CN102622755B - Plant limb identification method - Google Patents
Plant limb identification method Download PDFInfo
- Publication number
- CN102622755B CN102622755B CN201210049195.XA CN201210049195A CN102622755B CN 102622755 B CN102622755 B CN 102622755B CN 201210049195 A CN201210049195 A CN 201210049195A CN 102622755 B CN102622755 B CN 102622755B
- Authority
- CN
- China
- Prior art keywords
- image
- segmentation
- aberration
- segmentation image
- limb
- 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.)
- Expired - Fee Related
Links
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a plant limb identification method, which includes the following steps: A. obtaining an R-G chromatic aberration image of a collected image; B. utilizing a threshold method to obtain a segmentation image of the R-G chromatic aberration image; C. performing initial denoising on the segmentation image of the R-G chromatic aberration image by using mathematical morphology and area features, and obtaining an initially-denoised chromatic aberration segmentation image; D. processing the initially-denoised chromatic aberration segmentation image respectively according to front-lighting or back-lighting conditions of the collected image; E. performing Hough transformation line detection on the initially-denoised chromatic aberration segmentation image; and F. calculating shape features of different areas of line segmentation obtained through detection of the image, identifying the collected image according to the shape features, and finally obtaining a limb target. The plant limb identification method can identify plant limbs accurately, and can avoid problems such as device damage or fruit tree damage caused by obstacles of the limbs when an apple picking robot performs picking operation.
Description
Technical field
The present invention relates to technical field of image processing, particularly relate to a kind of plant limb identification method.
Background technology
Apple, as the topmost fruit tree species of China, accounts for 20% of the fruit tree total area.In apple production operation, results are plucked and are accounted for 40% of whole workload.The quality of plucking operation quality directly has influence on the storage of apple, processing and sale, thus finally affects the market price and economic benefit.Meanwhile, harvest operation is again an operation that labour intensity is large, elapsed time long, work is dull.Along with Rural Urbanization Construction and people in the countryside are to a large amount of transfers in city, easily occur the phenomenon of short-term manpower shortage in the ripe apples phase, the harvesting task adopting apple picking robot to complete apple has good application prospect.
Picking robot technology mainly comprises three aspects: identify location technology, picked technology and mobile technology.Wherein recognition technology comprises identification to ripening fruits and surrounding environment and location, and picked technology is the design of mechanical arm and motion control mainly, and mobile technology mainly refers to robot navigation's technology.
Realize apple picking robot to pluck apple, the link of most critical be to obtain apple the three-dimensional coordinate position in space and and fruit around the identification of the barrier such as limb and location, the i.e. three-dimensional reconstruction of apple tree, thus provide space position parameter accurately for mechanical arm, improve the harvesting quality of apple.Vision system is one of difficult point of picking robot research, and the accuracy of the three-dimensional reconstruction of apple tree is related to the picking efficiency of picking robot and plucks quality.
In recent years, researchist both domestic and external is studied, as citrus, apple and peach etc. the identification under field conditions (factors) of various fruits fruit and location.In numerous method, applying maximum is method based on machine vision, namely adopts camera acquisition image, utilizes the equipment such as computing machine, dsp chip to process image, identify and location fruit.Above-mentioned achievement in research alleviates the impact of available light to a certain extent, achieves identification and the location of fruit under natural conditions.
But in harvesting operation process, mechanical arm and mechanical arm can with the limb of fruit tree, orchard in support " barriers " such as the Metallic rod of fruit tree and iron wires and collide, thus make device damage or injury fruit tree.Therefore, only can not meet the needs plucking operation to the location of fruit, need equally identify these " barrier " and locate.
The aberration R-B image of the limb to citrus is adopted to carry out Ostu (maximum variance between clusters) adaptive threshold fuzziness in prior art, limb can be identified, but when light direct projection, illumination solar flare causes image segmentation obviously to decline, poor to the recognition effect of limb.
Summary of the invention
(1) technical matters that will solve
The technical problem to be solved in the present invention is: provide a kind of plant limb identification method, it can identify the limb of plant exactly, and apple picking robot can be made when plucking operation to avoid the obstruction because of limb to cause the generation of the problem such as device damage or injury fruit tree.
(2) technical scheme
For solving the problem, the invention provides a kind of plant limb identification method, comprising the following steps:
A: the R-G color difference image asking for the image collected;
B: utilize threshold method to obtain the segmentation image of described R-G color difference image;
C: utilize mathematical morphology and the segmentation image of area features to described R-G color difference image to carry out preliminary denoising, obtains the aberration segmentation image of preliminary denoising;
D: respectively according to described in light or the backlight situation aberration segmentation image to described preliminary denoising of meeting of image that collects process;
E: Hough transform straight-line detection is carried out to the aberration segmentation image of the described preliminary denoising that backlight situation obtains;
F: the aberration segmentation image calculating described preliminary denoising is detected the shape facility of the zones of different of the line segmentation obtained, and identifies the described image collected according to the shape facility of described zones of different, obtains final limb target.
Preferably, described steps A comprises further: the step of the value of the described image collected R passage in RGB color space and the value of G passage being subtracted each other.
Preferably, described step B comprises further: if the pixel value of R-G color difference image is between 0 ~ 10, this pixel value is set to 255, otherwise, this pixel value is set to the step of 0.
Preferably, described step D comprises further: if described in the image that collects for meeting light image, then the aberration segmentation image of described preliminary denoising is not processed, if described in the image that collects be backlight image, then adopt the edge of the aberration segmentation image of preliminary denoising described in canny operator extraction, obtain its edge image.
Preferably, in described step e, when carrying out Hough transform straight-line detection, angle step and the distance step-length of Hough transform are respectively 1 degree and 1 pixel, detect 5 straight lines that totalizer value is maximum.
Preferably, described step F comprises further: straight line region is identified as limb, utilizes form parameter to by other region of line segmentation
and excentricity
identify, meet F and be greater than 2.5 and the region recognition that e is greater than 4 is the step of limb, wherein, A and B is respectively the area and perimeter in region, c and a is respectively border long axis length and the minor axis length in region.
(3) beneficial effect
The present invention is by carrying out Threshold segmentation to the aberration R-B image of diseases on plant stalk, and according to meet light or the backlight situation of image, denoising is carried out to segmentation image, the limb of plant can being identified exactly, making apple picking robot avoid when plucking operation the obstruction because of limb to cause the generation of the problem such as device damage or injury fruit tree.
Accompanying drawing explanation
Fig. 1 is the process flow diagram of plant limb identification method described in embodiment of the present invention.
Embodiment
Below in conjunction with drawings and Examples, the specific embodiment of the present invention is described in further detail.Following examples for illustration of the present invention, but are not used for limiting the scope of the invention.
As shown in Figure 1, a kind of plant limb identification method of the present invention, comprises the following steps:
A: the R-G color difference image asking for the image collected;
In this step, the value of the described image collected R passage in RGB color space and the value of G passage are subtracted each other.
B: utilize threshold method to obtain the segmentation image of described R-G color difference image;
In this step, if the pixel value of R-G color difference image is between 0 ~ 10, this pixel value is set to 255, otherwise, this pixel value is set to 0.
C: utilize mathematical morphology and the segmentation image of area features to described R-G color difference image to carry out preliminary denoising, obtains the aberration segmentation image of preliminary denoising;
In this step, the length of side first can be adopted to be that the square structure element of 5 pixels carries out open and close operator twice to described R-G aberration segmentation image, to remove small noise spot.Zone marker is carried out to the R-G aberration segmentation image after removing small noise spot, and calculates the area of regional respectively, remove the region that area is less than maximum region area 1/50, obtain the aberration segmentation image of preliminary denoising.
D: respectively according to described in light or the backlight situation aberration segmentation image to described preliminary denoising of meeting of image that collects process;
In this step, if described in the image that collects for meeting light image, target (pixel value is 255 parts) in the aberration segmentation image of then obtained preliminary denoising is limb, remainder is background, the aberration segmentation image of described preliminary denoising is not processed, if described in the image that collects be backlight image, more noise is still comprised in the aberration segmentation image of described preliminary denoising, then adopt the edge of the aberration segmentation image of preliminary denoising described in canny operator extraction, obtain its edge image.
E: Hough transform straight-line detection is carried out to the aberration segmentation image of the described preliminary denoising that backlight situation obtains;
In this step, when carrying out Hough transform straight-line detection, angle step and the distance step-length of Hough transform are respectively 1 degree and 1 pixel, detect 5 straight lines that totalizer value is maximum.
In described 5 straight lines, the straight line corresponding to maximum 2 values got by totalizer, it is different regions that the aberration of described preliminary denoising is split Iamge Segmentation, described 5 straight line regions are limb, wherein the totalizer region of getting corresponding to maximum 2 values between straight line is trunk, and other 3 straight line regions are branch.Outside described 5 straight line regions, there is limb and noise.
F: the aberration segmentation image calculating described preliminary denoising is detected the shape facility of the zones of different of the line segmentation obtained, and identifies the described image collected according to the shape facility of described zones of different, obtains final limb target.
In this step, described shape facility has multiple parameter, and the present embodiment only calculates form parameter and excentricity, first the form parameter of defined range and excentricity, form parameter
wherein A and B is respectively the area and perimeter in region; Excentricity
wherein c and a is respectively border long axis length and the minor axis length in region, and major axis refers to distance 2 lines farthest on zone boundary; Zone boundary is called minor axis with the longest line segment in the line of long axis normal.
Straight line region is identified as limb, utilizes form parameter to by other region of line segmentation
and excentricity
identify, meet F and be greater than 2.5 and the region recognition that e is greater than 4 is limb.
Above embodiment is only for illustration of the present invention; and be not limitation of the present invention; the those of ordinary skill of relevant technical field; without departing from the spirit and scope of the present invention; can also make a variety of changes and modification; therefore all equivalent technical schemes also belong to category of the present invention, and scope of patent protection of the present invention should be defined by the claims.
Claims (3)
1. a plant limb identification method, is characterized in that, comprises the following steps:
A: the R-G color difference image asking for the image collected;
B: utilize threshold method to obtain the segmentation image of described R-G color difference image;
C: utilize mathematical morphology and the segmentation image of area features to described R-G color difference image to carry out preliminary denoising, obtains the aberration segmentation image of preliminary denoising;
D: respectively according to described in light or the backlight situation aberration segmentation image to described preliminary denoising of meeting of image that collects process;
E: Hough transform straight-line detection is carried out to the aberration segmentation image of the described preliminary denoising that backlight situation obtains;
F: the aberration segmentation image calculating described preliminary denoising is detected the shape facility of the zones of different of the line segmentation obtained, and identifies the described image collected according to the shape facility of described zones of different, obtains final limb target;
Wherein, described step B comprises further: if the pixel value of R-G color difference image is between 0 ~ 10, this pixel value is set to 255, otherwise, this pixel value is set to the step of 0;
Wherein, described step C specifically comprises: the employing length of side is that the square structure element of 5 pixels carries out open and close operator twice to described R-G aberration segmentation image, removes small noise spot; Zone marker is carried out to the R-G aberration segmentation image after removing small noise spot, and calculates the area of regional respectively, remove the region that area is less than maximum region area 1/50, obtain the aberration segmentation image of preliminary denoising;
Wherein, described step D comprises further: if described in the image that collects for meeting light image, then the aberration segmentation image of described preliminary denoising is not processed, if described in the image that collects be backlight image, then adopt the edge of the aberration segmentation image of preliminary denoising described in canny operator extraction, obtain its edge image;
Described step F comprises further: straight line region is identified as limb, utilizes form parameter to by other region of line segmentation
and excentricity
identify, meet F and be greater than 2.5 and the region recognition that e is greater than 4 is the step of limb, wherein, A and B is respectively the area and perimeter in region, c and a is respectively border long axis length and the minor axis length in region.
2. plant limb identification method according to claim 1, is characterized in that, described steps A comprises further: the step of the value of the described image collected R passage in RGB color space and the value of G passage being subtracted each other.
3. plant limb identification method according to claim 1, it is characterized in that, in described step e, when carrying out Hough transform straight-line detection, angle step and the distance step-length of Hough transform are respectively 1 degree and 1 pixel, detect 5 straight lines that totalizer value is maximum.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210049195.XA CN102622755B (en) | 2012-02-28 | 2012-02-28 | Plant limb identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210049195.XA CN102622755B (en) | 2012-02-28 | 2012-02-28 | Plant limb identification method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102622755A CN102622755A (en) | 2012-08-01 |
| CN102622755B true CN102622755B (en) | 2015-01-07 |
Family
ID=46562656
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210049195.XA Expired - Fee Related CN102622755B (en) | 2012-02-28 | 2012-02-28 | Plant limb identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102622755B (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103226709B (en) * | 2013-04-24 | 2015-12-02 | 聊城大学 | A kind of network curtain image recognition method of fall webworm larvae |
| CN103870816B (en) * | 2014-03-26 | 2016-11-23 | 中国科学院寒区旱区环境与工程研究所 | The method of the plants identification that a kind of discrimination is high |
| CN103996197A (en) * | 2014-05-30 | 2014-08-20 | 北京农业信息技术研究中心 | Character collecting method and system |
| CN106776675B (en) * | 2016-11-01 | 2020-12-22 | 吴建伟 | Plant identification method and system based on mobile terminal |
| CN106824651B (en) * | 2017-04-11 | 2023-07-14 | 极昊供应链(上海)有限公司 | Paint spraying and baking device and method for mobile automobile maintenance and repair service |
| CN107993243B (en) * | 2017-12-21 | 2020-06-23 | 北京林业大学 | Wheat tillering number automatic detection method based on RGB image |
| CN108901540A (en) * | 2018-06-28 | 2018-11-30 | 重庆邮电大学 | Fruit tree light filling and fruit thinning method based on artificial bee colony fuzzy clustering algorithm |
| CN109522901B (en) * | 2018-11-27 | 2020-11-03 | 中国计量大学 | Tomato plant stem edge identification method based on edge dual relation |
| CN110211133B (en) * | 2019-05-27 | 2021-01-15 | 中国农业大学 | Method and device for obtaining safety protection strategy of tree with leaves and electronic equipment |
| CN114027052A (en) * | 2021-10-20 | 2022-02-11 | 华南农业大学 | Illumination regulation and control system for plant reproductive development |
| CN113989253A (en) * | 2021-11-04 | 2022-01-28 | 广东皓行科技有限公司 | Farmland target object information acquisition method and device |
| CN114102591B (en) * | 2021-11-24 | 2023-04-07 | 北京市农林科学院智能装备技术研究中心 | Operation method and device for agricultural robot mechanical arm |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101216894A (en) * | 2008-01-08 | 2008-07-09 | 北京林业大学 | A branch identification method based on standing tree occlusion of single image |
| CN102036016A (en) * | 2010-12-03 | 2011-04-27 | 杭州海康威视系统技术有限公司 | Video detection method and system thereof |
-
2012
- 2012-02-28 CN CN201210049195.XA patent/CN102622755B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101216894A (en) * | 2008-01-08 | 2008-07-09 | 北京林业大学 | A branch identification method based on standing tree occlusion of single image |
| CN102036016A (en) * | 2010-12-03 | 2011-04-27 | 杭州海康威视系统技术有限公司 | Video detection method and system thereof |
Non-Patent Citations (3)
| Title |
|---|
| RGB与HIS颜色空间下番茄图像分割的对比研究;尹建军等;《农机化研究》;20061130;摘要 * |
| 基于整枝抚育目的的立木枝干自动识别研究;孙仁山等;《北京林业大学学报》;20050731;87页左栏－89页左栏 * |
| 柑橘采摘机器人成熟果实定位及障碍物检测研究;周小军;《中国优秀硕士学位论文全文数据库 信息科技辑》;20090930;第54－56页，图4-10 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102622755A (en) | 2012-08-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102622755B (en) | Plant limb identification method | |
| Longsheng et al. | Kiwifruit recognition at nighttime using artificial lighting based on machine vision | |
| Suo et al. | Improved multi-classes kiwifruit detection in orchard to avoid collisions during robotic picking | |
| Silwal et al. | Apple identification in field environment with over the row machine vision system | |
| CN103279762B (en) | Common growth form of fruit decision method under a kind of physical environment | |
| CN201600330U (en) | System for recognizing and locating mature pineapples | |
| CN103295018A (en) | Method for precisely recognizing fruits covered by branches and leaves | |
| CN110599507B (en) | Tomato identification and positioning method and system | |
| CN105719282A (en) | Fruit, branch and leaf region obtaining method of red apple images in garden | |
| Changyi et al. | Apple detection from apple tree image based on BP neural network and Hough transform | |
| Ji et al. | Research on key technology of truss tomato harvesting robot in greenhouse | |
| Chen et al. | Segmentation of field grape bunches via an improved pyramid scene parsing network | |
| Jin et al. | Far-near combined positioning of picking-point based on depth data features for horizontal-trellis cultivated grape | |
| CN103226709B (en) | A kind of network curtain image recognition method of fall webworm larvae | |
| Moghimi et al. | A robo-vision algorithm for automatic harvesting of green bell pepper | |
| CN104992448A (en) | Automatic positioning method for robot picking grapes in a damage-free way | |
| CN108090910B (en) | Night outdoor tomato plant image segmentation algorithm based on information entropy gradient simplified PCNN model | |
| Bortolotti et al. | A computer vision system for in-field quality evaluation: Preliminary results on peach fruit | |
| Longye et al. | Overlapping citrus segmentation and reconstruction based on mask R-CNN model and concave region simplification and distance analysis | |
| CN114347044B (en) | Pepper picking method and device, electronic equipment and storage medium | |
| Quan et al. | Selecting candidate regions of clustered tomato fruits under complex greenhouse scenes using RGB-D data | |
| CN115937314A (en) | Camellia oleifera fruit growth posture detection method | |
| Xiao-Lian et al. | Identification and location of picking tomatoes based on machine vision | |
| Nejati et al. | Kiwifruit detection in challenging conditions | |
| Xiang et al. | Tests of a recognition algorithm for clustered tomatoes based on mathematical morphology |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20150107 Termination date: 20190228 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |