CN104463184A - Statistical method of cell numbers in statistical area static picture or dynamic video - Google Patents
Statistical method of cell numbers in statistical area static picture or dynamic video Download PDFInfo
- Publication number
- CN104463184A CN104463184A CN201310419776.2A CN201310419776A CN104463184A CN 104463184 A CN104463184 A CN 104463184A CN 201310419776 A CN201310419776 A CN 201310419776A CN 104463184 A CN104463184 A CN 104463184A
- Authority
- CN
- China
- Prior art keywords
- image
- sorter
- black
- threshold values
- sorting parameter
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
Abstract
The invention is suitable for an image cell number recognition statistical method, in which an openCV is used as a model base and tree form character classification is used as a recognition basis, so as to calculate cell numbers of image data in an area collected by a device, to collect data to generate XML files, and to issue information in a multi-interface mode by utilizing the XML files covering information of cell numbers of all sites. The identification model has a relatively higher recognition rate of cells under various environments, operates faster and reaches a level of practical application.
Description
Technical field
The present invention relates to a kind of mode identification technology, specifically relates to a kind of cell quantity added up by computing machine or other equipment in the static images in certain region or dynamic video, and carries out the figure added up and counting statistics method thereof.
Background technology
The scientific research technology in the current world is increasingly flourishing, becomes feasible and important gradually to the study and utilization of microorganism.In the research of microorganism, often need to carry out statistical work to the quantity of cell, simply by virtue of naked eyes identification cell and statistical magnitude becomes the work of a time and effort consuming, greatly reduce Efficiency.The present invention, by being added up the cell quantity in the static images in certain region or dynamic video by computing machine or other equipment, can be greatly reduced the unnecessary work of people, make scientific research become comparatively efficient.
Summary of the invention
The object of the invention is to overcome deficiency of the prior art, provide one.
In order to solve the problems of the technologies described above, the present invention is achieved through the following technical solutions:
The present invention is a kind of, and to add up the principle of work of the statistical method of the cell quantity in the static images in certain region or dynamic video as follows:
A kind of image cell recognition statistical method, is characterized in that it comprises the following steps:
A () sets up cascade classifier: build artificial neural network by the computing machine vision storehouse openCV that increases income, set about, train cascade classifier from several samples:
The training step of this cascade classifier is:
Sorter is divided into some levels, every one-level sets a kind of method of partitioned image, with black and white region representation, calculate the integration differential of black and white area pixel value, the sorting parameter of this grade of sorter is calculated according to this integration differential, comprise total threshold values and two, left and right branch value, the sorting parameter of some grades, through combination, forms cascade classifier;
Gather several samples pictures, the geometric center of samples pictures object feature and cell body is overlapped, by the sample image after process, be divided into black and white two regions, according to the pixel in white portion and the pixel in black region, calculate integration differential, according to this difference, calculate the sorting parameter of sorter at different levels, comprise total threshold values and two branch values, wherein two branch values are the divisions of total threshold values, and every first-level class device is different for the careful degree of division in black and white region;
B () adopts the window onesize with training sample to divide view picture image to be identified, whenever marking off a part of image, namely according to the computing method of sample sample, classification calculates the sorting parameter of this image, total threshold values in the sorting parameter of respective stages in the total threshold values calculated and sorter is compared, thus selects left or right branch value;
C () enters next stage screening, when not meeting the parameter in sorter when the sorting parameter calculated, represent this image not containing object feature, now stop calculating, partition window is moved to next position, repeat step b-c, when the level all by sorter of the image in partition window time, the image in this partition window contains object feature.
In statistic processes of the present invention, if there is client to propose inquiry request, then inquire about, corresponding information is issued requesting party.
Samples pictures of the present invention comprises cells intact image, and this picture is through artificial treatment, and intercepting is got off, uniform sizes, and after intercepting, the cluster center of each characteristic overlaps, and typing neural network learning after process, the picture size after process is 25cm*25cm.
Compared with prior art, the invention has the beneficial effects as follows:
The present invention is a kind of adds up the statistical method of the cell quantity in the static images in certain region or dynamic video, employs Haar feature recognition mode.For cell number identification, this pattern is more applicable, relatively other two kinds of patterns, and setting up of recognition mode is more convenient.
Accompanying drawing explanation
Fig. 1 is the process flow diagram of cell quantity statistical method in a kind of statistical regions static images or dynamic video.
Embodiment
Generally acknowledge that good three kinds of feature recognition modes are: SIFT/SURF, Haar feature, generalized h ough transform characteristics at present in the world.
Three models is all based on strength information, is all characterization method.
The feature of SIFT/SURF is a kind of feature with strong directivity and brightness, and this makes it be applicable to rigidity deformation, slightly has an X-rayed the occasion of deformation;
Haar characteristic recognition method is with the meaning of some artificial intelligence, and have significantly for as cell is this, the object of the Haar feature of rock-steady structure is the most applicable, as long as even if structure fixes non-linear deformation still identifiable design such as being distorted relatively;
Generalized h ough conversion is accurate coupling completely, can obtain the parameter informations such as the locality of object.
Method designed by this patent, employs Haar feature recognition mode.For cell number identification, this pattern is more applicable, relatively other two kinds of patterns, and setting up of recognition mode is more convenient.
Below in conjunction with accompanying drawing and embodiment, the present invention is described in further detail:
As shown in Figure 1, a kind of image cell quantity identification statistical method, it comprises the following steps:
A () sets up cascade classifier: build artificial neural network by the computing machine vision storehouse openCV that increases income, set about, train cascade classifier from several samples:
The training step of this cascade classifier is:
Sorter is divided into some levels, every one-level sets a kind of method of partitioned image, with black and white region representation, calculate the integration differential of black and white area pixel value, the sorting parameter of this grade of sorter is calculated according to this integration differential, comprise total threshold values and two, left and right branch value, the sorting parameter of some grades, through combination, forms cascade classifier;
Gather several samples pictures, the geometric center of samples pictures object feature and cell body is overlapped, by the sample image after process, be divided into black and white two regions, according to the pixel in white portion and the pixel in black region, calculate integration differential, according to this difference, calculate the sorting parameter of sorter at different levels, comprise total threshold values and two branch values, wherein two branch values are the divisions of total threshold values, and every first-level class device is different for the careful degree of division in black and white region;
B () adopts the window onesize with training sample to divide view picture image to be identified, whenever marking off a part of image, namely according to the computing method of sample sample, classification calculates the sorting parameter of this image, total threshold values in the sorting parameter of respective stages in the total threshold values calculated and sorter is compared, thus selects left or right branch value;
C () enters next stage screening, when not meeting the parameter in sorter when the sorting parameter calculated, represent this image not containing object feature, now stop calculating, partition window is moved to next position, repeats step b-c, when the level all by sorter of the image in partition window time, illustrate that this part is identified with larger probability, the image namely in this partition window contains object feature.
In statistic processes, if there is client to propose inquiry request, then inquire about, corresponding information is issued requesting party.
Samples pictures comprises cells intact, and this picture is typing neural network learning after artificial treatment, and the picture size after process is 25cm*25cm.
The undeclared part related in the present invention is same as the prior art or adopt prior art to be realized.
Claims (3)
1. an image cell recognition statistical method, is characterized in that it comprises the following steps:
A () sets up cascade classifier: build artificial neural network by the computing machine vision storehouse openCV that increases income, set about, train cascade classifier from several samples:
The training step of this cascade classifier is:
Sorter is divided into some levels, every one-level sets a kind of method of partitioned image, with black and white region representation, calculate the integration differential of black and white area pixel value, the sorting parameter of this grade of sorter is calculated according to this integration differential, comprise total threshold values and two, left and right branch value, the sorting parameter of some grades, through combination, forms cascade classifier;
Gather several samples pictures, the geometric center of samples pictures object feature and cell body is overlapped, by the sample image after process, be divided into black and white two regions, according to the pixel in white portion and the pixel in black region, calculate integration differential, according to this difference, calculate the sorting parameter of sorter at different levels, comprise total threshold values and two branch values, wherein two branch values are the divisions of total threshold values, and every first-level class device is different for the careful degree of division in black and white region;
B () adopts the window onesize with training sample to divide view picture image to be identified, whenever marking off a part of image, namely according to the computing method of sample sample, classification calculates the sorting parameter of this image, total threshold values in the sorting parameter of respective stages in the total threshold values calculated and sorter is compared, thus selects left or right branch value;
C () enters next stage screening, when not meeting the parameter in sorter when the sorting parameter calculated, represent this image not containing object feature, now stop calculating, partition window is moved to next position, repeat step b-c, when the level all by sorter of the image in partition window time, the image in this partition window contains object feature.
2. image cell recognition statistical method according to claim 1, is characterized in that, in described statistic processes, if there is client to propose inquiry request, then inquiring about, corresponding information being issued requesting party.
3. image cell recognition statistical method according to claim 1, it is characterized in that described samples pictures comprises cells intact image, this picture is through artificial treatment, intercepting is got off, uniform sizes, after intercepting, the cluster center of each characteristic overlaps, and typing neural network learning after process, the picture size after process is 25cm*25cm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310419776.2A CN104463184A (en) | 2013-09-15 | 2013-09-15 | Statistical method of cell numbers in statistical area static picture or dynamic video |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310419776.2A CN104463184A (en) | 2013-09-15 | 2013-09-15 | Statistical method of cell numbers in statistical area static picture or dynamic video |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104463184A true CN104463184A (en) | 2015-03-25 |
Family
ID=52909198
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310419776.2A Pending CN104463184A (en) | 2013-09-15 | 2013-09-15 | Statistical method of cell numbers in statistical area static picture or dynamic video |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104463184A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106156734A (en) * | 2016-06-28 | 2016-11-23 | 浙江工业大学 | A kind of current speed-measuring method based on convolutional neural networks image recognition |
-
2013
- 2013-09-15 CN CN201310419776.2A patent/CN104463184A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106156734A (en) * | 2016-06-28 | 2016-11-23 | 浙江工业大学 | A kind of current speed-measuring method based on convolutional neural networks image recognition |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107038416B (en) | Pedestrian detection method based on binary image improved HOG characteristics | |
| CN105608456A (en) | Multi-directional text detection method based on full convolution network | |
| CN111950723A (en) | Neural network model training method, image processing method, device and terminal equipment | |
| CN109948566B (en) | Double-flow face anti-fraud detection method based on weight fusion and feature selection | |
| CN105469400A (en) | Rapid identification and marking method of electronic component polarity direction and system thereof | |
| CN104112143A (en) | Weighted hyper-sphere support vector machine algorithm based image classification method | |
| CN103745104A (en) | Examination paper marking method based on augmented reality technology | |
| CN109472193A (en) | Method for detecting human face and device | |
| CN112529090B (en) | Small target detection method based on improved YOLOv3 | |
| CN110163111A (en) | Method, apparatus of calling out the numbers, electronic equipment and storage medium based on recognition of face | |
| CN110287787B (en) | Image recognition method, image recognition device and computer-readable storage medium | |
| CN104143081A (en) | Smile recognition system and method based on mouth features | |
| CN112241679A (en) | Automatic garbage classification method | |
| CN102663401A (en) | Image characteristic extracting and describing method | |
| CN110866900A (en) | Water body color identification method and device | |
| CN112862849A (en) | Image segmentation and full convolution neural network-based field rice ear counting method | |
| CN109447111A (en) | A kind of remote sensing supervised classification method based on subclass training sample | |
| CN111008979A (en) | Robust night image semantic segmentation method | |
| CN102043963A (en) | Method for recognizing and counting number of people in image | |
| CN109800659A (en) | A kind of action identification method and device | |
| CN104537381A (en) | Blurred image identification method based on blurred invariant feature | |
| CN111027564A (en) | Low-illumination imaging license plate recognition method and device based on deep learning integration | |
| CN114332942A (en) | Night infrared pedestrian detection method and system based on improved YOLOv3 | |
| CN110135274B (en) | Face recognition-based people flow statistics method | |
| CN104463184A (en) | Statistical method of cell numbers in statistical area static picture or dynamic video |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20150325 |