CN109588114A - A kind of parallelism recognition picker system and method applied to fruit picking robot - Google Patents
A kind of parallelism recognition picker system and method applied to fruit picking robot Download PDFInfo
- Publication number
- CN109588114A CN109588114A CN201811562710.8A CN201811562710A CN109588114A CN 109588114 A CN109588114 A CN 109588114A CN 201811562710 A CN201811562710 A CN 201811562710A CN 109588114 A CN109588114 A CN 109588114A
- Authority
- CN
- China
- Prior art keywords
- image
- module
- data
- lamp
- host computer
- 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
Classifications
-
- A—HUMAN NECESSITIES
- A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
- A01D—HARVESTING; MOWING
- A01D46/00—Picking of fruits, vegetables, hops, or the like; Devices for shaking trees or shrubs
- A01D46/30—Robotic devices for individually picking crops
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/68—Food, e.g. fruit or vegetables
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Robotics (AREA)
- Environmental Sciences (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
- Manipulator (AREA)
Abstract
The invention discloses a kind of parallelism recognition picker systems and method applied to fruit picking robot, including upper computer and lower computer；Host computer includes images match identification module, image segmentation module, data processing module, path planning decision-making module, first communication module；Host computer is used to receive the image of slave computer upload, and carries out images match identification, segmentation, the positioning of picking position, path planning decision, and by path planning decision feedback to slave computer；Slave computer includes video image acquisition module, second communication module, control module, slave computer is for acquiring image, host computer is uploaded the images to, and receives the information of host computer feedback, realizes travelling control, mechanical arm space orientation control and the crawl of the end effector control of picking robot；Each module has independent calculating section, can complete independently control work accordingly, and can be realized each module collaboration processing.Scalability of the present invention is good, and the effect of parallelization is good.
Description
Technical field
The invention belongs to image identification technical fields, are related to a kind of parallelism recognition method, and in particular to one kind is applied to water
The parallelism recognition picker system and method for fruit picking machine device people.
Background technique
Although the identification picked technology of picking robot is difficult to realize currently, picking robot utilization is more and more wider
Parallelization strategies, picking process is slow at the same time is difficult to put into really production；And parallel (the prediction model of software module
Decision tree, image segmentation and matching) cannot achieve parallelization strategies, control section be also difficult to realize parallel control (running gear,
Mechanical arm positioning and end effector).
Summary of the invention
In order to solve the above-mentioned technical problems, the present invention provides a kind of parallelism recognitions applied to fruit picking robot to adopt
Pluck system and method.
Technical solution used by system of the invention is: a kind of parallelism recognition picking applied to fruit picking robot
System, it is characterised in that: including upper computer and lower computer；
The host computer includes images match identification module, image segmentation module, data processing module, path planning decision
Module, first communication module；The host computer be used for receive slave computer upload image, and carry out images match identification, segmentation,
Position positioning, path planning decision are picked, and gives path planning decision feedback to the slave computer；Wherein each module has solely
Vertical calculating section, can complete independently control work accordingly, and can be realized each module collaboration processing；
The slave computer includes video image acquisition module, second communication module, control module, and the slave computer is for adopting
Collect image, upload the images to the host computer, and receive the information of the host computer feedback, realizes the walking control of picking robot
System, mechanical arm space orientation control and the crawl of end effector control；Wherein each module has independent calculating section,
Can complete independently control work accordingly, and can be realized each module collaboration processing.
Technical solution used by method of the invention is: a kind of parallelism recognition picking applied to fruit picking robot
Method, which comprises the following steps:
Step 1: host computer receives the image of slave computer video image acquisition module acquisition；
Step 2: images match identification module carries out images match identification；
Step 3: image segmentation module is split matched image；
Step 4: path planning decision-making module carries out path decision and planning to picking robot；
Step 5: path decision and planning are fed back to slave computer by first communication module by data processing module；
Step 6: slave computer receives the information that host computer is fed back by second communication module；
Step 7: control module controls picking robot execution route decision and planning.
Compared to the prior art, the present invention have following features and the utility model has the advantages that
1. Scalable Performance is good, independent individual robot number that slave computer individual is constituted can according to job costs and
Task needs flexible setting；
2. clear in structure, it is easy to design and build；
3. data volume compares concentration, the effect of the parallelization of software is good；
4. building, module is complete, can carry out different modules according to different fruit is picked and call, have good
Versatility.
Detailed description of the invention
Fig. 1 is the picking robot concurrent technique route map of the embodiment of the present invention；
Fig. 2 is the method flow diagram of the embodiment of the present invention；
Fig. 3 is the matching process schematic diagram of the embodiment of the present invention；
Fig. 4 is the matching process flow chart of the embodiment of the present invention；
Fig. 5 be the embodiment of the present invention segmentation before decision flow chart is carried out to matched image；
Fig. 6 is the image partition method flow chart of the embodiment of the present invention；
Fig. 7 carries out piecemeal and task allocation schedule flow chart to data by decision tree for the embodiment of the present invention；
Fig. 8 is the parallel flow chart of the embodiment of the present invention, and s1, s2, s3 are respectively the logic control circuit of slave computer in figure
Switch, L1, L2, L3 are respectively the signal lamp that circuit is connected.
Specific embodiment
Understand for the ease of those of ordinary skill in the art and implement the present invention, with reference to the accompanying drawings and embodiments to this hair
It is bright to be described in further detail, it should be understood that implementation example described herein is merely to illustrate and explain the present invention, not
For limiting the present invention.
Referring to Fig.1, a kind of parallelism recognition picker system applied to fruit picking robot provided by the invention, including it is upper
Position machine and slave computer；
Host computer includes images match identification module, image segmentation module, data processing module, path planning decision model
Block, first communication module；Host computer is used to receive the image of slave computer upload, and carries out images match identification, segmentation, picking position
Positioning, path planning decision are set, and by path planning decision feedback to slave computer；Wherein each module has independent calculation part
Point, can complete independently control work accordingly, and can be realized each module collaboration processing；
Slave computer includes video image acquisition module, second communication module, control module, and slave computer is used to acquire image,
Host computer is uploaded the images to, and receives the information of host computer feedback, realizes that the travelling control of picking robot, mechanical arm space are fixed
Position control and the crawl of end effector control；Wherein each module has independent calculating section, can complete independently
Corresponding control work, and can be realized the collaboration processing of each module.
See Fig. 2, the present invention also provides a kind of parallelism recognition picking methods applied to fruit picking robot, including
Following steps:
Step 1: host computer receives the image of slave computer video image acquisition module acquisition；
Step 2: images match identification module carries out images match identification；
See Fig. 3, Fig. 4, what the images match identification module of the present embodiment acquired slave computer video image acquisition module
Image carries out the matching of template, by matching between search window and template, then the size relation between respective pixel will
It is consistent with template；On the contrary, not meeting this condition once, then it is assumed that do not matched that with its template；When image data matching, number
According to being stored in data storage, when image data is unmatched, data are deleted, and re-execute images match mistake
Journey；
Image data result after images match is divided into three classes: image data A, image data B, image data C；Wherein
Image data A is interference image, and image data B is navigation picture, and image data C is fruit figure；
By the method for quick One Dimensional Projection template matching, two-dimensional image is carried out one-dimensional throwing to X-axis and Y-axis respectively
Its one-dimensional projection value is transformed by one group 0,1 number using one-dimensional differential quantization method and is constituted to describe image by shadow
The character string of feature, then matched with template with characteristics of image character string；
When template character string and characteristics of image string matching, image is retained, is at this time image data B or figure
As data C, logic judgment parameter i=1 is taken；When template character string and characteristics of image character string mismatch, image is retained,
It is at this time image data A, takes logic judgment parameter i=0；
Fine match is carried out to image data B or image data C by feature vector again, image is further classified；This
In use feature vector b and feature vector c, wherein feature vector b is feature vector used in matching navigation picture, and feature vector c is
Match feature vector used in navigation picture；Each image is traversed with feature vector b and feature vector c respectively；
Feature vector b traverses image, as navigation characteristic vector b and images match, takes logic judgment parameter j=1；
Feature vector c traverses image, as fruit character vector c and images match, takes logic judgment parameter n=1；
By level-one logic judgment, when judging parameter i=1, regard as matching correlation be classified as image data B or
Image data C when judging parameter i=0, is regarded as matching and uncorrelated is classified as image data A；Two-level logic is carried out again to sentence
It is disconnected, when judging parameter j=1, regards as matching correlation and be classified as image data B, when judging parameter n=1, regard as
Matching correlation is classified as image data C.
Step 3: image segmentation module is split matched image；
See Fig. 5, Fig. 6, the specific implementation of step 3 includes following sub-step:
Step 3.1: matched image being judged before entering segmentation, it is determined whether be required process object；
It is required process object when logical relation is 1, is not required process object when logical relation is 0；
Step 3.2: to the figure of required processing, by the way that the grayscale information of image will be handled, its gradation data being optimized
Processing, the gray value of pixel (i, j) are Gij,0≤Gij≤255；
Processing wherein is optimized to its gradation data, it is assumed that image has n data point { x1, x2, x3......xn, point
For 3 clustering cluster cluster, what cluster to be done is exactly to minimize the image objective function J to be optimized；
When carrying out classification processing to data with objective function J, when data n is classified as a kind of clusterkWhen, majority is according to pass
It is variable Tnk1 is taken, is otherwise 0；
Meet formula when J minimum:
μkValue be all clusterkIn data point average value；Since iteration is all to get the minimum of J each time
Value, therefore J can be only steadily decreasing or constant, without will increase, obtain 3 clusters of gray level image by iteration repeatedly
Point μk1, μk2, μk3, μk1<μk2<μk3。
Step 3.3: n data point X of input1、X2、…、Xn, number k to be clustered；K cluster centre is obtained, in cluster
The heart is screened, and three optimal cluster value C are obtainedk1、Ck2、Ck3, choose (Ck1, Ck3) be optimization tonal range to image into
Row threshold division；
Step 3.4: image being split using threshold value, is obtained by the comparison of threshold value to judge whether this threshold value is most
Good segmentation threshold, just threshold value is σ here, by itself and cluster centre Ck2It is compared, obtains a judgment value W, closed by logic
System's judgement is split if logical relation is 1 by σ, if logical relation is 0, is split by optimizing threshold value；
In the present embodiment, if the pixel cluster center that the gray level [0,255] of piece image is obtained by top, take image grey
Degree grade is [μk1,μk3]；
It is C the pixel threshold value w in selected figure points0And C1Two classes, C0By gray value in [μk1, w] pixel composition, C1
Having gray value is [w, μk3] pixel composition, region C0And C1Average gray can seek μ0、μ1；P0For C0The probability in region, P1For
C1The probability in region；
μ is image in [μk1μk3] average gray:
σ2=p0(μ0-μ)2+p1(μ1-μ)2=p0p1(μ0-μ1)2
Allow K in [μk1 μk3] successively value in section, so that σ2Maximum K value is best region segmentation threshold.
K value will be acquired and cluster obtained cluster value μk2It is compared, if meeting formula:
Meet segmentation range threshold requirement；
If w≤σ, assignment 1 takes μk2For optimal segmenting threshold,
Otherwise assignment 0, taking K is optimal segmenting threshold.
Step 3.5: image data after segmentation being carried out to be stored in data storage, includes at least by the image after segmentation
Three kinds of image informations are classified as data information 1: fruit coordinate, data information 2: fruit number, data information 3: fruit size；
These three types of data informations are stored respectively, are remained in database, so as to subsequent robot's data call.
Step 4: path planning decision-making module carries out path decision and planning to picking robot；
After the present embodiment is by being sorted out to image data after segmentation and learning processing, by decision tree, to data into
Row piecemeal, task allocation schedule determine the following path of motion decision of picking robot and planning.
See Fig. 7, navigational parameter: lateral deviation λ and navigation deviation θ can get after image data B, identification segmentation；When
Image data C can get fruit sequence number n, fruit radius r and fruit coordinate (x, y, z) after identification segmentation；
Step 5: path decision and planning are fed back to slave computer by first communication module by data processing module；
Step 6: slave computer receives the information that host computer is fed back by second communication module；
Step 7: control module controls picking robot execution route decision and planning.
See Fig. 8, the specific implementation of step 7 the following steps are included:
Step 7.1: the video image acquisition module of slave computer by image transmitting to the process of host computer, by control circuit S0
Control transmits a low level signal 0 when Image Acquisition finishes, and S0 is connected at this time, and lamp L0 is shinny, starts to transmit pictorial information,
When image transmitting is completed, a high level signal 1 is received, S0 is disconnected at this time；
Step 7.2:S1 is connected, and lamp L1 shines, and host computer transmission control information waits control instruction to execute to running gear
When completion, a high level signal 1 is received, S1 is disconnected at this time；
Step 7.3:S2 is connected, and lamp L2 shines, and host computer transmission control information is positioned to mechanical arm, and control instruction is waited to hold
When row is completed, a high level signal 1 is received, S2 is disconnected at this time；
Step 7.4:S3 is connected, and lamp L3 shines, and host computer transmission control information waits control instruction to hold to end effector
When row is completed, a high level signal 1 is received, S3 is disconnected at this time；
Step 7.5:S0, S1, S2, S3 are connected, and lamp L0, L1, L2, L3 shine, and are waited control instruction to execute completion, are connect respectively
A high level signal 1 is received, S0, S1, S2, S3 are disconnected, and lamp L0, L1, L2, L3 extinguish；
Step 7.6:S0, S1, S3 are connected, and lamp L0, L1, L3 shine, and are waited control instruction to execute completion, are respectively received one
A high level signal 1, S0, S1, S3 are disconnected, and lamp L0, L1, L3 extinguish；
Step 7.7:S0, S2, S3 are connected, and lamp L0, L2, L3 shine, and are waited control instruction to execute completion, are respectively received one
A high level signal 1, S0, S2, S3 are disconnected, and lamp L0, L2, L3 extinguish；
Step 7.8:S0, S3 is connected, and lamp L0, L3 shine, and control instruction is waited to execute completion, is respectively received one high electricity
Ordinary mail number 1, S0, S3 are disconnected, and lamp L0, L3 extinguish；
Wherein, S0 is the control switch of image transmission circuit, and L0 is that image transmission circuit connects display lamp；S1 is walking dress
The control switch of circuits, L1 are that running gear circuit connects display lamp；S2 is the control switch of mechanical arm spatial location circuitry,
L2 is that mechanical arm spatial location circuitry connects display lamp；S3 is the control switch of end effector apparatus circuit, and L3 is that end is held
Row device device circuit connects display lamp.
In the present embodiment, fruit picking robot picks apple on apple tree between agriculture, video image acquisition mould
Block acquires two image informations, and piece image is orchard route map, and another width is fruit hum pattern, is accompanied with 5 on fruit tree
Single fruit, 1 pair of company fruit and 2 fruits blocked by branches and leaves, there are one fruits not in picking range.Robot passes through acquisition
Image information takes opposite decision rule.
In actual use: by the connection of image transmission circuit S0, L0 is lighted, and transmits and obtain image information, image
With image information playbacks three classes, data A is impurity figure, and data B is navigation picture, and data C is fruit figure；Image partition method is logical
It crosses and is handled navigation picture to obtain navigation information, transmit information to slave computer, running gear circuit S1 is connected at this time, L1 point
It is bright, robot ambulation to fruit tree side；Image partition method obtains fruit information by being handled fruit figure, is calculated by image
Method obtains single fruit, Lian Guo, branches and leaves and blocks fruit, plucks the spatial positional information less than fruit, transmits information to mechanical arm space orientation
Device, mechanical arm spatial location circuitry S2 is connected at this time, and L2 is lighted, positioned to mechanical arm；When mechanical arm reaches picking position
End effector apparatus circuit S3 connection, L3 at this time is set to light, pick fruit.Can certainly parallel control, respectively
Circuit S1, S2, S3, S4 are controlled, process is similar with above-mentioned Serial Control.
It should be understood that the part that this specification does not elaborate belongs to the prior art.
It should be understood that the above-mentioned description for preferred embodiment is more detailed, can not therefore be considered to this
The limitation of invention patent protection range, those skilled in the art under the inspiration of the present invention, are not departing from power of the present invention
Benefit requires to make replacement or deformation under protected ambit, fall within the scope of protection of the present invention, this hair
It is bright range is claimed to be determined by the appended claims.
Claims (9)
1. a kind of parallelism recognition picker system applied to fruit picking robot, it is characterised in that: including host computer and bottom
Machine；
The host computer includes images match identification module, image segmentation module, data processing module, path planning decision model
Block, first communication module；The host computer is used to receive the image of slave computer upload, and carries out images match identification, divide, adopt
Position positioning, path planning decision are plucked, and gives path planning decision feedback to the slave computer；Wherein each module has independence
Calculating section, can complete independently control work accordingly, and can be realized each module collaboration processing；
The slave computer includes video image acquisition module, second communication module, control module, and the slave computer is for acquiring figure
Picture uploads the images to the host computer, and receives the information of host computer feedback, realize picking robot travelling control,
Mechanical arm space orientation control and the crawl of end effector control；Wherein each module has independent calculating section, all
Can complete independently control work accordingly, and can be realized each module collaboration processing.
2. a kind of parallelism recognition picking method applied to fruit picking robot, which comprises the following steps:
Step 1: host computer receives the image of slave computer video image acquisition module acquisition；
Step 2: images match identification module carries out images match identification；
Step 3: image segmentation module is split matched image；
Step 4: path planning decision-making module carries out path decision and planning to picking robot；
Step 5: path decision and planning are fed back to slave computer by first communication module by data processing module；
Step 6: slave computer receives the information that host computer is fed back by second communication module；
Step 7: control module controls picking robot execution route decision and planning.
3. the parallelism recognition picking method according to claim 2 applied to fruit picking robot, it is characterised in that: step
In rapid 2, the image that images match identification module acquires slave computer video image acquisition module carries out the matching of template, by searching
Match between rope window and template, then the size relation between respective pixel will be consistent with template；On the contrary, not meeting once
This condition, then it is assumed that do not matched that with its template；When image data matching, data are stored in data storage, work as figure
When unmatched as data, data are deleted, and images match process is re-executed；
Image data result after images match is divided into three classes: image data A, image data B, image data C；Wherein image
Data A is interference image, and image data B is navigation picture, and image data C is fruit figure；
By the method for quick One Dimensional Projection template matching, two-dimensional image is carried out One Dimensional Projection, benefit to X-axis and Y-axis respectively
Its one-dimensional projection value is transformed by one group 0,1 number with one-dimensional differential quantization method and is constituted to describe characteristics of image
Character string, then matched with template with characteristics of image character string；
When template character string and characteristics of image string matching, image is retained, is at this time image data B or picture number
According to C, logic judgment parameter i=1 is taken；When template character string and characteristics of image character string mismatch, image is retained, at this time
For image data A, logic judgment parameter i=0 is taken；
Fine match is carried out to image data B or image data C by feature vector again, image is further classified；Here it uses
To feature vector b and feature vector c, wherein feature vector b is feature vector used in matching navigation picture, and feature vector c is matching
Feature vector used in navigation picture；Each image is traversed with feature vector b and feature vector c respectively；
Feature vector b traverses image, as navigation characteristic vector b and images match, takes logic judgment parameter j=1；
Feature vector c traverses image, as fruit character vector c and images match, takes logic judgment parameter n=1；
By level-one logic judgment, when judging parameter i=1, regards as matching correlation and be classified as image data B or image
Data C when judging parameter i=0, is regarded as matching and uncorrelated is classified as image data A；Two-level logic judgement is carried out again,
When judging parameter j=1, regards as matching correlation and be classified as image data B, when judging parameter n=1, regard as
Image data C is classified as with correlation.
4. the parallelism recognition picking method according to claim 3 applied to fruit picking robot, which is characterized in that step
Rapid 3 specific implementation includes following sub-step:
Step 3.1: matched image being judged before entering segmentation, it is determined whether be required process object；When patrolling
It is required process object when the relationship of collecting is 1, is not required process object when logical relation is 0；
Step 3.2: to the figure of required processing, by the way that the grayscale information of image will be handled, processing is optimized to its gradation data,
The gray value of pixel (i, j) is Gij,0≤Gij≤255；
Step 3.3: n data point X of input1、X2、…、Xn, number k to be clustered；Obtain K cluster centre, to cluster centre into
Row screening, obtains three optimal cluster value Ck1、Ck2、Ck3, choose (Ck1, Ck3) it is the tonal range of optimization to image progress threshold
Value segmentation；
Step 3.4: image being split using threshold value, is obtained by the comparison of threshold value to judge whether this threshold value is best point
Threshold value is cut, just threshold value is σ here, by itself and cluster centre Ck2It is compared, obtains a judgment value w, sentenced by logical relation
It is disconnected, if logical relation is 1, it is split by σ, if logical relation is 0, is split by optimizing threshold value；
Step 3.5: image data after segmentation being carried out to be stored in data storage, includes at least three kinds by the image after segmentation
Image information is classified as data information 1: fruit coordinate, data information 2: fruit number, data information 3: fruit size；By this
Three classes data information stores respectively, remaines in database, so as to subsequent robot's data call.
5. the parallelism recognition picking method according to claim 4 applied to fruit picking robot, it is characterised in that: step
Processing is optimized to its gradation data described in rapid 3.2, it is assumed that image has n data point { x1, x2, x3......xn, it is divided into
3 clustering cluster cluster, what cluster to be done is exactly to minimize the image objective function J to be optimized；
When carrying out classification processing to data with objective function J, when data n is classified as a kind of clusterkWhen, more data relationship variables
Tnk1 is taken, is otherwise 0；
Meet formula when J minimum:
μkValue be all clusterkIn data point average value；Since iteration is all to get the minimum value of J each time, because
This J can be only steadily decreasing or constant, without will increase, obtain 3 cluster point μ of gray level image by iteration repeatedlyk1,
μk2, μk3, μk1<μk2<μk3。
6. the parallelism recognition picking method according to claim 5 applied to fruit picking robot, it is characterised in that: step
In rapid 3.4, if the pixel cluster center that the gray level [0,255] of piece image is obtained by top, taking image gray levels is [μk1,
μk3]；
It is C the pixel threshold value w in selected figure points0And C1Two classes, C0By gray value in [μk1, w] pixel composition, C1There is ash
Angle value is [w, μk3] pixel composition, region C0And C1Average gray can seek μ0、μ1；P0For C0The probability in region, P1For C1Area
The probability in domain；
μ is image in [μk1μk3] average gray:
σ2=p0(μ0-μ)2+p1(μ1-μ)2=p0p1(μ0-μ1)2
Allow K in [μk1μk3] successively value in section, so that σ2Maximum K value is best region segmentation threshold.
7. the parallelism recognition picking method according to claim 6 applied to fruit picking robot, it is characterised in that: step
In rapid 3.4, K value will be acquired and cluster obtained cluster value μk2It is compared, if meeting formula:
Meet segmentation range threshold requirement；
If w≤σ, assignment 1 takes μk2For optimal segmenting threshold,
Otherwise assignment 0, taking K is optimal segmenting threshold.
8. the parallelism recognition picking method according to claim 2 applied to fruit picking robot, it is characterised in that: step
In rapid 4, after being sorted out to image data after segmentation and learning processing, by decision tree, piecemeal, task are carried out to data
Allocation schedule determines the following path of motion decision of picking robot and planning.
9. it is applied to the parallelism recognition picking method of fruit picking robot according to claim 2-8 any one,
Be characterized in that, the specific implementation of step 7 the following steps are included:
Step 7.1: the video image acquisition module of slave computer to the process of host computer, is controlled image transmitting by control circuit S0
System transmits a low level signal 0 when Image Acquisition finishes, and S0 is connected at this time, and lamp L0 is shinny, starts to transmit pictorial information, when
When image transmitting is completed, a high level signal 1 is received, S0 is disconnected at this time；
Step 7.2:S1 is connected, and lamp L1 shines, and host computer transmission control information waits control instruction to execute completion to running gear
When, a high level signal 1 is received, S1 is disconnected at this time；
Step 7.3:S2 is connected, and lamp L2 shines, and host computer transmission control information is positioned to mechanical arm, and control instruction is waited to execute
Cheng Shi receives a high level signal 1, and S2 is disconnected at this time；
Step 7.4:S3 is connected, and lamp L3 shines, and host computer transmission control information waits control instruction to execute to end effector
Cheng Shi receives a high level signal 1, and S3 is disconnected at this time；
Step 7.5:S0, S1, S2, S3 are connected, and lamp L0, L1, L2, L3 shine, and are waited control instruction to execute completion, are respectively received
One high level signal 1, S0, S1, S2, S3 are disconnected, and lamp L0, L1, L2, L3 extinguish；
Step 7.6:S0, S1, S3 are connected, and lamp L0, L1, L3 shine, and are waited control instruction to execute completion, are respectively received a height
Level signal 1, S0, S1, S3 are disconnected, and lamp L0, L1, L3 extinguish；
Step 7.7:S0, S2, S3 are connected, and lamp L0, L2, L3 shine, and are waited control instruction to execute completion, are respectively received a height
Level signal 1, S0, S2, S3 are disconnected, and lamp L0, L2, L3 extinguish；
Step 7.8:S0, S3 is connected, and lamp L0, L3 shine, and control instruction is waited to execute completion, is respectively received a high level letter
Number 1, S0, S3 are disconnected, and lamp L0, L3 extinguish；
Wherein, S0 is the control switch of image transmission circuit, and L0 is that image transmission circuit connects display lamp；S1 is running gear electricity
The control switch on road, L1 are that running gear circuit connects display lamp；S2 is the control switch of mechanical arm spatial location circuitry, and L2 is
Mechanical arm spatial location circuitry connects display lamp；S3 is the control switch of end effector apparatus circuit, and L3 is end effector
Device circuit connects display lamp.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811562710.8A CN109588114B (en) | 2018-12-20 | 2018-12-20 | Parallel recognition picking system and method applied to fruit picking robot |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811562710.8A CN109588114B (en) | 2018-12-20 | 2018-12-20 | Parallel recognition picking system and method applied to fruit picking robot |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109588114A true CN109588114A (en) | 2019-04-09 |
| CN109588114B CN109588114B (en) | 2021-07-06 |
Family
ID=65964148
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811562710.8A Expired - Fee Related CN109588114B (en) | 2018-12-20 | 2018-12-20 | Parallel recognition picking system and method applied to fruit picking robot |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109588114B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110125036A (en) * | 2019-04-25 | 2019-08-16 | 广东工业大学 | A kind of self-identifying sorting system and method based on template matching |
| CN112136505A (en) * | 2020-09-07 | 2020-12-29 | 华南农业大学 | Fruit picking sequence planning method based on visual attention selection mechanism |
| CN112528826A (en) * | 2020-12-04 | 2021-03-19 | 江苏省农业科学院 | Picking device control method based on 3D visual perception |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101726251A (en) * | 2009-11-13 | 2010-06-09 | 江苏大学 | Automatic fruit identification method of apple picking robot on basis of support vector machine |
| CN102914967A (en) * | 2012-09-21 | 2013-02-06 | 浙江工业大学 | Autonomous navigation and man-machine coordination picking operating system of picking robot |
| CN103503637A (en) * | 2013-10-14 | 2014-01-15 | 青岛农业大学 | Intelligent identification picking robot and picking method |
| CN103529855A (en) * | 2013-10-11 | 2014-01-22 | 华南农业大学 | Rotary adjustable binocular vision target recognition and positioning device and application thereof in agricultural fruit harvesting machinery |
| CN103999635A (en) * | 2014-05-21 | 2014-08-27 | 浙江工业大学 | Intelligent automatic cutting type tea-leaf picker based on machine vision and working method |
| WO2016132264A1 (en) * | 2015-02-22 | 2016-08-25 | Ffmh-Tech Ltd. | Multi-robot crop harvesting machine |
| CN106371446A (en) * | 2016-12-03 | 2017-02-01 | 河池学院 | Navigation and positioning system of indoor robot |
| CN108340374A (en) * | 2018-02-08 | 2018-07-31 | 西北农林科技大学 | A kind of control system and control method of picking mechanical arm |
-
2018
- 2018-12-20 CN CN201811562710.8A patent/CN109588114B/en not_active Expired - Fee Related
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101726251A (en) * | 2009-11-13 | 2010-06-09 | 江苏大学 | Automatic fruit identification method of apple picking robot on basis of support vector machine |
| CN102914967A (en) * | 2012-09-21 | 2013-02-06 | 浙江工业大学 | Autonomous navigation and man-machine coordination picking operating system of picking robot |
| CN103529855A (en) * | 2013-10-11 | 2014-01-22 | 华南农业大学 | Rotary adjustable binocular vision target recognition and positioning device and application thereof in agricultural fruit harvesting machinery |
| CN103503637A (en) * | 2013-10-14 | 2014-01-15 | 青岛农业大学 | Intelligent identification picking robot and picking method |
| CN103999635A (en) * | 2014-05-21 | 2014-08-27 | 浙江工业大学 | Intelligent automatic cutting type tea-leaf picker based on machine vision and working method |
| WO2016132264A1 (en) * | 2015-02-22 | 2016-08-25 | Ffmh-Tech Ltd. | Multi-robot crop harvesting machine |
| CN106371446A (en) * | 2016-12-03 | 2017-02-01 | 河池学院 | Navigation and positioning system of indoor robot |
| CN108340374A (en) * | 2018-02-08 | 2018-07-31 | 西北农林科技大学 | A kind of control system and control method of picking mechanical arm |
Non-Patent Citations (2)
| Title |
|---|
| 伍锡如等: "新型苹果采摘机器人的设计与试验", 《科学技术与工程》 * |
| 邹谜等: "一种用于机器人水果采摘的快速识别方法", 《农机化研究》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110125036A (en) * | 2019-04-25 | 2019-08-16 | 广东工业大学 | A kind of self-identifying sorting system and method based on template matching |
| CN112136505A (en) * | 2020-09-07 | 2020-12-29 | 华南农业大学 | Fruit picking sequence planning method based on visual attention selection mechanism |
| CN112528826A (en) * | 2020-12-04 | 2021-03-19 | 江苏省农业科学院 | Picking device control method based on 3D visual perception |
| CN112528826B (en) * | 2020-12-04 | 2024-02-02 | 江苏省农业科学院 | Control method of picking device based on 3D visual perception |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109588114B (en) | 2021-07-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106250812B (en) | A kind of model recognizing method based on quick R-CNN deep neural network | |
| CN109588114A (en) | A kind of parallelism recognition picker system and method applied to fruit picking robot | |
| Puigcerver et al. | Scalable transfer learning with expert models | |
| US7203360B2 (en) | Learnable object segmentation | |
| CN110163236A (en) | The training method and device of model, storage medium, electronic device | |
| CN110021033A (en) | A kind of method for tracking target based on the twin network of pyramid | |
| CN111783831A (en) | Complex image accurate classification method based on multi-source multi-label shared subspace learning | |
| CN106778501A (en) | Video human face ONLINE RECOGNITION method based on compression tracking with IHDR incremental learnings | |
| CN108537168A (en) | Human facial expression recognition method based on transfer learning technology | |
| CN113989944B (en) | Operation action recognition method, device and storage medium | |
| CN106204651A (en) | A kind of method for tracking target based on the judgement improved with generation conjunctive model | |
| D O’Riordan et al. | Object recognition within smart manufacturing | |
| CN108764447A (en) | A kind of group robot Majiang game intelligence dynamicization system and mahjong identification learning algorithm | |
| CN115147488B (en) | Workpiece pose estimation method and grabbing system based on dense prediction | |
| CN114179104B (en) | Picking robot control method and system based on visual identification | |
| CN117520933B (en) | Environment monitoring method and system based on machine learning | |
| CN108921186A (en) | A kind of fingerprint image categorizing system and method based on twin support vector machines | |
| CN112990154B (en) | Data processing method, computer equipment and readable storage medium | |
| CN108280516B (en) | Optimization method for mutual-pulsation intelligent evolution among multiple groups of convolutional neural networks | |
| Naseer et al. | Multimodal Objects Categorization by Fusing GMM and Multi-layer Perceptron | |
| Ahmed et al. | Robust Object Recognition with Genetic Algorithm and Composite Saliency Map | |
| Hoang et al. | Grasp configuration synthesis from 3D point clouds with attention mechanism | |
| CN107609586A (en) | A kind of visual characteristic learning method based on self-supervision | |
| CN113139945A (en) | Intelligent image detection method, equipment and medium for air conditioner outdoor unit based on Attention + YOLOv3 | |
| CN116778220A (en) | Three-dimensional point cloud classification and segmentation method and network based on enhanced point cloud rotation invariance |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20210706 Termination date: 20211220 |