CN109389139A - A kind of locust method of counting and device - Google Patents
A kind of locust method of counting and device Download PDFInfo
- Publication number
- CN109389139A CN109389139A CN201710686810.0A CN201710686810A CN109389139A CN 109389139 A CN109389139 A CN 109389139A CN 201710686810 A CN201710686810 A CN 201710686810A CN 109389139 A CN109389139 A CN 109389139A
- Authority
- CN
- China
- Prior art keywords
- region
- locust
- image
- target connected
- connected region
- 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
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides a kind of locust method of counting and device, the method comprise the steps that S1, clusters locust image using Meanshift algorithm, carries out binary conversion treatment to the locust image after cluster, obtains the target connected region in the locust image;S2 judges that each target connected region is monomer region or adhesion region, is split to the adhesion region;The number of the monomer region is added with the areal after the adhesion region segmentation, obtains the number of locust in the locust image by S3.The present invention has the automatic counting for realizing locust, and the beneficial effect that the accuracy counted is high.
Description
Technical field
The present invention relates to image processing application fields, more particularly, to a kind of locust method of counting and device.
Background technique
Locust is worldwide pest, and China migratory locusts generating region is related to 16 provinces, cities and regions at present, involves the county Jin200Ge, wherein
County, 100, recurrence area, the occurring area of summer and autumn migratory locusts is up to 2500~30,000,000 mu;Native locust generation area is related to more than 20
It saves, involves the county Jin500Ge, wherein the county of serious generating region 200, occurring area is at 200,000,000 mu or more.Locust occurrence frequency in recent years
Rise, the extent of injury aggravates.
Generation for the plague of locusts is different degrees of, and anti-locust station needs to transfer different grades of anti-locust resource, if transferred excessive
Resource it will cause the wastings of resources.Therefore, the accuracy for improving the estimation of plague of locusts degree is critically important.Generally by locust density into
The estimation of row plague of locusts degree.It in traditional method, needs manually to obtain the number of locust, locust is obtained according to locust number
Density, efficiency are very low.Currently, some methods obtain locust number by locust image, but only simply use a kind of algorithm pair
Locust image is split, and is counted according to the number of segmentation to locust, and the result precision of counting is low.
Summary of the invention
For the low efficiency for overcoming above-mentioned locust to count, the problem of accuracy the or it at least is partially solved the above problem,
The present invention provides a kind of locust method of counting and devices.
According to the first aspect of the invention, a kind of locust method of counting is provided, comprising:
S1 clusters locust image using Meanshift algorithm, carries out two-value to the locust image after cluster
Change processing, obtains each target connected region in the locust image;
S2 judges that each target connected region is monomer region or adhesion region, divides the adhesion region
It cuts;
The number of the monomer region is added with the areal after the adhesion region segmentation, obtains the locust by S3
The number of locust in worm image.
Specifically, before the step S1 further include:
Original locust image is obtained using the image acquiring device for loading filter plate;
The original locust image is pre-processed using median filtering algorithm.
Specifically, binary conversion treatment is carried out to the locust image after cluster in the step S1, obtains the locust
The step of each target connected region in image, specifically includes:
Gray level image is converted by the locust image after cluster, obtains the histogram of the gray level image;
Choose the histogram it is bimodal between the lowest point position gray value as gray threshold, according to the gray threshold
Thresholding processing is carried out to the gray level image;
According to thresholding processing as a result, obtaining each target connected region in the locust image.
Specifically, between the step S1 and S2 further include:
Corrosion or expansion process are carried out to each target connected region.
Specifically, the step of each target connected region is monomer region or adhesion region is judged in the step S2
It specifically includes:
According to the area and perimeter of each target connected region, the form factor of each target connected region is obtained;
According to the form factor of each target connected region, form factor threshold value is obtained;
According to the form factor threshold value and preset area threshold value, judge each target connected region be monomer region also
It is adhesion region.
Specifically, the shape factor S F of each target connected region is obtained by following formula:
Wherein, A indicates that the area of the target connected region, C indicate the perimeter of the target connected region.
Specifically, according to the form factor threshold value and preset area threshold value, judge that each target connected region is single
The step of body region or adhesion region, specifically includes:
When the form factor of the target connected region is greater than the form factor threshold value and the target connected region
When area is less than the preset area threshold value, the target connected region is monomer region;
When the form factor of the target connected region is less than or equal to the form factor threshold value and target connection
When the area in region is greater than or equal to the preset area threshold value, the target connected region is adhesion region.
Specifically, the step of being split in the step S2 to the adhesion region specifically includes:
S21 selects a pixel as sub-pixel, by described kind in each adhesion region from the adhesion region
The son region new as one;
S22 calculates the gray scale difference value between the sub-pixel and each neighborhood of the sub-pixel, and it is default poor to be less than
The corresponding neighborhood of the gray scale difference value of value threshold value is merged into the new region where the sub-pixel of selection as sub-pixel;
S23, iteration execute step S22, until for each sub-pixel, the sub-pixel and the sub-pixel
Gray scale difference value between each neighborhood is all larger than or is equal to the preset difference value threshold value;
S24, for remaining pixel in the adhesion region, iteration executes step S21-S23, until in the adhesion region
Each pixel is divided into new region.
According to the second aspect of the invention, a kind of locust counting device is provided, comprising:
Acquiring unit schemes the locust after cluster for being clustered using Meanshift algorithm to locust image
As carrying out binary conversion treatment, each target connected region in the locust image is obtained;
Cutting unit, for judging that each target connected region is monomer region or adhesion region, to described each viscous
Even region is split;
Counting unit, for by the areal phase after the number of the monomer region and each adhesion region segmentation
Add, obtains the number of locust in the locust image.
According to the third aspect of the invention we, a kind of non-transient computer readable storage medium is provided, for storing such as preceding institute
State the computer program of method.
The present invention provides a kind of locust method of counting and device, this method gather locust by Meanshift algorithm
Class, and binaryzation is carried out to the locust after cluster, so that target connected region is obtained, by judging each target connected region
It is monomer region or adhesion region, the adhesion region is split, after monomer region and each adhesion region segmentation
The number in region is added, and the number of locust in locust image is obtained, thus the standard realizing the automatic counting of locust, and counting
Exactness is high.The result that locust counts can be used for estimating the density of locust, and then detect to the severity of the plague of locusts, be locust
Prevention and treatment help is provided.
Detailed description of the invention
Fig. 1 is locust method of counting overall flow schematic diagram provided in an embodiment of the present invention;
Fig. 2 is locust counting device overall structure diagram provided in an embodiment of the present invention;
Fig. 3 is locust counting equipment structural schematic diagram provided in an embodiment of the present invention.
Specific embodiment
With reference to the accompanying drawings and examples, specific embodiments of the present invention will be described in further detail.Implement below
Example is not intended to limit the scope of the invention for illustrating the present invention.
A kind of locust method of counting is provided in one embodiment of the invention, and Fig. 1 is locust provided in an embodiment of the present invention
Worm method of counting overall flow schematic diagram.Generally, this method comprises: S1, carries out locust image using Meanshift algorithm
Cluster carries out binary conversion treatment to the locust image after cluster, obtains each target connected region in the locust image;
S2 judges that each target connected region is monomer region or adhesion region, is split to the adhesion region;S3, will
The number of the monomer region is added with the areal after the adhesion region segmentation, obtains locust in the locust image
Number.
Specifically, in S1, the Meanshift algorithm is a kind of iterative algorithm, i.e., first calculates the offset mean value of current point,
Then it as new starting point, continues to move to, until meeting preset constraint condition.Use the Meanshift algorithm pair
Locust image is clustered, and by color in the locust image, similar, closely located pixel combination is a region, thus will
The locust image is divided into multiple connected regions.Binary conversion treatment is carried out to the locust image after cluster, i.e., it will be described
Locust image is converted to two kinds of features of black and white, and one of feature is background, and another feature is target, to distinguish
Target and background in the locust image out, using the corresponding connected region of the target as target connected region.The mesh
It is designated as locust.In S2, judges that each target connected region is monomer region or adhesion region, include in the monomer region
Single locust includes multiple locusts in the adhesion region.The adhesion region is split, by each adhesion region segmentation
At multiple regions, each region after segmentation includes single locust.Can be used region growing algorithm to the adhesion region into
Row segmentation, but such partitioning algorithm is not limited in the present embodiment.In S3, by the number of the monomer region and each adhesion area
Areal after regional partition is added, and the sum of addition is the number of locust in the locust image.It can be to the monomer
Region behind region and shown each adhesion region segmentation is marked, and counts the number of label as locust in the locust image
Number.
The present embodiment clusters locust by Meanshift algorithm, and carries out binaryzation to the locust after cluster, from
And target connected region is obtained, by judging that each target connected region is monomer region or adhesion region, to described viscous
Even region is split, and monomer region is added with the number in the region after each adhesion region segmentation, obtains locust image
The number of middle locust, to realize the automatic counting of locust, and the accuracy counted is high.The result that locust counts can be used for estimating
The density of locust, and then the severity of the plague of locusts is detected, help is provided for the prevention and treatment of locust.
On the basis of the above embodiments, before step S1 described in the present embodiment further include: use the figure for loading filter plate
As acquisition device obtains original locust image;The original locust image is pre-processed using median filtering algorithm.
Specifically, the original locust image can obtain in field, can also obtain under experimental situation.When testing
When obtaining under environment, experimental material is got out, has configured experimental situation, sets the parameter of image acquiring device, such as described
The time for exposure of image acquiring device is 30s, and aperture is adjusted to maximum etc..Described image acquisition device can for camera, mobile phone,
Camera etc., but the present embodiment is not limited to these image acquiring devices.Described image acquisition device is mounted with filter plate, and selection is closed
The filter plate of suitable wave band, keeps the difference between background and target as big as possible, such as wave band is the filter plate of 763nm.Pass through filter
The original locust image that wave plate obtains is single band image, since uneven illumination is even, background border in the original locust image
Gray value it is close with the gray value of locust target area, the effect of subsequent processing is influenced, so scheming to the original locust
As carrying out median filtering to the gray level image of the original locust image, inhibiting local gray noise, increase before being handled
Add the accuracy of image segmentation, to improve the accuracy of locust counting.
Based on any of the above embodiments, to the locust image after cluster in step S1 described in the present embodiment
The step of carrying out binary conversion treatment, obtaining each target connected region in the locust image specifically includes: by the institute after cluster
It states locust image and is converted into gray level image, obtain the histogram of the gray level image;Choose the histogram it is bimodal between paddy
The gray value of bottom position carries out thresholding processing to the gray level image as gray threshold, according to the gray threshold;According to
Thresholding processing as a result, obtaining each target connected region in the locust image.
Specifically, since the locust image that filter plate obtains is band image, the locust image after cluster is also wave
Section image, converts gray level image for the locust image after cluster, obtains the histogram of the gray level image.Described in selection
Histogram it is bimodal between the lowest point position gray value as gray threshold.According to the gray threshold to the locust after the cluster
Worm image carries out thresholding processing, the function of the thresholding processing are as follows:
Wherein, (x, y) is the position of each pixel in the gray level image, and g (x, y) is each picture in the gray level image
Plain thresholding treated gray value, f (x, y) are the gray value of each pixel in the gray level image, and T is the gray scale threshold
Value.According to thresholding processing as a result, the prospect and target in the gray level image are distinguished, to obtain in the locust image
Target connected region.
The present embodiment is obtained in the locust image by carrying out binary conversion treatment to the locust image after cluster
Target connected region improves the accuracy of locust counting to accurately distinguish out the prospect and target in the locust image.
Based on any of the above embodiments, step S1 described in the present embodiment and between S2 further include: to described each
Target connected region carries out corrosion or expansion process.
Specifically, by carrying out corrosion or expansion process to the target connected region, tiny on image make an uproar can be eliminated
Sound, and smoothed image boundary more accurately obtain the target connected region in the locust image, to improve locust counting
Accuracy.
Based on any of the above embodiments, each target connected region is judged in step S2 described in the present embodiment
The step of being monomer region or adhesion region, specifically includes: according to the area and perimeter of each target connected region, obtaining
The form factor of each target connected region;According to the form factor of each target connected region, form factor threshold is obtained
Value;According to the form factor threshold value and preset area threshold value, judge that each target connected region is monomer region or glues
Even region.
Specifically, the area and perimeter for obtaining each target connected region, according to the face of each target connected region
Long-pending and perimeter obtains the form factor of each target connected region, and the form factor is for describing the target connected region
The shape in domain.According to the difference between the monomer region and the form factor in adhesion region, form factor threshold value is obtained.It is described
Form factor threshold value with the form factor of each connected region for being compared, to distinguish monomer region and adhesion region.It is default
Area threshold be it is preset, for being compared with the area of each connected region, to distinguish monomer region and adhesion region.
According to the form factor threshold value and the preset area threshold value, judge that each target connected region is monomer region or glues
Even region.
The present embodiment judges that each target connected region is single by the form factor threshold value and preset area threshold value
Body region or adhesion region improve the accuracy of locust counting to accurately distinguish monomer region and adhesion region.
On the basis of the above embodiments, the shape factor S F of each target connected region described in the present embodiment passes through following formula
It obtains:
Wherein, A indicates that the area of the target connected region, C indicate the perimeter of the target connected region.
On the basis of the above embodiments, sentenced in the present embodiment according to the form factor threshold value and preset area threshold value
The step of each target connected region is monomer region or adhesion region of breaking specifically includes: when the target connected region
When form factor is less than the preset area threshold value greater than the area of the form factor threshold value and the target connected region, institute
Stating target connected region is monomer region;When the form factor of the target connected region is less than or equal to the form factor threshold
When the area of value and the target connected region is greater than or equal to the preset area threshold value, the target connected region is adhesion
Region.
On the basis of the above embodiments, the step adhesion region being split in step S2 described in the present embodiment
Suddenly specifically include: the step of being split in the step S2 to the adhesion region specifically includes: S21, for each adhesion
Region selects a pixel as sub-pixel, using the seed as a new region from the adhesion region;S22 is calculated
Gray scale difference value between the sub-pixel and each neighborhood of the sub-pixel will be less than the gray scale of preset difference value threshold value
The corresponding neighborhood of difference is merged into the new region where the sub-pixel of selection as sub-pixel;S23, iteration execute step
S22, until for each sub-pixel, the gray scale difference value between the sub-pixel and each neighborhood of the sub-pixel is big
In or equal to the preset difference value threshold value;S24, for remaining pixel in the adhesion region, iteration executes step S21-S23,
Until each pixel in the adhesion region is divided into new region.
It specifically, will be in judging result after judging each target connected region for monomer region or adhesion region
Adhesion region be split.It is split in the present embodiment using region growing algorithm.Specifically, for each adhesion area
Domain selects a pixel as sub-pixel from the adhesion region.Using the sub-pixel region new as one.It calculates
Gray scale difference value between the sub-pixel and each neighborhood of the sub-pixel will be less than the gray scale of preset difference value threshold value
The corresponding neighborhood of difference is merged into the new region as sub-pixel.The neighborhood of the sub-pixel can be neighbours domain
Or eight neighborhood.The preset difference value threshold value be it is preset, for being compared with each gray scale difference value, to obtain new seed
Pixel.Iteration executes the step of obtaining sub-pixel according to gray scale difference value, until for each sub-pixel, the seed picture
Gray scale difference value between element and each neighborhood of the sub-pixel is all larger than or is equal to the preset difference value threshold value.It is obtained in each iteration
New sub-pixel is taken, when no longer obtaining new sub-pixel, stops iteration.Thus by the sub-pixel distance with selection
In region close, where the similar pixel combination to the sub-pixel of selection of feature.For remaining picture in the adhesion region
Element continues to execute above-mentioned selection seed and combined operation, until all seeds are all divided into new region.It will be described
The seed selected is as different new regions, the number in the region for selecting the number of seed as to divide.
The present embodiment is split the adhesion region in judging result by using region growing algorithm, the area after segmentation
Include single target in domain, locust is counted according to the result after segmentation, improves the accuracy of technology.
A kind of locust counting device is provided in another embodiment of the present invention, and Fig. 2 is provided in an embodiment of the present invention
Locust counting device overall structure diagram.Generally, which includes acquiring unit 1, cutting unit 2 and counting unit 3,
In:
The acquiring unit 1 is for clustering locust image using Meanshift algorithm, to the locust after cluster
Worm image carries out binary conversion treatment, obtains each target connected region in the locust image;The cutting unit 2 is for judging
Each target connected region is monomer region or adhesion region, is split to each adhesion region;The counting is single
Member 3 obtains the locust for the number of the monomer region to be added with the areal after each adhesion region segmentation
The number of locust in image.
Specifically, the Meanshift algorithm is a kind of iterative algorithm, i.e., first calculates the offset mean value of current point, then
It as new starting point, continues to move to, until meeting preset constraint condition.Described in 1 use of acquiring unit
Meanshift algorithm clusters locust image, is by similar, the closely located pixel combination of color in the locust image
One region, so that the locust image is divided into multiple connected regions.The locust after 1 pair of acquiring unit cluster
Image carries out binary conversion treatment, i.e., the locust image is converted to two kinds of features of black and white, and one of feature is back
Scape, another feature is target, so that the target and background in the locust image is distinguished, by the corresponding connection of the target
Region is as target connected region.The target is locust.The cutting unit 2 judges that each target connected region is monomer
Region or adhesion region include single locust in the monomer region, include multiple locusts in the adhesion region.Described point
It cuts unit 2 to be split the adhesion region, each region packet by each adhesion region segmentation at multiple regions, after segmentation
Containing single locust.Region growing algorithm can be used to be split the adhesion region, but be not limited in the present embodiment such
Partitioning algorithm.The counting unit 3 is by the areal phase after the number of the monomer region and each adhesion region segmentation
Add, the sum of addition is the number of locust in the locust image.It can be to the monomer region and shown each adhesion region
Region after segmentation is marked, and counts number of the number of label as locust in the locust image.
The present embodiment clusters locust by Meanshift algorithm, and carries out binaryzation to the locust after cluster, from
And target connected region is obtained, by judging that each target connected region is monomer region or adhesion region, to described viscous
Even region is split, and monomer region is added with the number in the region after each adhesion region segmentation, obtains locust image
The number of middle locust, to realize the automatic counting of locust, and the accuracy counted is high.
On the basis of the above embodiments, device described in the present embodiment further includes pretreatment unit, for using loading
The image acquiring device of filter plate obtains original locust image;The original locust image is carried out using median filtering algorithm pre-
Processing.
On the basis of the above embodiments, acquiring unit described in the present embodiment is specifically used for: by the locust after cluster
Worm image is converted into gray level image, obtains the histogram of the gray level image;Choose the histogram it is bimodal between the lowest point position
The gray value set carries out thresholding processing to the gray level image as gray threshold, according to the gray threshold;According to threshold value
Change processing as a result, obtaining the target connected region in the locust image.
On the basis of the above embodiments, device described in the present embodiment further includes optimization unit, for the target
Connected region carries out corrosion or expansion process.
On the basis of the above embodiments, cutting unit described in the present embodiment is specifically used for: being connected according to each target
The area and perimeter in logical region, obtain the form factor of each target connected region;According to each target connected region
Form factor obtains form factor threshold value;According to the form factor threshold value and preset area threshold value, judge that each target connects
Logical region is monomer region or adhesion region.
On the basis of the above embodiments, the shape factor S F of each target connected region described in the present embodiment passes through following formula
It obtains:
Wherein, A indicates that the area of the target connected region, C indicate the perimeter of the target connected region.
On the basis of the above embodiments, cutting unit described in the present embodiment is further specifically used for: when the target
The form factor of connected region is greater than the form factor threshold value and the area of the target connected region is less than the default face
When product threshold value, the target connected region is monomer region;When the form factor of the target connected region is less than or equal to institute
When stating the area of form factor threshold value and the target connected region and being greater than or equal to the preset area threshold value, the target connects
Logical region is adhesion region.
On the basis of the above embodiments, cutting unit described in the present embodiment is specifically used for: to institute in the step S2
It states the step of adhesion region is split to specifically include: for each adhesion region, a pixel is selected from the adhesion region
As sub-pixel, using the seed region new as one;Calculate each neighbour of the sub-pixel Yu the sub-pixel
The corresponding neighborhood of the gray scale difference value for being less than preset difference value threshold value is merged by the gray scale difference value between domain as sub-pixel
New region where the sub-pixel of selection;The step of iteration execution is merged according to the gray scale difference value, until for institute
Each sub-pixel is stated, the gray scale difference value between the sub-pixel and each neighborhood of the sub-pixel is all larger than or is equal to described pre-
If difference threshold;For remaining pixel in the adhesion region, iteration is held selected seed and is closed according to the gray scale difference value
And the step of, until each pixel in the adhesion region is divided into new region.
The present embodiment provides a kind of locust counting equipment, Fig. 3 is locust counting equipment structure provided in an embodiment of the present invention
Schematic diagram, the equipment include: at least one processor 31, at least one processor 32 and bus 33;Wherein,
The processor 31 and memory 32 complete mutual communication by the bus 33;
The memory 32 is stored with the program instruction that can be executed by the processor 31, and the processor calls the journey
Sequence instruction is able to carry out method provided by above-mentioned each method embodiment, for example, S1, using Meanshift algorithm to locust
Worm image is clustered, and is carried out binary conversion treatment to the locust image after cluster, is obtained each mesh in the locust image
Mark connected region;S2 judges that each target connected region is monomer region or adhesion region, carries out to the adhesion region
Segmentation;The number of the monomer region is added by S3 with the areal after the adhesion region segmentation, obtains the locust figure
The number of locust as in.
The present embodiment provides a kind of non-transient computer readable storage medium, the non-transient computer readable storage medium
Computer instruction is stored, the computer instruction makes the computer execute method provided by above-mentioned each method embodiment, example
Such as include: S1, locust image is clustered using Meanshift algorithm, two-value is carried out to the locust image after cluster
Change processing, obtains each target connected region in the locust image;S2 judges that each target connected region is monomer region
Or adhesion region, is split the adhesion region;S3, by the number of the monomer region and the adhesion region segmentation
Areal afterwards is added, and obtains the number of locust in the locust image.
Those of ordinary skill in the art will appreciate that: realize that all or part of the steps of above method embodiment can pass through
The relevant hardware of program instruction is completed, and program above-mentioned can be stored in a computer readable storage medium, the program
When being executed, step including the steps of the foregoing method embodiments is executed;And storage medium above-mentioned includes: ROM, RAM, magnetic disk or light
The various media that can store program code such as disk.
Locust counting equipment embodiment described above is only schematical, wherein described be used as separate part description
Unit may or may not be physically separated, component shown as a unit may or may not be
Physical unit, it can it is in one place, or may be distributed over multiple network units.It can be according to the actual needs
Some or all of the modules therein is selected to achieve the purpose of the solution of this embodiment.Those of ordinary skill in the art are not paying
In the case where creative labor, it can understand and implement.
Through the above description of the embodiments, those skilled in the art can be understood that each embodiment can
It realizes by means of software and necessary general hardware platform, naturally it is also possible to pass through hardware.Based on this understanding, on
Stating technical solution, substantially the part that contributes to existing technology can be embodied in the form of software products in other words, should
Computer software product may be stored in a computer readable storage medium, such as ROM/RAM, magnetic disk, CD, including several fingers
It enables and using so that a computer equipment (can be personal computer, server or the network equipment etc.) executes each implementation
Method described in certain parts of example or embodiment.
Finally, the present processes are only preferable embodiment, it is not intended to limit the scope of the present invention.It is all
Within the spirit and principles in the present invention, any modification, equivalent replacement, improvement and so on should be included in protection of the invention
Within the scope of.
Claims (10)
1. a kind of locust method of counting characterized by comprising
S1 clusters locust image using Meanshift algorithm, carries out at binaryzation to the locust image after cluster
Reason, obtains each target connected region in the locust image;
S2 judges that each target connected region is monomer region or adhesion region, is split to the adhesion region;
The number of the monomer region is added by S3 with the areal after the adhesion region segmentation, obtains the locust figure
The number of locust as in.
2. locust method of counting according to claim 1, which is characterized in that before the step S1 further include:
Original locust image is obtained using the image acquiring device for loading filter plate;
The original locust image is pre-processed using median filtering algorithm.
3. locust method of counting according to claim 1 or 2, which is characterized in that the institute after cluster in the step S1
The step of locust image carries out binary conversion treatment, obtains each target connected region in the locust image is stated to specifically include:
Gray level image is converted by the locust image after cluster, obtains the histogram of the gray level image;
Choose the histogram it is bimodal between the lowest point position gray value as gray threshold, according to the gray threshold to institute
It states gray level image and carries out thresholding processing;
According to thresholding processing as a result, obtaining each target connected region in the locust image.
4. locust method of counting according to claim 1 or 2, which is characterized in that between the step S1 and S2 further include:
Corrosion or expansion process are carried out to each target connected region.
5. locust method of counting according to claim 1 or 2, which is characterized in that judge each mesh in the step S2
The step of mark connected region is monomer region or adhesion region specifically includes:
According to the area and perimeter of each target connected region, the form factor of each target connected region is obtained;
According to the form factor of each target connected region, form factor threshold value is obtained;
According to the form factor threshold value and preset area threshold value, judge that each target connected region is monomer region or glues
Even region.
6. locust method of counting according to claim 5, which is characterized in that the form factor of each target connected region
SF is obtained by following formula:
Wherein, A indicates that the area of the target connected region, C indicate the perimeter of the target connected region.
7. locust method of counting according to claim 5, which is characterized in that according to the form factor threshold value and default face
Product threshold value, judges that the step of each target connected region is monomer region or adhesion region specifically includes:
When the form factor of the target connected region is greater than the area of the form factor threshold value and the target connected region
When less than the preset area threshold value, the target connected region is monomer region;
When the form factor of the target connected region is less than or equal to the form factor threshold value and the target connected region
Area be greater than or equal to the preset area threshold value when, the target connected region be adhesion region.
8. locust method of counting according to claim 1 or 2, which is characterized in that the adhesion area in the step S2
The step of domain is split specifically includes:
S21 selects a pixel as sub-pixel from the adhesion region, the seed is made for each adhesion region
For a new region;
S22 calculates the gray scale difference value between the sub-pixel and each neighborhood of the sub-pixel, will be less than preset difference value threshold
The corresponding neighborhood of the gray scale difference value of value is merged into the new region where the sub-pixel of selection as sub-pixel;
S23, iteration executes step S22, until for each sub-pixel, each neighbour of the sub-pixel and the sub-pixel
Gray scale difference value between domain is all larger than or is equal to the preset difference value threshold value;
S24, for remaining pixel in the adhesion region, iteration executes step S21-S23, until each of the adhesion region
Pixel is all divided into new region.
9. a kind of locust counting device characterized by comprising
Acquiring unit, for being clustered using Meanshift algorithm to locust image, to the locust image after cluster into
Row binary conversion treatment obtains the target connected region in the locust image;
Cutting unit, for judging that each target connected region is monomer region or adhesion region, to each adhesion area
Domain is split;
Counting unit is obtained for the number of the monomer region to be added with the areal after each adhesion region segmentation
Take the number of locust in the locust image.
10. a kind of non-transient computer readable storage medium, which is characterized in that the non-transient computer readable storage medium is deposited
Computer instruction is stored up, the computer instruction makes the computer execute method as described in any of the claims 1 to 8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710686810.0A CN109389139A (en) | 2017-08-11 | 2017-08-11 | A kind of locust method of counting and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710686810.0A CN109389139A (en) | 2017-08-11 | 2017-08-11 | A kind of locust method of counting and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109389139A true CN109389139A (en) | 2019-02-26 |
Family
ID=65413948
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710686810.0A Pending CN109389139A (en) | 2017-08-11 | 2017-08-11 | A kind of locust method of counting and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109389139A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111178354A (en) * | 2019-12-23 | 2020-05-19 | 深圳市铁汉生态环境股份有限公司 | Mangrove pest monitoring method and system |
CN111209824A (en) * | 2019-12-30 | 2020-05-29 | 江苏省血吸虫病防治研究所 | Method for monitoring ampullaria gigas by using unmanned aerial vehicle |
CN111696127A (en) * | 2020-06-02 | 2020-09-22 | 中国联合网络通信集团有限公司 | Locust plague analysis and early warning system and method |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101777124A (en) * | 2010-01-29 | 2010-07-14 | 北京新岸线网络技术有限公司 | Method for extracting video text message and device thereof |
JP2011212307A (en) * | 2010-03-31 | 2011-10-27 | Shiseido Co Ltd | Cosmetic-applied skin evaluation device, cosmetic-applied skin evaluation method, and cosmetic-applied skin evaluation program |
CN104484877A (en) * | 2014-12-12 | 2015-04-01 | 山东大学 | AML cell segmentation method based on Meanshift cluster and morphological operations |
CN105095958A (en) * | 2014-05-12 | 2015-11-25 | 浙江理工大学 | Cocoon counting method |
CN106447672A (en) * | 2016-09-14 | 2017-02-22 | 燕山大学 | Identifying and counting method for intersecting and shielded fish |
-
2017
- 2017-08-11 CN CN201710686810.0A patent/CN109389139A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101777124A (en) * | 2010-01-29 | 2010-07-14 | 北京新岸线网络技术有限公司 | Method for extracting video text message and device thereof |
JP2011212307A (en) * | 2010-03-31 | 2011-10-27 | Shiseido Co Ltd | Cosmetic-applied skin evaluation device, cosmetic-applied skin evaluation method, and cosmetic-applied skin evaluation program |
CN105095958A (en) * | 2014-05-12 | 2015-11-25 | 浙江理工大学 | Cocoon counting method |
CN104484877A (en) * | 2014-12-12 | 2015-04-01 | 山东大学 | AML cell segmentation method based on Meanshift cluster and morphological operations |
CN106447672A (en) * | 2016-09-14 | 2017-02-22 | 燕山大学 | Identifying and counting method for intersecting and shielded fish |
Non-Patent Citations (4)
Title |
---|
BYOUYO: "灰度图像二值化阈值选取常用方法", 《HTTPS://WENKU.BAIDU.COM/VIEW/52C113650B1C59EEF8C7B4F6.HTML?SXTS=1583830390201》 * |
SUMMER_ZJU: "图像处理算法—区域生长法", 《HTTPS://BLOG.CSDN.NET/XIAQUNFENG123/ARTICLE/DETAILS/16971913》 * |
李文勇 等: "基于形状因子和分割点定位的粘连害虫图像分割方法", 《农业工程学报》 * |
高宏伟 等: "《电子制造装备技术》", 30 September 2015 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111178354A (en) * | 2019-12-23 | 2020-05-19 | 深圳市铁汉生态环境股份有限公司 | Mangrove pest monitoring method and system |
CN111178354B (en) * | 2019-12-23 | 2024-02-27 | 中节能铁汉生态环境股份有限公司 | Mangrove pest monitoring method and system |
CN111209824A (en) * | 2019-12-30 | 2020-05-29 | 江苏省血吸虫病防治研究所 | Method for monitoring ampullaria gigas by using unmanned aerial vehicle |
CN111696127A (en) * | 2020-06-02 | 2020-09-22 | 中国联合网络通信集团有限公司 | Locust plague analysis and early warning system and method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109389139A (en) | A kind of locust method of counting and device | |
US20120224789A1 (en) | Noise suppression in low light images | |
CN109146891B (en) | Hippocampus segmentation method and device applied to MRI and electronic equipment | |
US11120307B2 (en) | Multi-task learning for dense object detection | |
CN109740721A (en) | Wheat head method of counting and device | |
CN116012364B (en) | SAR image change detection method and device | |
CN110533632B (en) | Image blurring tampering detection method and device, computer equipment and storage medium | |
CN102999922A (en) | Multi-cell automatic tracking method and system based on plurality of task ant systems | |
CN109685806A (en) | Image significance detection method and device | |
CN105869175A (en) | Image segmentation method and system | |
CN109658378A (en) | Hole discrimination method and system based on soil CT image | |
Deshmukh et al. | Segmentation of microscopic images: A survey | |
CN105354575B (en) | Image binaryzation threshold value determines method in a kind of sea horizon detection | |
CN108961316A (en) | Image processing method, device and server | |
CN107392887A (en) | A kind of heterogeneous method for detecting change of remote sensing image based on the conversion of homogeneity pixel | |
CN109934784A (en) | Image enchancing method, device, equipment and computer readable storage medium | |
CN114783021A (en) | Intelligent detection method, device, equipment and medium for wearing of mask | |
CN110110736A (en) | Increment clustering method and device | |
CN105913451B (en) | A kind of natural image superpixel segmentation method based on graph model | |
KR20210013659A (en) | Improvement of image classification using boundary-bitmaps | |
CN110288025A (en) | Frequency spectrum sensing method, device and equipment based on information geometry and spectral clustering | |
CN102903104A (en) | Subtractive clustering based rapid image segmentation method | |
CN105589896B (en) | Data digging method and device | |
Chang et al. | Batch-invariant nuclear segmentation in whole mount histology sections | |
CN111047579A (en) | Characteristic quality evaluation method and image characteristic uniform extraction method |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190226 |