CN104132650A - Method for detection of fruit tree leaf density by maximum contour rectangle technique - Google Patents
Method for detection of fruit tree leaf density by maximum contour rectangle technique Download PDFInfo
- Publication number
- CN104132650A CN104132650A CN201410246400.0A CN201410246400A CN104132650A CN 104132650 A CN104132650 A CN 104132650A CN 201410246400 A CN201410246400 A CN 201410246400A CN 104132650 A CN104132650 A CN 104132650A
- Authority
- CN
- China
- Prior art keywords
- fruit tree
- image
- leaf
- area
- shared
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
- G06T2207/30188—Vegetation; Agriculture
Abstract
The invention discloses a method for detection of fruit tree leaf density by a maximum contour rectangle technique. Compared with the prior art, the maximum contour rectangle technique is employed to detect and calculate a fruit tree image area so as to obtain the fruit tree leaf density, and the result varies with the actual fruit tree image. The method is free of the problem of small detected leaf density caused by uniform adoption of a camera set image size as the maximum profile in existing methods, and is a digital image processing method that is used for fruit tree leaf density measurement and can more accurately reflect the biomass density level of fruit trees. Being simple and easy, the method provided by the invention can timelier and more effectively detect the fruit tree leaf density.
Description
Technical field
The present invention relates to technical field of image processing, especially a kind of method that adopts largest contours Rectangular Method to detect the rare close degree of fruit tree leaf.
Background technology
Detecting real-time of the rare close degree of fruit tree leaf is the gordian technique of the variable rate spray based on real-time.About the concept of individual plant fruit tree biomass density, also there is no at present definite definition both at home and abroad.The external biomass density research that mainly concentrates on forest or timber biological metric density and area crops, but the scholar who studies for individual plant fruit tree biomass density is seldom.The rare close degree of fruit tree leaf refers to that leaf and trunk in a width fruit tree image account for the Area Ratio of whole fruit tree image.Existing method camera when gathering image sets the size of image as the area of whole fruit tree image, but this value is larger than fruit tree real profile area, thereby causes the rare close degree value of fruit tree leaf calculated less than normal than reality; The method also requires collecting sample under same image-forming range as standard, to adopt thereafter BP Data fusion technique to eliminate the impact of image-forming range on the rare close degree detecting result of fruit tree leaf again.
Summary of the invention
Technical matters to be solved by this invention is: a kind of method that adopts largest contours Rectangular Method to detect the rare close degree of fruit tree leaf is provided, it can solve existing method and detect the problem that the rare close degree value of leaf is less than normal than actual value and affected by image-forming range, the biomass density level that can accurately reflect fruit tree, to overcome the deficiencies in the prior art.
The present invention is achieved in that the method that adopts largest contours Rectangular Method to detect the rare close degree of fruit tree leaf, comprises the steps:
1) by the recognition technology of image, utilize super green method to be combined with Otsu algorithm, identify fruit tree binary image clearly;
2) the fruit tree binary image in step 1) is carried out to medium filtering denoising,
3) by step 2) in denoising image by opening computing, impurity is eliminated in continuous 4 corrosion and 4 expansions respectively;
4) image step 3) being obtained carries out dilation operation processing, with the cavity in removal of images; Increase the white portion that represents fruit tree, solved the problem of wrong minute;
5) the imagery exploitation largest contours Rectangular Method of processing through step 4) is detected to its shared largest contours, comprise the length of fruit tree image and wide, then with this, calculate the contour area of fruit tree image;
5) calculate the shared area of leaf trunk of fruit tree in image;
6) utilize the shared area of the leaf of the fruit tree that step 4) calculates to be the rare close degree of fruit tree leaf divided by the value of the contour area of fruit tree image.
Step 2) morphology described in is processed specifically and is referred to: 1) the fruit tree image after medium filtering is first carried out continuously to 4 corrosion, then carry out continuously 4 times and expand, corrosion is all identical with the mechanism's element expanding.Continuous 4 corrosion are removed the independent image vegetarian refreshments in fruit tree image (weeds and soil) preferably with expansion energy, and fruit tree image outline becomes smooth, have highlighted fruit tree image.
Described in step 4), utilize largest contours Rectangular Method to detect its shared largest contours specifically to refer to, by detecting step 3) in whole the area A that fruit tree largest contours is shared in the image processed, detect the shared pixel quantity of maximum boundary of fruit tree horizontal and vertical directions, then calculate this two border and be the area A of the whole image of fruit tree for long and wide rectangular area; Detect again the area m of leaf and trunk in entire image, in this fruit tree bianry image, the pixel value that fruit tree leaf and trunk are corresponding is 1, and statistical pixel values is that 1 shared pixel count is the shared area m of leaf and trunk in entire image, according to formula (1), calculates the rare close degree η of leaf:
In formula (1), η represents the rare density of fruit tree leaf, and m represents the shared area of trunks of fruit trees and leaf in this width figure; A represents the area of the whole image of fruit tree.The area of A is to become according to the size of fruit tree reality, and therefore, m and A are variable values, only relevant with the upgrowth situation of fruit tree self.
Flow process of the present invention as shown in Figure 6, is utilized super green method and the clear fruit tree binary image 1 that identifies of OTSU algorithm by original colored fruit tree image process; To process fruit tree binary image 1 through medium filtering denoising through image recognition technology, the enhancing that morphology computing obtains the fruit tree binary image 2 of identity; The function of setting fruit tree bianry image 2 is after treatment f (x, y); If the size of f (x, y) equals M * N; And i=1,2 ..., M; J=1,2 ..., N.The point of f (i, j)=1 represents the point of fruit tree leaf or trunk image.From f (i, 1), start until f (i, N) position of the pixel that the pixel of adding up every a line is 1 (row coordinate) sequence H, 1≤H≤N; From f (1, j) start until f (M, j) position (row-coordinate) the sequence L of the pixel that the pixel of adding up each row is 1,1≤L≤M; To every one-row pixels, all try to achieve the columns f (I, first) of the position that occurs for the first time 1 and the difference Di of place, the position columns f (i, end) of last appearance 1, Di=f (i, end)-f (i, first); To each row pixel, all try to achieve the columns f (first, j) of the position that occurs for the first time 1 and the difference Dj of place, the position columns f (end, j) of last appearance 1, Dj=f (end, j)-f (first, j); Obtain maximal value Dh in Di and the maximal value Dl in Dj; Detect the shared maximum row distance of fruit tree image and maximum column apart from all comprising initial value, so the length of largest contours rectangle and wide value are respectively (Dh+1) and (Dl+1), the area 3 that can obtain whole fruit tree largest contours rectangle according to rectangular area computing rule is (Dh+1) and product (Dl+1); The shared area 4 of leaf trunk in detected image, the pixel number that bianry image intermediate value is 1; The rare close degree value 5 of fruit tree leaf is the business of the shared area of leaf trunk 4 and the area 3 of fruit tree largest contours rectangle in image.
Compared with prior art, the present invention adopts largest contours Rectangular Method that fruit tree image area is detected and calculated, to obtain the rare close degree of fruit tree leaf, this result because of fruit tree real image different, do not exist the unified image size that adopts camera to set of existing method to cause the rare close degree of detected leaf problem less than normal as largest contours, more can accurately reflect the digital image processing method of the rare density of measurement fruit tree leaf of the biomass density level of fruit tree.The present invention simply, easily goes, and result of use is good.
Accompanying drawing explanation
Accompanying drawing 1 is the fruit tree image of the super green method gray processing of process in embodiments of the invention;
Accompanying drawing 2 is the fruit tree binary image after OTSU is cut apart in embodiments of the invention;
Accompanying drawing 3 is the fruit tree image after medium filtering in embodiments of the invention;
Accompanying drawing 4 is the fruit tree image after having carried out continuous 4 corrosion in embodiments of the invention and having expanded;
Accompanying drawing 5 has carried out the fruit tree image after dilation operation in embodiments of the invention;
The process flow diagram that accompanying drawing 6 is embodiments of the invention.
Embodiment
Embodiments of the invention 1: adopt largest contours Rectangular Method to detect the method for the rare close degree of fruit tree leaf, comprise the steps:
1) by original colored fruit tree image (existing take set size be 640(pixel) * 480(pixel) fruit tree image be example) through utilizing super green method and the clear fruit tree binary image that identifies of OTSU algorithm, fruit tree is clearly extracted from the backgrounds such as soil, as shown in Figure 1, 2;
2) the fruit tree binary image in step 1) is utilized to medium filtering, the image of acquisition as shown in Figure 3;
3) Fig. 3 is first carried out continuously to 4 corrosion, then carry out continuously 4 times and expand, corrosion is all identical with the mechanism's element expanding.Continuous 4 corrosion are removed the independent image vegetarian refreshments in fruit tree image (weeds and soil) preferably with 4 expansion energies, and independent image vegetarian refreshments has obtained good removal, and fruit tree image outline becomes smooth, have highlighted fruit tree image, as shown in Figure 4;
4) image of processing in step 3) (Fig. 4) is carried out to dilation operation, increased the white portion that represents fruit tree, solved the problem of wrong minute, as shown in Figure 5;
5) utilize largest contours Rectangular Method detecting step 4) in whole the area A that fruit tree largest contours is shared in the image processed, detect the shared pixel quantity of maximum boundary of fruit tree horizontal and vertical directions, then calculate this two border and be the area A of the whole image of fruit tree for long and wide rectangular area; Detect again the area m of leaf and trunk in entire image, in this fruit tree bianry image, the pixel value that fruit tree leaf and trunk are corresponding is 1, and statistical pixel values is that 1 shared pixel count is the shared area m of leaf and trunk in entire image, according to formula (1), calculates the rare close degree η of leaf:
In formula (1), η represents the rare density of fruit tree leaf, and m represents the shared area of trunks of fruit trees and leaf in this width figure; A represents the area of the whole image of fruit tree.The area of A is to become according to the size of fruit tree reality, and therefore, m and A are variable values, only relevant with the upgrowth situation of fruit tree self; Detect every a line, each row in Fig. 5 and occur that for the first time and for the last time pixel value is 1 difference, if the shared vertical direction maximum pixel of fruit tree image spacing is 1(max), find out the maximal value in row difference and row difference, the product that two maximal values add respectively after 1 is the shared largest contours area of fruit tree image, and in the present embodiment, the result of calculation of the shared largest contours area of fruit tree image is 231336; The pixel number that detection Fig. 5 intermediate value is 1 is the shared area of leaf trunk of fruit tree; In the present embodiment, the shared area result of calculation of the leaf trunk of fruit tree is 133487.
5) the rare close degree value of fruit tree leaf is the business of the shared largest contours area of the shared area of fruit tree leaf trunk and fruit tree, and result is 0.577.
Claims (3)
1. adopt largest contours Rectangular Method to detect a method for the rare close degree of fruit tree leaf, it is characterized in that: comprise the steps:
1) by the recognition technology of image, utilize super green method to be combined with Otsu algorithm, identify fruit tree binary image clearly;
2) the fruit tree binary image in step 1) is carried out to medium filtering denoising,
3) by step 2) in denoising image by opening computing, impurity is eliminated in continuous 4 corrosion and 4 expansions respectively;
4) image step 3) being obtained carries out dilation operation processing, with the cavity in removal of images;
5) the imagery exploitation largest contours Rectangular Method of processing through step 4) is detected to its shared largest contours, comprise the length of fruit tree image and wide, then with this, calculate the contour area of fruit tree image, and calculate the shared area of leaf trunk of fruit tree in image;
6) utilize the shared area of the leaf of the fruit tree that step 4) calculates to be the rare close degree of fruit tree leaf divided by the value of the contour area of fruit tree image.
2. employing largest contours Rectangular Method according to claim 1 detects the method for the rare close degree of fruit tree leaf, it is characterized in that: step 2) described in morphology process specifically and refer to: the fruit tree image after medium filtering is first carried out to 4 times continuously and corrodes, then carry out continuously 4 times and expand, corrosion is all identical with the structural element expanding.
3. employing largest contours Rectangular Method according to claim 1 detects the method for the rare close degree of fruit tree leaf, it is characterized in that: described in step 4), utilize largest contours Rectangular Method to detect its shared largest contours specifically to refer to, by detecting step 3) in whole the area A that fruit tree largest contours is shared in the image processed, detect the shared pixel quantity of maximum boundary of fruit tree horizontal and vertical directions, then calculate this two border and be the area A of the whole image of fruit tree for long and wide rectangular area; Detect again the area m of leaf and trunk in entire image, in this fruit tree bianry image, the pixel value that fruit tree leaf and trunk are corresponding is 1, and statistical pixel values is that 1 shared pixel count is the shared area m of leaf and trunk in entire image, according to formula (1), calculates the rare close degree η of leaf:
In formula (1), η represents the rare density of fruit tree leaf, and m represents the shared area of trunks of fruit trees and leaf in this width figure; A represents the area of the whole image of fruit tree.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410246400.0A CN104132650A (en) | 2014-06-05 | 2014-06-05 | Method for detection of fruit tree leaf density by maximum contour rectangle technique |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410246400.0A CN104132650A (en) | 2014-06-05 | 2014-06-05 | Method for detection of fruit tree leaf density by maximum contour rectangle technique |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104132650A true CN104132650A (en) | 2014-11-05 |
Family
ID=51805424
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410246400.0A Pending CN104132650A (en) | 2014-06-05 | 2014-06-05 | Method for detection of fruit tree leaf density by maximum contour rectangle technique |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104132650A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106614484A (en) * | 2017-01-19 | 2017-05-10 | 成都市宏德永兴养殖有限公司 | Vehicle-mounted fruit tree pesticide spraying method |
| CN109068597A (en) * | 2016-02-04 | 2018-12-21 | 菲特杰夫有限公司 | Horticultural light device |
| CN110309762A (en) * | 2019-06-26 | 2019-10-08 | 扆亮海 | A kind of forestry health assessment system based on air remote sensing |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060221417A1 (en) * | 2005-03-15 | 2006-10-05 | Omron Corporation | Image processing method, three-dimensional position measuring method and image processing apparatus |
| CN101298986A (en) * | 2007-04-30 | 2008-11-05 | 中国科学院沈阳应用生态研究所 | Field measuring method of sand broad leaf plant leaf area |
| CN104008365A (en) * | 2014-03-31 | 2014-08-27 | 贵州大学 | Method for detecting sparse degree of fruit tree leaves based on image processing technology |
-
2014
- 2014-06-05 CN CN201410246400.0A patent/CN104132650A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060221417A1 (en) * | 2005-03-15 | 2006-10-05 | Omron Corporation | Image processing method, three-dimensional position measuring method and image processing apparatus |
| CN101298986A (en) * | 2007-04-30 | 2008-11-05 | 中国科学院沈阳应用生态研究所 | Field measuring method of sand broad leaf plant leaf area |
| CN104008365A (en) * | 2014-03-31 | 2014-08-27 | 贵州大学 | Method for detecting sparse degree of fruit tree leaves based on image processing technology |
Non-Patent Citations (1)
| Title |
|---|
| 张富贵 等: ""基于图像处理技术的果树树叶稀密程度的检测"", 《山地农业生物学报》, vol. 32, no. 6, 31 December 2013 (2013-12-31) * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109068597A (en) * | 2016-02-04 | 2018-12-21 | 菲特杰夫有限公司 | Horticultural light device |
| CN109068597B (en) * | 2016-02-04 | 2022-04-08 | 菲特杰夫有限公司 | Gardening lighting device |
| CN106614484A (en) * | 2017-01-19 | 2017-05-10 | 成都市宏德永兴养殖有限公司 | Vehicle-mounted fruit tree pesticide spraying method |
| CN110309762A (en) * | 2019-06-26 | 2019-10-08 | 扆亮海 | A kind of forestry health assessment system based on air remote sensing |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105740773A (en) | Deep learning and multi-scale information based behavior identification method | |
| CN103310218B (en) | A kind of overlap blocks fruit precise recognition method | |
| CN103218605A (en) | Quick eye locating method based on integral projection and edge detection | |
| Wang et al. | DSE-YOLO: Detail semantics enhancement YOLO for multi-stage strawberry detection | |
| CN103914836B (en) | Farmland machinery guidance line drawing algorithm based on machine vision | |
| CN103164694A (en) | Method for recognizing human motion | |
| CN108960011B (en) | Partially-shielded citrus fruit image identification method | |
| Zheng et al. | Research on tomato detection in natural environment based on RC-YOLOv4 | |
| CN105335967A (en) | Back-of-hand vein line extraction method based on local maximum between-class variance and mathematical morphology | |
| CN103870808A (en) | Finger vein identification method | |
| CN104268853A (en) | Infrared image and visible image registering method | |
| CN104361330A (en) | Crop row identification method for precise corn pesticide application system | |
| CN102542560B (en) | Method for automatically detecting density of rice after transplantation | |
| CN105117701A (en) | Corn crop row skeleton extraction method based on largest square principle | |
| CN103177257A (en) | Image identification method and image classification method for coprinus comatus | |
| CN108090485A (en) | Display foreground extraction method based on various visual angles fusion | |
| CN112990103A (en) | String mining secondary positioning method based on machine vision | |
| CN103729621B (en) | Plant leaf image automatic recognition method based on leaf skeleton model | |
| CN103914829B (en) | Method for detecting edge of noisy image | |
| CN103226824A (en) | Video retargeting system for maintaining visual saliency | |
| CN104132650A (en) | Method for detection of fruit tree leaf density by maximum contour rectangle technique | |
| Saravanakumar et al. | Plant syndrome recognition by Gigapixel Image using convolutional neural network | |
| CN107169932A (en) | A kind of image recovery method based on Gauss Poisson mixed noise model suitable for neutron imaging system diagram picture | |
| CN104008365A (en) | Method for detecting sparse degree of fruit tree leaves based on image processing technology | |
| Ma et al. | YOLOv5-lotus an efficient object detection method for lotus seedpod in a natural environment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into 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: 20141105 |