CN106295789A - A kind of crop seed method of counting based on image procossing - Google Patents
A kind of crop seed method of counting based on image procossing Download PDFInfo
- Publication number
- CN106295789A CN106295789A CN201510323529.1A CN201510323529A CN106295789A CN 106295789 A CN106295789 A CN 106295789A CN 201510323529 A CN201510323529 A CN 201510323529A CN 106295789 A CN106295789 A CN 106295789A
- Authority
- CN
- China
- Prior art keywords
- result
- seed
- connected domain
- recorded
- area
- 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
Landscapes
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to computer image processing technology field, be specifically related to the method for counting of a kind of crop seed.The method mainly includes, drawing of seeds picture is carried out pretreatment, and main method has image gray processing, binaryzation, Canny method to strengthen the boundary information etc. of drawing of seeds；Utilizing morphology and area features that the connected domain in image is divided into two classes, the first kind is the seed separated, and Equations of The Second Kind is the seed of adhesion；According to concave point feature, adhesion seed is cut；Number and the position of Sub Connected is planted after finally finding out separation.Instant invention overcomes traditional breeding method slow to seed counting, the shortcomings such as efficiency is low, and successfully image processing techniques is applied to breeding field, through verifying on 5 kinds of crop seeds, discrimination is all more than 99%, the process time on common PC, less than 3 seconds, has therefore reached quick accurately to seed counting, has provided strong technical support for breeding counting.
Description
Technical field
The present invention relates to computer image processing technology field, be specifically related to the method for counting of a kind of crop seed.
Background technology
In order to carry out seed testing, measure the scientific experiment such as yield, species test, several is indispensable step, and the precision of several has strong influence to experimental result.Breeding is that the development to plant husbandry and animal husbandry has very important significance, and counting is an important step of breeding to reach to cultivate the technology of excellent animals and plants new varieties by creation hereditary variation, improvement inherited character.It is domestic Digital image technology is started late in terms of the breeding, in recent years, the importance of Chinese agricultural development is more and more prominent, wherein breeding technique is exactly the very important part of agricultural development, by concern and the research of a lot of research worker, achieve certain achievement in research in image segmentation context of detection simultaneously.Cereal seed research essentially consists in Quality Detection and counting, and research concentrates on and splits adhesion cereal seed mostly, and the method that wherein range conversion and watershed combine is the most common；Additive method includes segmentation based on elliptic curve matching, based on segmentation utilizing seed contour curvature etc..So far, the image segmentation algorithm that scientific research personnel proposes can be divided into the types such as rim detection, thresholding, pixel classifications and the artificial neural network of comprehensive use, fuzzy set theory.Currently the research to corn counting concentrates on the corn segmentation retaining the original profile of corn substantially, but segmentation effect is barely satisfactory, counts and requires more high to environment and picture, and speed is the most relatively slow；Thus its convenience, accuracy, agility all has much room for improvement.At present, the mode of artificial counting is the most universal, and advantage is that real-time is high, and tool demands is low, and shortcoming to be efficiency low, easily make mistakes, and labour force is big, easily cause visual fatigue.Also there is electromechanical integration number the equipment of minority, can substitute for artificial counting, but also exist that error is big, manufacture the problems such as complicated, expensive, it is difficult to popularization and application widely.In sum, in order to solve above-mentioned the deficiencies in the prior art, a kind of convenient cereal seed method of counting accurately is provided, the present invention provides a kind of crop seed method of counting based on image procossing on the basis of image processing techniques, through the 5 kinds of cereal seeds having Breeding value have been done applied research, it was demonstrated that the method can count seed amount, simple in construction quickly and accurately, easy and simple to handle, it is adaptable to various crop seed.
Summary of the invention
(1) to solve the technical problem that
A kind of crop seed method of counting based on image procossing of offer is provided, overcome image segmentation speed slow, the problem that accuracy is low, overcome that Traditional Man counting efficiency is low simultaneously, easily make mistakes, labour force is big and electromechanical integration number equipment exists error is big, manufacture the problems such as complicated, expensive, image processing techniques is more preferably applied to breeding field, reach quick accurately to seed counting, provided strong technical support for breeding counting.
(2) technical scheme
The present invention is achieved by the following technical solutions, including step:
S1, acquisition drawing of seeds picture, be recorded as figure bmp1；
S2, Image semantic classification, obtain figure bmp1, image carry out gray processing, and smothing filtering, binaryzation, Canny method strengthen binary map boundary information etc., and pre-processed results is recorded in figure bmp2；
S3, image are classified, and obtain figure bmp2, classify image, first with corrosion, the seed of weak adhesion is separated, kind of a Sub Connected is divided into two classes by recycling area features, and the first kind is the connected domain class only comprising a seed in a connected domain, this type of is recorded in figure dstBmp；Equations of The Second Kind is the connected domain class comprising multiple seed in a connected domain, this type of is recorded in figure bmp3；
S4, cutting according to concave point, obtain figure bmp3, the detection concave point feature of connected domain and area features, utilize the two feature that kind of a Sub Connected is circulated cutting, cutting result is recorded in figure bmp4；
S5, class merge, and will merge by two class images, obtain result bmp4 that S4 processes, corrode this figure and go little connected domain to operate；The result figure dstBmp that obtaining step S3 processes again, is added bmp4 and dstBmp, and will add up result and be stored in figure bmp5.
S6, counting display, obtain figure bmp5, searches number and the position of connected domain in figure, and show and check.
Preferential described step S2 includes:
S2.1, acquirement S1 result figure bmp1, carry out gray processing process to it, and recycling bilateral filtering and medium filtering are smoothed, and result is recorded in figure img21；
Figure img21 after S2.2, acquisition S2.1 process, utilizes maximum variance between clusters that it is carried out binaryzation, and result figure is recorded as img22；
Figure img21 after S2.3, acquisition S2.1 process, utilizes canny method to find out its marginal information, and is recorded in figure img23；
Binary map img22 after S2.4, acquisition S2.2 process and the result figure img23 after S2.3 process, then img23 is deducted with img22, strengthening the marginal information of bianry image with this, the binary map strengthening edge feature is recorded in figure bmp2, and passes it to next step S3.
Preferential, described step S3 includes:
Figure bmp2 after S3.1, acquisition S2 process, utilizes caustic solution to be separated by the kind Sub Connected of adhesion weak in figure, and its result is recorded in figure img31；
Figure img31 after S3.2, acquisition S3.1 process, the area features of detection connected domain, and record area features；
Area features after figure img31 and S3.2 process after S3.3, acquisition S3.1 process, utilizes area features that kind of a Sub Connected is divided into two classes, and the first kind is the connected domain class only comprising a seed in a connected domain, this type of is recorded in figure dstBmp；Equations of The Second Kind is the connected domain class comprising multiple seed in a connected domain, this type of is recorded in figure bmp3；
S3.4, step S3.1, S3.2, S3.3 circulation perform three times.
Preferential, described step S4 includes:
S4.1, area of detection feature, obtain S3 result figure bmp3, count out the area area [i] of its i-th connected domain；Obtain the result figure dstBmp that S3 processes, count the average area averarea of its single seed；
S4.2, detection concave point feature, obtain S3 result figure bmp3, utilize the method for convex defect to detect its concave point feature, including concave point position, concave point number, the concave point degree of depth；
S4.3, acquisition S3 result figure bmp3, the seed number comprised according to i-th connected domain in area features judgement figure, it may be assumed that
If if 10 ＜ area [i]≤(3 × averarea/2), then judging i-th connected domain only comprises 1 seed, not cutting；
2 (if 3 × averarea/2) ＜ area [i]≤(5 × averarea/2), then judge i-th connected domain comprises 2 seeds, once cut according to concave point；
3 (if 5 × averarea/2) ＜ area [i], then judge i-th connected domain comprises multiple seed, once cut according to concave point；
S4.4, repetition step S4.1, S4.2, S4.3, termination condition is that cutting times reaches 5 times or no longer have in connected domain to comprise two or more seed number.Image after cutting is recorded in figure bmp4；
Preferential, described step S5 includes:
S5.1 obtains S4 result figure bmp4, corrodes them 2 times, and the little noise stayed after removing cutting, its result is recorded in figure img51；
S5.2 obtains S5.1 result figure img51, utilizes area features to remove its little connected domain, even has the meansigma methods that the area value of connected domain is less than 1/3 times, then it is assumed that being the little connected domain removed, result is recorded in figure img52；
S5.3 obtains S3 result figure dstBmp, obtains S5.2 result figure img52, is added by dstBmp and img52, is added the separating resulting just having obtained whole seed, and result figure is recorded in figure bmp5.
Preferential, described step S6 includes:
Obtain S5 result figure bmp5, find out the minimum enclosed rectangle of each connected domain in image, and using the center of boundary rectangle as the center of seed, boundary rectangle number, as seed number, finally shows result.
(3) beneficial effect
The present invention is a kind of crop seed method of counting based on image procossing, and it has the beneficial effect that:
(1) digital image processing techniques are applied to the seed counting in breeding field by the present invention, replace artificial counting and the mode of mechanical count, save human cost and equipment cost, work efficiency is significantly increased；
(2) instant invention overcomes image and split processing speed when being applied to seed counting slowly, the shortcomings such as accuracy is low, reach speed fast, the advantages such as accuracy is high, and stability is strong, repeatable batch processing；
(3) inventive algorithm is to combine opencv exploitation with C/C++ language, there is the strongest portability, applied in Android panel computer now, take pictures conveniently during plate count, the graphic interface of Android is also convenient for man-machine interaction, conveniently check counting effect, and can artificially revise, it is ensured that the accuracy of counting.
(4) present invention is simple, easily operation, several the most efficiently, solve the problems such as cost height, maintenance difficult, and applied at various crop seed (such as Semen Maydis, Semen phaseoli radiati, Semen Tritici aestivi, Semen Sesami, Oryza sativa L. etc.) and verify.
Detailed description of the invention
Below in conjunction with the accompanying drawings and detailed description of the invention, the detailed description of the invention of invention is described further.See Fig. 1, it is seen that the piecemeal of the method used of the present invention and each method links, and method of mainly using in image processing module has smothing filtering method, maximum variance between clusters, Canny method etc.；Image classification module has mainly used area features detection and morphological method etc., according in concave point cutting module, mainly according to concave point feature and area features, kind of Sub Connected being carried out cutting etc., last class merges mainly has run morphology and area features removal noise in module.The handling process of the present invention is shown in Fig. 2, mainly has six parts, mainly has the acquisition of drawing of seeds picture, Image semantic classification, and image is classified, and cuts according to concave point, and two class images merge, counting display；It is described in greater detail below.
(1) drawing of seeds picture is obtained
Obtaining drawing of seeds picture is module S1, and the image of acquisition is recorded as bmp1.
Obtaining image has condition to limit:
Seeing the application interface figure on the android of Fig. 3, the present invention is applied on panel computer, and image can obtain with panel computer camera, and image background to be white；When taking figure, need to add backlight, seed needs tiling, it is impossible to overlapping, can adhesion, quantity is optimal between 50-1500, mutually confidential upright, and photo wants clear, and pixel is 8,000,000 optimal, and camera and receiver distance are between 20cm-30cm.
(2) Image semantic classification
Image semantic classification is module S2, obtains the result figure bmp1 of S1, and image carries out gray processing, and smothing filtering, Canny method strengthen binary map boundary information etc., and pre-processed results figure is recorded in figure bmp2, and it participates in Fig. 6.
The detailed step of Image semantic classification is as follows:
S2.1, acquirement S1 result figure bmp1, carry out gray processing process to it, utilizing bilateral filtering and medium filtering to be smoothed, and result figure is recorded as img21；
1, gray processing is i.e. that the R of coloured image, G, channel B brightness value are converted into single channel brightness value and characterize the image that each pixel is constituted, and conversion formula is:
Gray (x, y)=0.299*R (x, y)+0.587*G (x, y)+0.114*B (x, y) (1)
2, smothing filtering mainly uses medium filtering and bilateral filtering；Each pixel value median pixel value in the square field of center pixel is replaced by median filter, and camera lens noise is had good treatment effect；Bilateral filtering can well retain the margin signal of image, so both filtering can well be filtered denoising on the premise of retaining margin signal, the filter window size of wave filter is 3 × 3；
Figure img21 after S2.2, acquisition S2.1 process, utilizes maximum variance between clusters that image carries out binaryzation, and result figure is recorded as img22, and it sees Fig. 4.
Herein, using the overall situation binarization method, i.e. maximum variance between clusters, but it needs to ensure that seed occupies certain quantity in the drawings, seed number is optimal at 50-1500.
Figure img21 after S2.3, acquisition S2.1 process, utilizes canny method to find out its marginal information, and is recorded in figure img23, and it sees Fig. 5；
In the binary map obtained after Binary Sketch of Grey Scale Image, the marginal information of seed can lose a part, and canny method is very sensitive to border, can well find out the boundary information in gray level image, the boundary information lost when just may be used for making up image binaryzation.
Binary map img22 after S2.4, acquisition S2.2 process, obtains S2.3 result figure img23, then deducts img23 with img22, strengthen the marginal information of bianry image with this, and result is recorded in figure bmp2.
Can not well preserve boundary information during visible binaryzation, and canny can well find out border, through strengthening binary map BORDER PROCESSING, it is possible to obtain the binary map having fine boundary characteristic.
(3) image classification
Image is categorized as module S3, obtain the result figure bmp2 that S2 processes, first with caustic solution, the seed of weak adhesion is separated, kind of a Sub Connected is divided into two classes by recycling area features, the first kind is the connected domain class only comprising a seed in a connected domain, this type of being recorded in figure dstBmp, it sees Fig. 7；Equations of The Second Kind is the connected domain class comprising multiple seed in a connected domain, this type of is recorded in figure bmp3, and it sees Fig. 8；
The detailed step of image classification is as follows:
Figure bmp2 after S3.1, acquisition S2 process, utilizes caustic solution to be separated by the kind Sub Connected of adhesion weak in figure, and its result is recorded in figure img31；Corrosion number of times is different according to varying in size of seed, and used herein is paddy seeds, and corrosion number of times is 4 times, and corrosion kernel is 3*3；
Figure img31 after S3.2, acquisition S3.1 process, corrodes 2 times it, and result is recorded in figure img32；
Corrosion purpose is to be separated by the kind Sub Connected of adhesion weak in figure, and corrosion kernel is 3*3, and this step will be used in the middle of cyclic sort；
Figure img31 after S3.3, acquisition S3.1 process, the area features of detection connected domain, and record area features；
The area recording i-th connected domain is area [i], counts its average area averarea, connected domain number n simultaneously, and average area is:
The area features after figure img32 and S3.3 process after S3.4, acquisition S3.2 process, utilizes area features that kind of a Sub Connected is divided into two classes, and the first kind is the connected domain class only comprising a seed in a connected domain, this type of is recorded in figure dstBmp；Equations of The Second Kind is the connected domain class comprising multiple seed in a connected domain, this type of is recorded in figure bmp3；
Have the following steps:
1, the figure img32 after S3.2 processes is obtained, it is judged that i-th connected domain comprises seed number, it is judged that method is to carry out judging according to the area features after obtaining S3.1 process；
If 20 ＜ area [i] ＜ (3 × averarea/2), it is judged that comprise 1 seed in i-th connected domain, then this type of connected domain is recorded in figure dstBmp；
3 (if 3 × averarea/2) ＜ area [i], it is judged that comprise multiple seed in i-th connected domain, this type of is recorded in figure bmp3；
S3.5, step S3.2, S3.3, S3.4, S3.5 circulation perform three times.
Circulation execution number of times is different according to kind of varying in size of Sub Connected, and used herein is rice paddy seed, the photo of 8,000,000 pixels, its circulation three times.
(4) cut according to concave point
Being module S4 according to concave point cutting, obtain S3 result figure bmp3, the concave point feature of detection connected domain and area features, utilize the two feature that kind of a Sub Connected is circulated cutting, cutting result figure is bmp4, and it sees Fig. 9.
Detailed step according to concave point cutting is as follows:
S4.1, area of detection feature, obtain S3 result figure bmp3, find out the area area [i] of its i-th connected domain；Obtaining S3 result figure dstBmp, count the average area averarea of its single seed, its counting mode sees formula (2)；Wherein the area of connected domain refers to the number of pixels comprised in connected domain.
S4.2, detection concave point feature, utilize convex defect method to detect concave point feature, including concave point position, concave point number, the concave point degree of depth etc.；
S4.3, acquisition S3 result figure bmp3, the seed number comprised according to i-th connected domain in area features judgement figure, it may be assumed that
If 10 ＜ area [i]≤(3 × averarea/2), then judge i-th connected domain only comprises 1 seed, do not cut；
Here it is circulation cutting, when single seed is cut out, avoids the need for cutting again, i.e. meet the kind Sub Connected of conditions above not cut.
2 (if 3 × averarea/2) ＜ area [i]≤(5 × averarea/2), then judge i-th connected domain comprises 2 seeds, once cut according to concave point；
The connected domain meeting conditions above is considered in a connected domain to comprise two seeds, needs once to cut, and main cutting method is to connect the deepest two concave point to cut；
3 (if 5 × averarea/2) ＜ area [i], then judge i-th connected domain comprises multiple seed, once cut according to concave point；
The connected domain meeting conditions above is considered in a connected domain to comprise multiple seed, this connected domain is the most once cut, the remaining part not having to cut will enter circulation cutting, main cutting method is when the degree of depth of concave point meets more than 15 pixels, find out nearest two concave point met in condition, link the two concave point, cut with this.
S4.4, repetition step S4.1, S4.2, S4.3, termination condition is that cutting times reaches 5 times or no longer have in connected domain to comprise two and above seed number.Image after cutting is recorded in figure bmp4.
(5) class merges
Class merges into module S5, will merge by two class images, obtains result bmp4 that S4 processes, corrode this figure and go little connected domain to operate；Obtaining the result figure dstBmp of the process of S3 again, be added by bmp4 and dstBmp, and will add up result and be stored in figure bmp5, its amalgamation result sees Figure 10.
The detailed step that class merges is as follows:
S5.1 obtains S4 result figure bmp4, corrodes them 2 times, and the little noise stayed after removing cutting, its result is recorded in figure img51；
After cutting according to concave point, can leave over down some and cross the spiced salt noise that cutting causes in image, so the method here with corrosion carries out little noise removal, caustic solution kernel is 3 × 3.
S5.2 obtains S5.1 result figure img51, utilizes area features to remove its little connected domain, when even having the average area that the area of connected domain is less than 1/3 times, then it is assumed that be the little connected domain removed, go division result to be recorded in figure img52；Execution step is as follows:
1, in calculating figure, the area features of i-th connected domain is area [i], counts its average area averarea simultaneously, and method of counting is shown in formula (2)；
If 2 area [i] ＜ (averarea/3), then it is assumed that i-th connected domain is noise, removes it；
S5.3 obtains S3 result figure dstBmp, and obtains S5.2 result figure img52, is added by dstBmp and img52, is added the separating resulting just having obtained whole seed, and result figure is recorded in figure bmp5.
(6) counting display
Counting is shown as module S6, obtains S5 result figure bmp5, finds out the minimum enclosed rectangle of each connected domain in image, and using the center of boundary rectangle as the center of seed, boundary rectangle number, as seed number, finally shows that result, display result are shown in Figure 11.
Accompanying drawing explanation
Fig. 1 functional block diagram
Illustrate: figure illustrates the main method and technology used in each module, and the binding sequence between each module；
Fig. 2 process chart
Illustrate: figure illustrates the handling process of each image and the process structure of each method and priority；
Fig. 3 android surface chart
Illustrate: the inventive method applies the counting design sketch on Android flat board, the User Interface in Android, in that context it may be convenient to check the effect of seed counting；
Fig. 4 binaryzation result figure
Illustrate: figure illustrates the result figure of smooth rear directly overall situation binaryzation；
The edge feature figure that Fig. 5 Canny extracts
Illustrate: figure illustrates and utilizes Canny method that seed gray-scale map is carried out the result of marginal information lookup.
Fig. 6 strengthens the binary map of edge feature
Illustrate: figure illustrates the result figure that binary map is strengthened through Canny method marginal information, be i.e. Fig. 4 result of deducting Fig. 5.
The single drawing of seeds of Fig. 7
Illustrate: figure illustrates the connected domain class only comprising a seed in a connected domain, the most single seed result figure.
Fig. 8 adhesion drawing of seeds
Illustrate: figure illustrates the connected domain class comprising multiple seed in a connected domain, i.e. adhesion drawing of seeds.
The result figure that Fig. 9 cuts according to concave point
Illustrating: illustrate the result figure according to concave point cutting in figure, it is seen that had the left little connected domain of cutting, it needs to remove.
Figure 10 amalgamation result figure
Illustrate: this figure illustrates the result that each seed is all separated, be the result figure cut.
Figure 11 counts display figure
Illustrate: figure illustrates counting effect, first cutting result figure is carried out connected domain lookup, the effect finally shown together with artwork by lookup result.
Claims (6)
1. a crop seed method of counting based on image procossing, it is characterised in that include step:
S1, acquisition drawing of seeds picture, be recorded as figure bmp1；
S2, Image semantic classification, obtain figure bmp1, image carry out gray processing, and smothing filtering, binaryzation, Canny method strengthen binary map boundary information etc., and pre-processed results is recorded in figure bmp2；
S3, image are classified, and obtain figure bmp2, classify image, first with corrosion, the seed of weak adhesion is separated, kind of a Sub Connected is divided into two classes by recycling area features, and the first kind is the connected domain class only comprising a seed in a connected domain, this type of is recorded in figure dstBmp；Equations of The Second Kind is the connected domain class comprising multiple seed in a connected domain, this type of is recorded in figure bmp3；
S4, cutting according to concave point, obtain figure bmp3, the detection concave point feature of connected domain and area features, utilize the two feature that kind of a Sub Connected is circulated cutting, cutting result is recorded in figure bmp4；
S5, class merge, and will merge by two class images, obtain result bmp4 that S4 processes, corrode this figure and go little connected domain to operate；The result figure dstBmp that obtaining step S3 processes again, is added bmp4 and dstBmp, and will add up result and be stored in figure bmp5；
S6, counting display, obtain figure bmp5, searches number and the position of connected domain in figure, and show and check.
A kind of crop seed method of counting based on image procossing the most according to claim 1, it is characterised in that described step S2 includes:
S2.1, acquirement S1 result figure bmp1, carry out gray processing process to it, and recycling bilateral filtering and medium filtering are smoothed, and result is recorded in figure img21；
Figure img21 after S2.2, acquisition S2.1 process, utilizes maximum variance between clusters that it is carried out binaryzation, and result figure is recorded as img22；
Figure img21 after S2.3, acquisition S2.1 process, utilizes canny method to find out its marginal information, and is recorded in figure img23；
Binary map img22 after S2.4, acquisition S2.2 process and the result figure img23 after S2.3 process, then deduct img23 with img22, strengthen the marginal information of bianry image with this, and the binary map strengthening edge feature is recorded in figure bmp2.
A kind of crop seed method of counting based on image procossing the most according to claim 1, it is characterised in that described step S3 includes:
Figure bmp2 after S3.1, acquisition S2 process, utilizes caustic solution to be separated by the kind Sub Connected of adhesion weak in figure, and its result is recorded in figure img31；
Figure img31 after S3.2, acquisition S3.1 process, the area features of detection connected domain, and record area features；
Area features after figure img31 and S3.2 process after S3.3, acquisition S3.1 process, utilizes area features that kind of a Sub Connected is divided into two classes, and the first kind is the connected domain class only comprising a seed in a connected domain, this type of is recorded in figure dstBmp；Equations of The Second Kind is the connected domain class comprising multiple seed in a connected domain, this type of is recorded in figure bmp3；
S3.4, step S3.1, S3.2, S3.3 circulation perform three times.
A kind of crop seed method of counting based on image procossing the most according to claim 1, it is characterised in that described step S4 includes:
S4.1, area of detection feature, obtain S3 result figure bmp3, count out the area area [i] of its i-th connected domain；Obtain the result figure dstBmp that S3 processes, count the average area averarea of its single seed；
S4.2, detection concave point feature, obtain S3 result figure bmp3, utilize the method for convex defect to detect its concave point feature, including concave point position, concave point number, the concave point degree of depth；S4.3, acquisition S3 result figure bmp3, the seed number comprised according to i-th connected domain in area features judgement figure, it may be assumed that
If 10 ＜ area [i]≤(3 × averarea/2), then judge i-th connected domain only comprises 1 seed, do not cut；
2 (if 3 × averarea/2) ＜ area [i]≤(5 × averarea/2), then judge i-th connected domain comprises 2 seeds, once cut according to concave point；
3 (if 5 × averarea/2) ＜ area [i], then judge i-th connected domain comprises multiple seed, once cut according to concave point；
S4.4, repetition step S4.1, S4.2, S4.3, termination condition is that cutting times reaches 5 times or no longer have in connected domain to comprise two or more seed number.Image after cutting is recorded in figure bmp4.
A kind of crop seed method of counting based on image procossing the most according to claim 1, it is characterised in that described step S5 includes:
S5.1 obtains S4 result figure bmp4, corrodes them 2 times, and the little noise stayed after removing cutting, its result is recorded in figure img51；
S5.2 obtains S5.1 result figure img51, utilizes area features to remove its little connected domain, even has the average area that the area of connected domain is less than 1/3 times, then it is assumed that this connected domain is the little connected domain removed, and result is recorded in figure img52；
S5.3 obtains S3 result figure dstBmp, obtains S5.2 result figure img52, is added by dstBmp and img52, is added the separating resulting just having obtained whole seed, and result figure is recorded in figure bmp5.
A kind of crop seed method of counting based on image procossing the most according to claim 1, it is characterised in that described step S6 includes:
Obtain S5 result figure bmp5, find out the minimum enclosed rectangle of each connected domain in image, and using the center of boundary rectangle as the center of seed, boundary rectangle number, as seed number, finally shows result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510323529.1A CN106295789B (en) | 2015-06-10 | 2015-06-10 | Crop seed counting method based on image processing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510323529.1A CN106295789B (en) | 2015-06-10 | 2015-06-10 | Crop seed counting method based on image processing |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106295789A true CN106295789A (en) | 2017-01-04 |
| CN106295789B CN106295789B (en) | 2021-07-09 |
Family
ID=57649957
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510323529.1A Active CN106295789B (en) | 2015-06-10 | 2015-06-10 | Crop seed counting method based on image processing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106295789B (en) |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106898010A (en) * | 2017-03-01 | 2017-06-27 | 上海市农业科学院 | Particle copies the method and device planted |
| CN106918595A (en) * | 2017-03-22 | 2017-07-04 | 扬州大学 | A kind of head rice rate batch assay method and its equipment |
| CN107300360A (en) * | 2017-08-10 | 2017-10-27 | 山西农业大学 | A kind of shaft size fast algorithm of detecting of rain fed crops seed three |
| CN107564001A (en) * | 2017-09-13 | 2018-01-09 | 电子科技大学 | A kind of magnetic sheet unfilled corner detection method based on concave point search |
| CN107578414A (en) * | 2017-08-18 | 2018-01-12 | 东南大学 | A kind of processing method of pavement crack image |
| CN107767361A (en) * | 2017-08-28 | 2018-03-06 | 江苏理工学院 | A kind of carborundum line grain count method, storage device and terminal device |
| CN107909138A (en) * | 2017-11-14 | 2018-04-13 | 江苏大学 | A kind of class rounded grain thing method of counting based on Android platform |
| CN108052886A (en) * | 2017-12-05 | 2018-05-18 | 西北农林科技大学 | A kind of puccinia striiformis uredospore programming count method of counting |
| CN110910403A (en) * | 2019-11-16 | 2020-03-24 | 厦门梓蔓生物科技有限公司 | Industrial hemp seed counting method based on image processing |
| CN111860038A (en) * | 2019-04-25 | 2020-10-30 | 河南中原光电测控技术有限公司 | Crop front-end recognition device and method |
| CN113298768A (en) * | 2021-05-20 | 2021-08-24 | 山东大学 | Cotton detection, segmentation and counting method and system |
| CN113642694A (en) * | 2021-08-11 | 2021-11-12 | 海南青峰生物科技有限公司 | Rice seed counting method based on 5G communication and image recognition processing |
| CN114495098A (en) * | 2022-01-30 | 2022-05-13 | 南水北调中线干线工程建设管理局 | Diaxing algae cell statistical method and system based on microscope image |
| CN114937077A (en) * | 2022-04-22 | 2022-08-23 | 南通荣华包装材料有限公司 | Peanut seed screening method |
| CN115546462A (en) * | 2022-12-01 | 2022-12-30 | 南京维拓科技股份有限公司 | Method for extracting shape features of product and counting based on image recognition |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH08214619A (en) * | 1995-02-08 | 1996-08-27 | Kubota Corp | Apparatus for evaluating seeding |
| CN101441721A (en) * | 2008-11-28 | 2009-05-27 | 江苏大学 | Device and method for counting overlapped circular particulate matter |
| CN101976447A (en) * | 2010-09-29 | 2011-02-16 | 广州中医药大学 | Method for cutting off petioles under condition of keeping images of lamina main bodies undamaged |
| CN104240243A (en) * | 2014-09-05 | 2014-12-24 | 南京农业大学 | Adhered piglet automatic counting method based on ellipse fitting |
-
2015
- 2015-06-10 CN CN201510323529.1A patent/CN106295789B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH08214619A (en) * | 1995-02-08 | 1996-08-27 | Kubota Corp | Apparatus for evaluating seeding |
| CN101441721A (en) * | 2008-11-28 | 2009-05-27 | 江苏大学 | Device and method for counting overlapped circular particulate matter |
| CN101976447A (en) * | 2010-09-29 | 2011-02-16 | 广州中医药大学 | Method for cutting off petioles under condition of keeping images of lamina main bodies undamaged |
| CN104240243A (en) * | 2014-09-05 | 2014-12-24 | 南京农业大学 | Adhered piglet automatic counting method based on ellipse fitting |
Non-Patent Citations (1)
| Title |
|---|
| 张亚秋 等: "基于逐步改变阈值方法的玉米种子图像分割", 《农业工程学报》 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106898010A (en) * | 2017-03-01 | 2017-06-27 | 上海市农业科学院 | Particle copies the method and device planted |
| CN106918595A (en) * | 2017-03-22 | 2017-07-04 | 扬州大学 | A kind of head rice rate batch assay method and its equipment |
| CN107300360A (en) * | 2017-08-10 | 2017-10-27 | 山西农业大学 | A kind of shaft size fast algorithm of detecting of rain fed crops seed three |
| CN107578414B (en) * | 2017-08-18 | 2021-09-07 | 东南大学 | Method for processing pavement crack image |
| CN107578414A (en) * | 2017-08-18 | 2018-01-12 | 东南大学 | A kind of processing method of pavement crack image |
| CN107767361A (en) * | 2017-08-28 | 2018-03-06 | 江苏理工学院 | A kind of carborundum line grain count method, storage device and terminal device |
| CN107564001A (en) * | 2017-09-13 | 2018-01-09 | 电子科技大学 | A kind of magnetic sheet unfilled corner detection method based on concave point search |
| CN107564001B (en) * | 2017-09-13 | 2020-09-25 | 电子科技大学 | Magnetic sheet corner defect detection method based on pit search |
| CN107909138A (en) * | 2017-11-14 | 2018-04-13 | 江苏大学 | A kind of class rounded grain thing method of counting based on Android platform |
| CN108052886A (en) * | 2017-12-05 | 2018-05-18 | 西北农林科技大学 | A kind of puccinia striiformis uredospore programming count method of counting |
| CN111860038A (en) * | 2019-04-25 | 2020-10-30 | 河南中原光电测控技术有限公司 | Crop front-end recognition device and method |
| CN111860038B (en) * | 2019-04-25 | 2023-10-20 | 河南中原光电测控技术有限公司 | Crop front end recognition device and method |
| CN110910403A (en) * | 2019-11-16 | 2020-03-24 | 厦门梓蔓生物科技有限公司 | Industrial hemp seed counting method based on image processing |
| CN113298768A (en) * | 2021-05-20 | 2021-08-24 | 山东大学 | Cotton detection, segmentation and counting method and system |
| CN113642694A (en) * | 2021-08-11 | 2021-11-12 | 海南青峰生物科技有限公司 | Rice seed counting method based on 5G communication and image recognition processing |
| CN113642694B (en) * | 2021-08-11 | 2024-05-14 | 海南青峰生物科技有限公司 | Rice seed counting method based on 5G communication and image recognition processing |
| CN114495098A (en) * | 2022-01-30 | 2022-05-13 | 南水北调中线干线工程建设管理局 | Diaxing algae cell statistical method and system based on microscope image |
| CN114937077A (en) * | 2022-04-22 | 2022-08-23 | 南通荣华包装材料有限公司 | Peanut seed screening method |
| CN115546462A (en) * | 2022-12-01 | 2022-12-30 | 南京维拓科技股份有限公司 | Method for extracting shape features of product and counting based on image recognition |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106295789B (en) | 2021-07-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106295789A (en) | A kind of crop seed method of counting based on image procossing | |
| CN107330465B (en) | A kind of images steganalysis method and device | |
| Quelhas et al. | Cell nuclei and cytoplasm joint segmentation using the sliding band filter | |
| CN102043950B (en) | Vehicle outline recognition method based on canny operator and marginal point statistic | |
| CN104021574B (en) | Pavement disease automatic identifying method | |
| CN104408707B (en) | Rapid digital imaging fuzzy identification and restored image quality assessment method | |
| CN103996018B (en) | Face identification method based on 4DLBP | |
| CN109859181A (en) | A kind of PCB welding point defect detection method | |
| CN106548463A (en) | Based on dark and the sea fog image automatic defogging method and system of Retinex | |
| CN103699900B (en) | Building horizontal vector profile automatic batch extracting method in satellite image | |
| CN107273896A (en) | A kind of car plate detection recognition methods based on image recognition | |
| CN101901342B (en) | Method and device for extracting image target region | |
| CN103984958A (en) | Method and system for segmenting cervical caner cells | |
| CN105389581B (en) | A kind of rice germ plumule integrity degree intelligent identifying system and its recognition methods | |
| CN107871125A (en) | Architecture against regulations recognition methods, device and electronic equipment | |
| CN103593670A (en) | Copper sheet and strip surface defect detection method based on-line sequential extreme learning machine | |
| CN107871316B (en) | Automatic X-ray film hand bone interest area extraction method based on deep neural network | |
| CN107945200A (en) | Image binaryzation dividing method | |
| CN112614062A (en) | Bacterial colony counting method and device and computer storage medium | |
| CN112750106A (en) | Nuclear staining cell counting method based on incomplete marker deep learning, computer equipment and storage medium | |
| CN109544564A (en) | A kind of medical image segmentation method | |
| CN108109133A (en) | A kind of silkworm seed automatic counting method based on digital image processing techniques | |
| CN108416814A (en) | Quick positioning and recognition methods and the system on a kind of pineapple head | |
| CN103914829B (en) | Method for detecting edge of noisy image | |
| CN110276759A (en) | A kind of bad line defect diagnostic method of Mobile phone screen based on machine vision |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | 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 |