CN106960444A - The automatic testing method of coccus in a kind of leukorrhea based on Hopfield neutral nets - Google Patents
The automatic testing method of coccus in a kind of leukorrhea based on Hopfield neutral nets Download PDFInfo
- Publication number
- CN106960444A CN106960444A CN201710209693.9A CN201710209693A CN106960444A CN 106960444 A CN106960444 A CN 106960444A CN 201710209693 A CN201710209693 A CN 201710209693A CN 106960444 A CN106960444 A CN 106960444A
- Authority
- CN
- China
- Prior art keywords
- coccus
- threshold value
- image
- leukorrhea
- gray
- 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.)
- Pending
Links
Landscapes
- Image Processing (AREA)
- Investigating Or Analysing Biological Materials (AREA)
Abstract
The invention discloses a kind of automatic testing method of coccus in leukorrhea based on Hopfield neutral nets, comprise the following steps, step one:Extract the gray level image of the standard coccus in existing leukorrhea micro-image;Threshold value is tried to achieve with OSTU algorithms;Each pixel gray value of image is compared with threshold value, 1,1 two values matrix of standard one by one is obtained;Create a Hopfield neutral net;The matrix data is inputted into the network, runs to network poised state, obtains data network weights and threshold value and output vector;Step 2:Gray level image micro- to salt solution leukorrhea obtains bianry image, and carry out connected domain demarcation, threshold value is tried to achieve with OSTU algorithms, a two values matrix is obtained by the way that each pixel gray value of the small figure of gray scale is compared with threshold value, gained two values matrix is inputted into Hopfield neutral nets, a result is exported during operation poised state.
Description
Technical field
The invention belongs to biological cell Processing Technology in Microscopic Images field, the method for use is a kind of refreshing based on Hopfield
The automatic testing method of coccus in leukorrhea through network.
Background technology
For Accurate Diagnosis gynecological disease, leukorrhea routine inspection is an essential inspection.Nowadays, leukorrhea is conventional
Judge that vagina cleanness degree is mainly dissolved the leukorrhea of patient by this method of cervical arthroplasty, i.e. clinical laboratory medical worker in inspection
Leukorrhea solution is uniformly applied on slide in physiological saline, then with cotton swab, with having in micro- sem observation salt solution leukorrhea
Formed and divide to carry out cleannes judgement, and then determine patient with the presence or absence of gynecological diseases such as vaginitis.Coccus is influence cleannes
An important indicator because coccus volume smallest number is more, easily judges by accident, manually can not accurately judge with fat globule and bacillus etc.
Its number, and the reason such as subjective factor and experience deficiency due to medical worker, it is possible to cause cleannes to judge by accident.And
By digital image processing techniques, the quantity of coccus can be accurately judged, judge to provide practical foundation for cleannes.
The content of the invention
The invention provides the method that the coccus in a kind of micro-image to salt solution leukorrhea carries out automatic detection, overcome
Medical worker can not judge the specific number of coccus and the defect easily judged by accident by Traditional Man microscopy, substantially increase medical matters people
The diagnosis accuracy of member.
The technical solution adopted by the present invention is a kind of automatic detection of coccus in leukorrhea based on Hopfield neutral nets
Method, this method includes training step and identification step, and training step is as follows:
Step 1:Manually extract the width of gray level image 100 of the standard coccus in existing leukorrhea micro-image, unified scaling
To 15*15 size;
Step 2:Threshold value is tried to achieve with OSTU algorithms respectively to 100 width coccus gray level images.
Step 3:Each pixel gray value in figure and the threshold value of the figure are compared, if more than threshold value, by the pixel
The gray value of point is set to 1, if less than threshold value, the gray value of the pixel is set into -1, obtain a standard -1,1 two
Value matrix.
Step 4:100 width figures are all taken steps 3 operation, obtain -1,1 two values matrix of 100 standards, will be all
Two values matrix preserves into training sample T, T=[A1;A2;A3;…;A100] ' (A1, A2, A3 ... A100 are gained two-value square
Battle array);
Step 5:A Hopfield neutral net is created, T is inputted into the network;
Step 6:Operational network preserves network weight and threshold value and output vector Y, made for cognitive phase to poised state
With training terminates.
Detecting step is as follows:
Step 1:Gray level image micro- to salt solution leukorrhea carries out bottom cap conversion;
Step 2:Image after being converted to bottom cap tries to achieve threshold value with OSTU algorithms, and binary map is obtained further according to Threshold segmentation
Picture;
Step 3:Connected domain demarcation is carried out to gained bianry image;
Step 4:Each connected domain is screened according to coccus morphological feature (area, girth, circularity), reservation meets
The boundary rectangle coordinate of the connected domain of coccus morphological feature;
Step 5:According to the coordinate retained in step 3, the gray scale of correspondence boundary rectangle is small under being cut in former gray level image
Figure;
Step 6:By the small figure scaling of gray scale to one size (15*15) of standard coccus image, tried to achieve with OSTU algorithms
Threshold value.Each pixel gray value of the small figure of gray scale is compared with threshold value, if gray value is more than the threshold value of the figure, its gray value put
For 1, if gray value is less than the threshold value of the figure, its gray value is set to -1, a two values matrix is obtained;
Step 7:The two values matrix that step 6 is obtained is as the input of Hopfield neutral nets, and Y is characterized storehouse, sends into
Network;
Step 8:When the network operation reaches poised state, output result;
Step 9:Interpretation of result, determines whether coccus, retains the region for being identified as coccus.
Brief description of the drawings
Fig. 1 is the system flow chart of the invention based on the coccus in Hopfield neutral net automatic detection leukorrhea.
Embodiment
Below in conjunction with the accompanying drawings, to proposed by the present invention automatic based on coccus in Hopfield neutral net automatic detection leukorrhea
Detection method is described in detail:
Step 1:Gather the micro- gray level image of salt solution leukorrhea;
Step 2:Manually intercept the width of gray level image 100 of the standard coccus in existing leukorrhea micro-image, unified scaling
To 15*15 size;
Step 3:Threshold value is tried to achieve with OSTU algorithms to 100 width coccus gray level images respectively;
Step 4:Each pixel gray value in figure and the threshold value of the figure are compared, if more than threshold value, by the pixel
The gray value of point is set to 1, if less than threshold value, the gray value of the pixel is set into -1, obtain a standard -1,1 two
Value matrix;
Step 5:100 width figures are all taken steps 4 operation, obtain -1,1 two values matrix of 100 standards, will be all
Two values matrix preserves into training sample T, T=[A1;A2;A3;…;A100] ' (A1, A2, A3 ... A100 are gained two-value square
Battle array);
Step 6:A Hopfield neutral net is created, T is inputted into the network;
Step 7:Operational network preserves network weight and threshold value and output vector Y, made for cognitive phase to poised state
With training terminates;
Step 8:Gray level image micro- to salt solution leukorrhea to be detected carries out bottom cap conversion;
Step 8-1:Expansion process is carried out to gray level image for 3 disc template with radius;
Step 8-2:Corrosion treatment is carried out to the image after expansion for 3 disc template with radius;
Step 8-3:Former gray level image is subtracted with the image after dilation erosion, the image of cap conversion on earth is obtained.
Step 9:Image after being converted to bottom cap tries to achieve threshold value with OSTU algorithms, and binary map is obtained further according to Threshold segmentation
Picture;
Step 9-1:Image after being converted to bottom cap tries to achieve threshold value with OSTU algorithms;
Step 9-2:Each pixel gray value in figure is compared with threshold value, if more than threshold value, by the pixel gray value
255 are set to, if less than threshold value, the pixel gray value being set into 0, bianry image is obtained.
Step 10:Connected domain demarcation is carried out to gained bianry image;
Step 11:Area, girth and the circularity of connected domain are calculated, each connected domain is carried out according to coccus morphological feature
Screening, the boundary rectangle coordinate of the connected domain of aperture syncoccus morphological feature;
Step 11-1:Calculate the area and girth of each connected domain;
Step 11-2:Circularity is calculated, circularity computing formula is:
Wherein, C is circularity, and S is the area of connected region, and L is the girth of connected region;
Step 11-3:Screened by area, Retention area scope is the boundary rectangle coordinate of 5~50 connected domain;
Step 11-4:Screened by girth, retain boundary rectangle coordinate of the peripheral extent for 10~150 connected domain;
Step 11-5:Screened by circularity, retain boundary rectangle coordinate of the circularity scope for 0.8~1 connected domain;
Step 12:According to the coordinate retained in step 11, the gray scale of correspondence boundary rectangle under being cut in former gray level image
Small figure;
Step 13:By the small figure scaling of gray scale to one size (15*15) of standard coccus image, tried to achieve with OSTU algorithms
Threshold value.Each pixel gray value of the small figure of gray scale is compared with threshold value, if gray value is more than the threshold value of the figure, its gray value put
For 1, if gray value is less than the threshold value of the figure, its gray value is set to -1, a two values matrix is obtained;
Step 14:The two values matrix that step 6 is obtained is as the input of Hopfield neutral nets, and Y is characterized storehouse, sends into
Network;
Step 15:When the network operation reaches poised state, output result Y ';
Step 16:Interpretation of result, determines whether coccus, retains the region for being identified as coccus;
Step 16 is concretely comprised the following steps:
Step 16-1:Result of calculation Y ' and the characteristic vector Y of feature database Plays coccus error rate P;Error rate is calculated
Formula is as follows:
(i=1,2,3 ... 15;J=1,2,3 ... is 15)
Step 16-2:If error rate is less than 0.005, the input sample is coccus;, should if error rate is more than 0.005
Input sample is impurity.Retain the region for being identified as coccus.
As described above, the present invention just can be realized well.
Claims (7)
1. the automatic testing method of coccus in a kind of leukorrhea based on Hopfield neutral nets, it is characterised in that including training
Step and identification step.
2. the automatic testing method of coccus in a kind of leukorrhea based on Hopfield neutral nets according to claim 1,
It is characterized in that the training step detailed process is:
Step 1:The width of gray level image 100 of the standard coccus in existing leukorrhea micro-image is manually extracted, 15* is uniformly zoomed to
15 size;
Step 2:Threshold value is tried to achieve with OSTU algorithms respectively to 100 width coccus gray level images;
Step 3:Each pixel gray value in figure and the threshold value of the figure are compared, if more than threshold value, by the pixel
Gray value is set to 1, if less than threshold value, the gray value of the pixel being set into -1, -1,1 two-value square of a standard is obtained
Battle array;
Step 4:100 width figures are all taken steps 3 operation, -1,1 two values matrix of 100 standards is obtained, by all two-values
Matrix preserves into training sample T, T=[A1;A2;A3;…;A100] ' (A1, A2, A3 ... A100 are gained two values matrix);
Step 5:A Hopfield neutral net is created, T is inputted into the network;
Step 6:Operational network preserves network weight and threshold value and output vector Y, used for cognitive phase to poised state,
Training terminates.
3. the automatic testing method of coccus in a kind of leukorrhea based on Hopfield neutral nets according to claim 1,
It is characterized in that the identification step detailed process is:
Step 1:Gray level image micro- to salt solution leukorrhea carries out bottom cap conversion;
Step 2:Image after being converted to bottom cap tries to achieve threshold value with OSTU algorithms, and bianry image is obtained further according to Threshold segmentation;
Step 3:Connected domain demarcation is carried out to gained bianry image;
Step 4:Each connected domain is screened according to coccus morphological feature (area, girth, circularity), aperture syncoccus
The boundary rectangle coordinate of the connected domain of morphological feature;
Step 5:According to the coordinate retained in step 3, the small figure of gray scale of correspondence boundary rectangle under being cut in former gray level image;
Step 6:By the small figure scaling of gray scale to one size (15*15) of standard coccus image, try to achieve threshold value with OSTU algorithms;
Each pixel gray value of the small figure of gray scale is compared with threshold value, if gray value is more than the threshold value of the figure, its gray value is set to 1,
If gray value is less than the threshold value of the figure, its gray value is set to -1, a two values matrix is obtained;
Step 7:The two values matrix that step 6 is obtained is as the input of Hopfield neutral nets, and Y is characterized storehouse, sends into network;
Step 8:When the network operation reaches poised state, output result;
Step 9:Interpretation of result, determines whether coccus, retains the region for being identified as coccus.
4. the automatic testing method of coccus in a kind of leukorrhea based on Hopfield neutral nets according to claim 3,
Characterized in that, the step 1 in the identification step carries out the specific steps of bottom cap conversion to the micro- gray level image of salt solution leukorrhea
For:
Step 1:Expansion process is carried out to gray level image for 3 disc template with radius;
Step 2:Corrosion treatment is carried out to the image after expansion for 3 disc template with radius;
Step 3:Former gray level image is subtracted with the image after dilation erosion, the image of cap conversion on earth is obtained.
5. the automatic testing method of coccus in a kind of leukorrhea based on Hopfield neutral nets according to claim 3,
Characterized in that, the specific acquisition methods of bianry image are in step 2 in the identification step:
Step 1:Image after being converted to bottom cap tries to achieve threshold value with OSTU algorithms;
Step 2:Each pixel gray value in figure is compared with threshold value, if more than threshold value, the pixel gray value is set to
255, if less than threshold value, the pixel gray value being set into 0, bianry image is obtained.
6. the automatic testing method of coccus in a kind of leukorrhea based on Hopfield neutral nets according to claim 3,
Characterized in that, meeting the tool of the boundary rectangle coordinate of the connected domain of coccus morphological feature in step 4 in the identification step
Body acquisition methods are:
Step 1:Calculate the area and girth of each connected domain;
Step 2:Circularity is calculated, circularity computing formula is:
Wherein, C is circularity, and S is the area of connected region, and L is the girth of connected region;
Step 3:Screened by area, Retention area scope is the boundary rectangle coordinate of 5~50 connected domain;
Step 4:Screened by girth, retain boundary rectangle coordinate of the peripheral extent for 10~150 connected domain;
Step 5:Screened by circularity, retain boundary rectangle coordinate of the circularity scope for 0.8~1 connected domain.
7. the automatic testing method of coccus in a kind of leukorrhea based on Hopfield neutral nets according to claim 3,
Characterized in that, determining whether coccus in step 9 in the identification step, retain the specific side in the region for being identified as coccus
Method is:
Step 1:Output result Y ' during system operation poised state and the characteristic vector Y of feature database Plays coccus are missed
Rate P is analyzed;Error rate computing formula is as follows:
(i=1,2,3 ... 15;J=1,2,3 ... is 15)
Step 2:If error rate is less than 0.005, the input sample is coccus;If error rate is more than 0.005, the input sample
For impurity.Retain the region for being identified as coccus.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710209693.9A CN106960444A (en) | 2017-03-31 | 2017-03-31 | The automatic testing method of coccus in a kind of leukorrhea based on Hopfield neutral nets |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710209693.9A CN106960444A (en) | 2017-03-31 | 2017-03-31 | The automatic testing method of coccus in a kind of leukorrhea based on Hopfield neutral nets |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106960444A true CN106960444A (en) | 2017-07-18 |
Family
ID=59483167
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710209693.9A Pending CN106960444A (en) | 2017-03-31 | 2017-03-31 | The automatic testing method of coccus in a kind of leukorrhea based on Hopfield neutral nets |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106960444A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115547412A (en) * | 2022-11-09 | 2022-12-30 | 内蒙古大学 | Method and device for evaluating cell differentiation potential based on Hopfield network |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104881679A (en) * | 2015-05-22 | 2015-09-02 | 电子科技大学 | Method for automatically detecting erythrocytes in leukorrhea based on improved fuzzy recognition |
CN106056118A (en) * | 2016-06-12 | 2016-10-26 | 合肥工业大学 | Recognition and counting method for cells |
CN106295715A (en) * | 2016-08-22 | 2017-01-04 | 电子科技大学 | A kind of leucorrhea cleannes automatic classification method based on BP neural network classifier |
CN106295588A (en) * | 2016-08-17 | 2017-01-04 | 电子科技大学 | The automatic identifying method of leukocyte in a kind of leucorrhea micro-image |
-
2017
- 2017-03-31 CN CN201710209693.9A patent/CN106960444A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104881679A (en) * | 2015-05-22 | 2015-09-02 | 电子科技大学 | Method for automatically detecting erythrocytes in leukorrhea based on improved fuzzy recognition |
CN106056118A (en) * | 2016-06-12 | 2016-10-26 | 合肥工业大学 | Recognition and counting method for cells |
CN106295588A (en) * | 2016-08-17 | 2017-01-04 | 电子科技大学 | The automatic identifying method of leukocyte in a kind of leucorrhea micro-image |
CN106295715A (en) * | 2016-08-22 | 2017-01-04 | 电子科技大学 | A kind of leucorrhea cleannes automatic classification method based on BP neural network classifier |
Non-Patent Citations (1)
Title |
---|
尹敏 等: "" Hopfield神经网络在字符识别中的应用"", 《电脑知识与技术》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115547412A (en) * | 2022-11-09 | 2022-12-30 | 内蒙古大学 | Method and device for evaluating cell differentiation potential based on Hopfield network |
CN115547412B (en) * | 2022-11-09 | 2024-02-02 | 内蒙古大学 | Method and device for evaluating cell differentiation potential based on Hopfield network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108764286B (en) | Classification and identification method of feature points in blood vessel image based on transfer learning | |
CN109190567A (en) | Abnormal cervical cells automatic testing method based on depth convolutional neural networks | |
CN107330263A (en) | A kind of method of area of computer aided breast invasive ductal carcinoma histological grading | |
CN104881679B (en) | Red blood cell automatic testing method in a kind of leukorrhea based on improvement fuzzy diagnosis | |
CN108717554A (en) | A kind of thyroid tumors histopathologic slide image classification method and its device | |
CN108961208A (en) | A kind of aggregation leucocyte segmentation number system and method | |
CN103218797B (en) | The method and system of blood-vessel image treatment and analyses | |
CN109272492A (en) | A kind of processing method and system of cell pathology smear | |
CN106875404A (en) | The intelligent identification Method of epithelial cell in a kind of leukorrhea micro-image | |
CN110135271A (en) | A kind of cell sorting method and device | |
CN104846054A (en) | Method for automatically detecting moulds in leucorrhea based on morphological characteristics | |
CN109389129A (en) | A kind of image processing method, electronic equipment and storage medium | |
CN110276763A (en) | It is a kind of that drawing generating method is divided based on the retinal vessel of confidence level and deep learning | |
CN111862020B (en) | Method and device for predicting physiological age of anterior ocular segment, server and storage medium | |
CN107507197A (en) | A kind of pulmonary parenchyma extracting method based on clustering algorithm and convolutional neural networks | |
CN105069803A (en) | Classifier for micro-angioma of diabetes lesion based on colored image | |
CN110473166A (en) | A kind of urinary formed element recognition methods based on improvement Alexnet model | |
CN107871316A (en) | A kind of X-ray hand bone interest region extraction method based on deep neural network | |
CN108961242A (en) | A kind of fluorescent staining image CTC intelligent identification Method | |
CN111950812B (en) | Method and device for automatically identifying and predicting rainfall | |
CN110246579B (en) | Pathological diagnosis method and device | |
CN104299242A (en) | Fluorescence angiography fundus image extraction method based on NGC-ACM | |
CN110929728A (en) | Image region-of-interest dividing method, image segmentation method and device | |
CN108229584A (en) | A kind of Multimodal medical image recognition methods and device based on deep learning | |
CN108052886A (en) | A kind of puccinia striiformis uredospore programming count method of counting |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170718 |
|
WD01 | Invention patent application deemed withdrawn after publication |