unit is related to different component in blood, assists in removing the information unrelated with counting, and further abundance figure is carried out at binaryzation
Reason and holes filling operate to automatically extract cytoplasm bianry image, sentence by cytoplasmic identification and to cytoplasm size
Not, influence of the AC to count results is solved, the accuracy rate of the automatic count results of red blood cell is improve.
Brief description of the drawings
A kind of flow chart of adhesion red blood cell automatic counting method based on high light spectrum image-forming that Fig. 1 is provided for the present invention；
The holes filling operational flowchart that Fig. 2 is provided for the present invention；
Fig. 3 chooses flow chart for the cytoplasm bianry image that the present invention is provided；
The adhesion Erythrocyte Recognition flow chart that Fig. 4 is provided for the present invention；
Fig. 5 is the abundance image of 5 end members of the embodiment of the present invention；
Fig. 6 is 5 bianry images of end member of the embodiment of the present invention；
Fig. 7 is 5 bianry images of end member by holes filling of the embodiment of the present invention；
Fig. 8 is cytoplasm bianry image (a) and adhesion Erythrocyte Recognition result figure of the embodiment of the present invention.
Specific embodiment
Technological means of the present invention, technological improvement and beneficial effect in order to become more apparent are illustrated, is below tied
The present invention will be described in detail to close accompanying drawing.
A kind of adhesion red blood cell automatic counting method based on high light spectrum image-forming provided by the present invention, referring to Fig. 1, Fig. 2,
Fig. 3 and Fig. 4, including following steps：
S101：Read in the hyperspectral image data of blood film and be compressed, using continuous maximum angular convex cone method to warp
The hyperspectral image data of overcompression treatment is decomposed, and obtains the abundance figure of blood film preset number end member.
The step is specially：
Read in the hyperspectral image data Data (2x, 2y, λ) of blood film；
The row, column number of each band image in the hyperspectral image data is compressed successively using quadratic linear interpolation method
It is original half, obtains the hyperspectral image data Data ' (x, y, λ) after compression；
The hyperspectral image data Data ' (x, y, λ) after the compression is divided using continuous maximum angular convex cone method
Solution, obtains the abundance figure I of 5 end members of blood filmj(x, y) (j=1,2 ..., 5)；
Wherein, according to being configured the need in practical application, the embodiment of the present invention is not done to this for the selection of preset number n
Limitation, this is illustrated as a example by sentencing n=5.
S102：Binary conversion treatment is carried out respectively to the abundance figure of the blood film preset number end member with reference to Da-Jin algorithm, and
Tiny noise spot is removed using Mathematical Morphology Method erosion operation, the bianry image of preset number end member abundance figure is obtained.
The step is specially：
Adaptively obtain the abundance figure I of 5 end members of the blood film respectively using Da-Jin algorithmj(x, y) (j=1,
2 ..., segmentation threshold T 5)j(j=1,2 ..., 5)；
By 5 end members abundance figure Ij(x, y) (j=1,2 ..., 5) in each pixel pixel value with segmentation threshold
Value Tj(5) j=1,2 ..., be compared, if the pixel value of pixel is more than Tj(j=1,2 ..., 5), then by the pixel
Pixel value is set to 1, and the pixel value of the pixel otherwise is set into 0, and 5 abundance figure I of end member are obtained respectivelyj(x, y) (j=1,
2 ..., 5) corresponding initial binary image Oj(x, y) (j=1,2 ..., 5)；
Using the square structure element of 3 × 3 sizes respectively to the initial binary image Oj(x, y) (j=1,2 ...,
5) carry out mathematical morphology erosion operation and remove tiny noise spot, obtain 5 abundance figure I of end memberj(x, y) (j=1,2 ...,
5) corresponding bianry image Bj(x, y) (j=1,2 ..., 5).
S103：Bianry image to the preset number end member abundance figure carries out holes filling operation respectively, and according to warp
The number of connected domain and the size in largest connected domain select cytoplasm binary map in each end member bianry image crossed after holes filling
Picture.
The step is specially：
To the bianry image Bj(x, y) (5) j=1,2 ..., carry out inversion operation respectively, obtains inverse bianry image Rj
(x, y) (j=1,2 ..., 5)；
Using connected component labeling algorithm to the inverse bianry image Rj(x, y) (j=1,2 ..., 5) in connected domain point
It is not marked, obtains the inverse bianry image L of markj(x, y) (j=1, then 2 ..., 5), bianry image Lj(x, y) (j=1,
2 ..., 5) in mjThe pixel value of pixel is just m in individual connected domainj, wherein mj=1,2 ..., Mj；
The inverse bianry image L of the mark is counted respectivelyj(x, y) (j=1,2 ..., 5) in pixel value be mjPixel
The number of point, is designated as Num (mj)(mj=1,2 ..., Mj), and find out Num (mj)(mj=1,2 ..., Mj) in maximum institute
The pixel value of corresponding pixel is designated as pj；
By the inverse bianry image L of the markj(x, y) (j=1,2 ..., 5) in pixel value be pjPixel picture
Plain value is set to 0, and the pixel value of rest of pixels point keeps constant, obtains bianry image Rj' (x, y) (j=1,2 ..., 5)；
To the bianry image Bj5) and corresponding bianry image R (x, y) (j=1,2 ...,j' (x, y) (j=1,
2 ..., 5) carry out xor operation, obtain the bianry image B after holes fillingj' (x, y) (j=1,2 ..., 5)；
Using connected component labeling algorithm to the bianry image Bj' (x, y) (and j=1,2 ..., connected domain 5) enter rower
Note, obtains the bianry image L of markj' (x, y) (j=1, then 2 ..., 5), the bianry image L of the markj' (x, y) (j=1,
2 ..., 5) in kthjThe pixel value of pixel is just k in individual connected domainj, wherein kj=1,2 ..., Nj, Nj(j=1,2 ...,
5) it is the bianry image Bj' (x, y) (and j=1,2 ..., 5) in connected domain number；The two-value of the mark is counted respectively
Image Lj' (x, y) (and j=1,2 ..., 5) in pixel value be kjPixel number, be designated as Num (kj)(kj=1,2 ...,
Nj), Num (kj)(kj=1,2 ..., Nj) in maximum be the bianry image Bj' (x, y) (and j=1,2 ..., 5) in most
The area of big connected domain, is designated as Sj；
If meeting NjMore than 100 and SjMore than 1000 and less than 4000, then the bianry image B after described hole is filledj’
(x, y) (j=1,2 ..., 5) be cytoplasm bianry image Bc(x, y).
S104：Cytoplasm connected domain to the cytoplasm bianry image is marked and area statistics, chooses cytoplasm
The intermediate value of connected domain area is used as reference value R.
The step is specially：
Using connected component labeling algorithm to the cytoplasm bianry image BcEach cytoplasm connected domain in (x, y) is carried out
Mark, obtains the cytoplasm bianry image L of markc(x, y), then LcM in (x, y)cPixel in individual cytoplasm connected domain
Pixel value is mc, wherein mc=1,2 ..., Mc；
Count the cytoplasm bianry image L of the markcPixel value is m in (x, y)cPixel number Num (mc)
(mc=1,2 ..., Mc)；
The pixel value for taking the statistics is mcPixel number Num (mc)(mc=1,2 ..., Mc) intermediate value, as
The reference value R of cytoplasm area.
S105：All cytoplasm connected domains are identified and counting.
The step is specially：
It is successively m by the pixel value of the statisticscPixel number Num (mc)(mc=1,2 ..., Mc) with it is described
The reference value R of cytoplasm area is compared；
If the number Num (m of the pixelc) (m=1,2 ..., M ') be less than 0.5 × R, then do not counted；If big
In 0.5 × R and less than 1.9 × R, individual cells are designated as；
If the number Num (m of the pixelc)(mc=1,2 ..., Mc) it is more than 1.9 × R, processed using morphological images
In convex hull algorithm to Num (mc)(mc=1,2 ..., Mc) corresponding cytoplasm connected domain carries out Least Chimb shape fitting, and unite
Count the area S of the fitting convex-edge shapecIf, Sc2.5 × R of ＞, are designated as two cells, are otherwise designated as a cell.
Fig. 5 for the embodiment of the present invention 5 end members abundance figure, respectively correspond to white space (a), stain (b), other
(c), cell membrane (d) and cytoplasm (e).
Fig. 6 is 5 bianry images of end member of the embodiment of the present invention.
Fig. 7 is 5 bianry images of end member after the holes filling of the embodiment of the present invention.
Fig. 8 recognizes knot for cytoplasm bianry image (a) and corresponding cytoplasm after the holes filling of the embodiment of the present invention
Really (b)-(d), wherein the connected domain in (b) is the cytoplasm not counted, the connected domain in (c) is red thin to be identified as one
The cytoplasm of born of the same parents, the connected domain in (d) is to be identified as two cytoplasm of red blood cell, it can be seen that Erythrocyte Recognition effect compared with
It is good.
In sum, the beneficial effect of the technical scheme that the present invention is provided is：One kind proposed by the invention is based on bloom
The adhesion red blood cell automatic counting method of imaging is composed, by the abundance for hyperspectral image data decompose the main end member of acquisition
Figure, because different end members are related to different component in blood, assists in removing the information unrelated with counting, further to abundance figure
Carry out binary conversion treatment and holes filling operation to automatically extract cytoplasm bianry image, by counting the individual of cytoplasm connected domain
Number, cytoplasm size, and operation is carried out to statistical value reach to cytoplasmic indirect identification, AC is solved to counting
The influence of result, improves the accuracy rate of the automatic count results of red blood cell.According to different application backgrounds, the present invention is by appropriate
Modification be equally applicable to other associated picture process fields.
The foregoing is only presently preferred embodiments of the present invention, be not intended to limit the invention, it is all it is of the invention spirit and
Within principle, any modification, equivalent substitution and improvements made etc. should be included within the scope of the present invention.
Claims (6)
1. a kind of adhesion red blood cell automatic counting method based on high light spectrum image-forming, it is characterised in that methods described includes following
Step：
(1) read in the hyperspectral image data of blood film and be compressed, using continuous maximum angular convex cone method to through overcompression
The hyperspectral image data for the treatment of is decomposed, and obtains the abundance figure of blood film preset number end member；
(2) binary conversion treatment is carried out respectively to the abundance figure of the blood film preset number end member with reference to Da-Jin algorithm, and uses number
Learn morphological method erosion operation and remove tiny noise spot, obtain the bianry image of preset number end member abundance figure；
(3) bianry image to the preset number end member abundance figure carries out holes filling operation respectively, and according to by hole
The number of connected domain and the size in largest connected domain select cytoplasm bianry image in each end member bianry image after filling；
(4) the cytoplasm connected domain to the cytoplasm bianry image is marked and area statistics, chooses cytoplasm connected domain
The intermediate value of area is used as reference value R；
(5) all cytoplasm connected domains are identified and counting.
2. the adhesion red blood cell automatic counting method based on high light spectrum image-forming according to claim 1, it is characterised in that institute
The detailed process of the step of stating (1) is：
Read in the hyperspectral image data Data (2x, 2y, λ) of blood film；
The image of each wave band in the hyperspectral image data is compressed using quadratic linear interpolation method, after obtaining compression
Hyperspectral image data Data ' (x, y, λ)；
The hyperspectral image data Data ' (x, y, λ) after the compression is decomposed using continuous maximum angular convex cone method, is obtained
Take the abundance figure I of n end member of blood filmj(x, y) (j=1,2 ..., n).
3. the adhesion red blood cell automatic counting method based on high light spectrum image-forming according to claim 1, it is characterised in that institute
The detailed process of the step of stating (2) is：
Adaptively obtain the abundance figure I of n end member of the blood film respectively using Da-Jin algorithmj(x, y) (j=1,2 ..., n)
Segmentation threshold Tj(j=1,2 .., n)；
According to the segmentation threshold Tj(j=1,2 .., n) respectively to the abundance figure I of the n end memberj(x, y) (j=1,2 ..,
N) binary conversion treatment is carried out, and uses the structural element of 3 × 3 sizes to carry out mathematical morphology erosion operation to remove tiny noise
Point, obtains the abundance figure I of the n end memberj(x, y) (j=1,2 .., n) corresponding bianry image Bj(x, y) (j=1,2 ..,
n)。
4. the adhesion red blood cell automatic counting method based on high light spectrum image-forming according to claim 1, it is characterised in that institute
The detailed process of the step of stating (3) is：
To the bianry image Bj(x, y) (j=1,2 .., n) carry out inversion operation respectively, obtain inverse bianry image Rj(x, y)
(j=1,2 .., n)；
Respectively by the inverse bianry image Rj(x, y) (j=1,2 .., n) in the pixel value in largest connected domain negate, other companies
The pixel value in logical domain keeps constant, obtains image Rj' (x, y) (j=1,2 .., n), to the bianry image Bj(x, y) (j=1,
2 .., n) and corresponding described image Rj' (x, y) (j=1,2 .., n) carry out xor operation, obtain the two-value after holes filling
Image Bj' (x, y) (j=1,2 .., n)；
The bianry image B after described hole filling is obtained respectivelyj' (x, y) (and j=1,2 .., n) in connected domain number Nj(j=
1,2 .., n) and largest connected domain area Sj(j=1,2 .., n), and are judged, if N respectivelyjMore than 100 and Sj
More than 1000 and less than 4000, then the bianry image B after described hole is filledj' (x, y) be cytoplasm bianry image Bc(x, y).
5. the adhesion red blood cell automatic counting method based on high light spectrum image-forming according to claim 1, it is characterised in that institute
The detailed process of the step of stating (4) is：
To the cytoplasm bianry image BcCytoplasm connected domain in (x, y) is marked, and each phase is counted respectively according to mark
Answer the size of marked region, and using the intermediate value of size as reference value R.
6. the adhesion red blood cell automatic counting method based on high light spectrum image-forming according to claim 1, it is characterised in that institute
The detailed process of the step of stating (5) is：
Successively to the cytoplasm bianry image BcThe area of the cytoplasm connected domain of each mark is judged in (x, y)；
If the area of the cytoplasm connected domain is less than 0.5 × R, do not counted；If more than 0.5 × R and being less than 1.9 × R,
It is designated as individual cells；If being more than 1.9 × R, it is fitted using Least Chimb shape, if fitting convex-edge shape area is more than 2.5 × R,
Two ACs are then designated as, individual cells are otherwise designated as.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710076460.6A CN106875413B (en) | 2017-02-13 | 2017-02-13 | A kind of adhesion red blood cell automatic counting method based on high light spectrum image-forming |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710076460.6A CN106875413B (en) | 2017-02-13 | 2017-02-13 | A kind of adhesion red blood cell automatic counting method based on high light spectrum image-forming |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106875413A true CN106875413A (en) | 2017-06-20 |
| CN106875413B CN106875413B (en) | 2019-05-17 |
Family
ID=59165887
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710076460.6A Active CN106875413B (en) | 2017-02-13 | 2017-02-13 | A kind of adhesion red blood cell automatic counting method based on high light spectrum image-forming |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106875413B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110110709A (en) * | 2019-05-27 | 2019-08-09 | 湖南开启时代智能科技有限公司 | A kind of red white corpuscle differential counting method, system and equipment based on image procossing |
| CN110956612A (en) * | 2019-10-31 | 2020-04-03 | 华侨大学 | Cell rapid counting method, computer equipment and readable storage medium |
| CN113222969A (en) * | 2021-05-28 | 2021-08-06 | 大连海事大学 | Imaging flow cytometry adhesion counting and activity detection method and device based on bidirectional background difference method |
| CN113269745A (en) * | 2021-05-21 | 2021-08-17 | 燕山大学 | Aerial photography automobile counting method based on OpenCv |
| CN114140827A (en) * | 2021-12-06 | 2022-03-04 | 华东师范大学 | Hole identification method and system based on U-Net network model |
| CN116309497A (en) * | 2023-03-26 | 2023-06-23 | 湖南医药学院 | Image recognition-based auxiliary analysis method for cancer cell counting and prognosis prediction |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101692125A (en) * | 2009-09-10 | 2010-04-07 | 复旦大学 | Fisher judged null space based method for decomposing mixed pixels of high-spectrum remote sensing image |
| CN101866424A (en) * | 2010-05-20 | 2010-10-20 | 复旦大学 | Hyperspectral remote sensing image mixed pixel decomposition method based on independent component analysis |
| CN102193090A (en) * | 2010-03-19 | 2011-09-21 | 复旦大学 | Mixed pixel decomposition method for remote sensing images |
| CN103604737A (en) * | 2013-12-10 | 2014-02-26 | 山东大学 | Automatic blood cell recognition device and operating method thereof |
| CN103985119A (en) * | 2014-05-08 | 2014-08-13 | 山东大学 | Method for partitioning cytoplasm and cell nucleuses of white blood cells in color blood cell image |
| CN106056118A (en) * | 2016-06-12 | 2016-10-26 | 合肥工业大学 | Recognition and counting method for cells |
| EP3089074A1 (en) * | 2015-04-22 | 2016-11-02 | The Boeing Company | Hyperspectral demixing using foveated compressive projections |
-
2017
- 2017-02-13 CN CN201710076460.6A patent/CN106875413B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101692125A (en) * | 2009-09-10 | 2010-04-07 | 复旦大学 | Fisher judged null space based method for decomposing mixed pixels of high-spectrum remote sensing image |
| CN102193090A (en) * | 2010-03-19 | 2011-09-21 | 复旦大学 | Mixed pixel decomposition method for remote sensing images |
| CN101866424A (en) * | 2010-05-20 | 2010-10-20 | 复旦大学 | Hyperspectral remote sensing image mixed pixel decomposition method based on independent component analysis |
| CN103604737A (en) * | 2013-12-10 | 2014-02-26 | 山东大学 | Automatic blood cell recognition device and operating method thereof |
| CN103985119A (en) * | 2014-05-08 | 2014-08-13 | 山东大学 | Method for partitioning cytoplasm and cell nucleuses of white blood cells in color blood cell image |
| EP3089074A1 (en) * | 2015-04-22 | 2016-11-02 | The Boeing Company | Hyperspectral demixing using foveated compressive projections |
| CN106056118A (en) * | 2016-06-12 | 2016-10-26 | 合肥工业大学 | Recognition and counting method for cells |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110110709A (en) * | 2019-05-27 | 2019-08-09 | 湖南开启时代智能科技有限公司 | A kind of red white corpuscle differential counting method, system and equipment based on image procossing |
| CN110956612A (en) * | 2019-10-31 | 2020-04-03 | 华侨大学 | Cell rapid counting method, computer equipment and readable storage medium |
| CN110956612B (en) * | 2019-10-31 | 2023-03-31 | 华侨大学 | Cell rapid counting method, computer equipment and readable storage medium |
| CN113269745A (en) * | 2021-05-21 | 2021-08-17 | 燕山大学 | Aerial photography automobile counting method based on OpenCv |
| CN113222969A (en) * | 2021-05-28 | 2021-08-06 | 大连海事大学 | Imaging flow cytometry adhesion counting and activity detection method and device based on bidirectional background difference method |
| CN113222969B (en) * | 2021-05-28 | 2024-05-14 | 大连海事大学 | Imaging flow type cell adhesion counting and activity detecting method and device based on bidirectional background difference method |
| CN114140827A (en) * | 2021-12-06 | 2022-03-04 | 华东师范大学 | Hole identification method and system based on U-Net network model |
| CN116309497A (en) * | 2023-03-26 | 2023-06-23 | 湖南医药学院 | Image recognition-based auxiliary analysis method for cancer cell counting and prognosis prediction |
| CN116309497B (en) * | 2023-03-26 | 2023-10-03 | 湖南医药学院 | Image recognition-based auxiliary analysis method for cancer cell counting and prognosis prediction |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106875413B (en) | 2019-05-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106875413B (en) | A kind of adhesion red blood cell automatic counting method based on high light spectrum image-forming | |
| CN110428432B (en) | Deep neural network algorithm for automatically segmenting colon gland image | |
| JP7198577B2 (en) | Image analysis method, device, program, and method for manufacturing trained deep learning algorithm | |
| KR101989202B1 (en) | Method and software for analysing microbial growth | |
| CN111798416B (en) | Intelligent glomerulus detection method and system based on pathological image and deep learning | |
| CN110084150B (en) | Automatic white blood cell classification method and system based on deep learning | |
| WO2021139258A1 (en) | Image recognition based cell recognition and counting method and apparatus, and computer device | |
| Wang et al. | Assisted diagnosis of cervical intraepithelial neoplasia (CIN) | |
| CN106709958A (en) | Gray scale gradient and color histogram-based image quality evaluation method | |
| KR20080016847A (en) | Methods of chromogen separation-based image analysis | |
| JP2006520972A (en) | Image pattern recognition system and method | |
| WO2009125678A1 (en) | Method and device for dividing area of image of particle in urine | |
| CN110807754B (en) | Fungus microscopic image segmentation detection method and system based on deep semantic segmentation | |
| CN113129281B (en) | Wheat stem section parameter detection method based on deep learning | |
| CN115170518A (en) | Cell detection method and system based on deep learning and machine vision | |
| CN108830874A (en) | A kind of number pathology full slice Image blank region automatic division method | |
| CN113609984A (en) | Pointer instrument reading identification method and device and electronic equipment | |
| CN115909006A (en) | Mammary tissue image classification method and system based on convolution Transformer | |
| CN111489369B (en) | Helicobacter pylori positioning method and device and electronic equipment | |
| CN110348533A (en) | A kind of planktonic organism partial size spectrum detection method based on SVM | |
| CN103927544A (en) | Machine vision grading method for ginned cotton rolling quality | |
| CN113869124A (en) | Deep learning-based blood cell morphology classification method and system | |
| Khong et al. | A novel approach for efficient implementation of nucleus detection and segmentation using correlated dual color space | |
| CN113192089A (en) | Bidirectional cross-connected convolutional neural network for image segmentation | |
| CN113723441B (en) | Intelligent analysis system and method for lip gland pathology |
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 | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210823 Address after: 215332 room 9, No. 18, Jinxing Road, Huaqiao Town, Kunshan City, Suzhou City, Jiangsu Province Patentee after: JIANGSU HUACHUANG HIGH-TECH MEDICAL TECHNOLOGY Co.,Ltd. Address before: 200241 No. 500, Dongchuan Road, Shanghai, Minhang District Patentee before: EAST CHINA NORMAL University |
|
| TR01 | Transfer of patent right |