CN104636722B - A kind of overlapping fruit quick Tracking Recognition method of picking robot - Google Patents

A kind of overlapping fruit quick Tracking Recognition method of picking robot Download PDF

Info

Publication number
CN104636722B
CN104636722B CN201510038828.0A CN201510038828A CN104636722B CN 104636722 B CN104636722 B CN 104636722B CN 201510038828 A CN201510038828 A CN 201510038828A CN 104636722 B CN104636722 B CN 104636722B
Authority
CN
China
Prior art keywords
circle
overlapping
fruit
image
center
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.)
Active
Application number
CN201510038828.0A
Other languages
Chinese (zh)
Other versions
CN104636722A (en
Inventor
赵德安
沈甜
陈玉
贾伟宽
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jiangsu University
Original Assignee
Jiangsu University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Jiangsu University filed Critical Jiangsu University
Priority to CN201510038828.0A priority Critical patent/CN104636722B/en
Publication of CN104636722A publication Critical patent/CN104636722A/en
Application granted granted Critical
Publication of CN104636722B publication Critical patent/CN104636722B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G06T7/66—Analysis of geometric attributes of image moments or centre of gravity
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/12—Edge-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/40—Scenes; Scene-specific elements in video content
    • G06V20/49—Segmenting video sequences, i.e. computational techniques such as parsing or cutting the sequence, low-level clustering or determining units such as shots or scenes
    • 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 kind of quick Tracking Recognition method of the overlapping fruit of picking robot, passes through the overlapping Apple image of newest 10 frame of camera continuous acquisition；The first two field picture collected is split, removes background；The position in the overlapping apple center of circle is determined by calculating the maximum for selecting contour edge minimum range in circle, the center of circle is calculated and determines radius to the distance of contour edge；According to the center of circle and the template of radius interception subsequent match；Determine the center of circle of overlapping apple in newest 10 two field picture of continuous acquisition, the motion path of robot is fitted according to the center of circle of every two field picture, prejudged, radius of integration determines the position of overlapping apple in next two field picture with anticipation path, and intercepts overlapping apple region.Finally, match cognization is carried out using quick normalized crosscorrelation matching algorithm.Tracking Recognition to the overlapping fruit of the near-sphericals such as overlapping apple can be realized by this method, and run time is short, can effectively improve the picking efficiency of picking robot.

Description

A kind of overlapping fruit quick Tracking Recognition method of picking robot
Technical field
The invention belongs to agricultural mechanical field, it is related to the image-recognizing method of fruit and vegetable picking robot, particularly to apple The quick Tracking Recognition method of fruit overlapping etc. near-spherical.
Background technology
From nineteen eighty-three the 1st tomato picking robot since the U.S. is born, the research and development of picking robot after More than 30 years, all kinds of vegetables and fruits harvesting intelligent robots of the successive registration study in various countries.However, because discrimination and harvesting rate is not high asks Topic, picking robot also have a certain distance from practical and commercialization, therefore, improve the picking efficiency of picking robot, increase The practical performance of strong picking robot is the key of current research.
The identification and positioning of fruit are the top priority and design difficulty of fruit picking robot, and it is accurate to identify and position Operating efficiency of the sexual intercourse to picking robot.Domestic and foreign scholars have carried out substantial amounts of research for overlapping fruit, and achieve Some preliminary achievements.Item is flourish to wait (2012) to propose a kind of overlapping tomato of method identification based on edge, for slight The recognition correct rate for the overlapping tomato blocked is 90.9%.Song Huaibo etc. (2013) is adopted on the basis of K-means cluster segmentations To split overlapping apple with the method based on convex hull, the experiment proved that, the apple target registration that this method obtains is 85.08%, Average localization error is 14.15%.Xu Yongwei etc. (2013) are using the overlapping grass of the HOG operator identifications of SVMs The certain kind of berries, it identifies that successful accuracy rate is 87%.However, these researchs, both under static conditions, static state knows method for distinguishing The dynamic that picking robot can not be completely suitable in motion process is plucked.Identifications of the Lv Jidong etc. (2014) to dynamic fruit Primary Study is carried out, experiment proves that the correlation of image before and after utilizing can effectively reduce the time of image procossing, but right It is less in the dynamic Tracking Recognition research of overlapping fruit.
The content of the invention
The purpose of the present invention is：A kind of quick Tracking Recognition method for overlapping fruits of near-spherical such as apples, solution are provided Certainly due to fruit grow naturally cause it is overlapping and influence robotic tracking identification the problem of.Its method is simple, and versatility is good, can It is accurately positioned out overlapping seed ball and improves the speed of picking robot.
The technical scheme of the overlapping quick Tracking Recognition method of fruit of picking robot of the present invention comprises the following steps：
A kind of quick Tracking Recognition method of overlapping fruit of picking robot, comprises the following steps：
Step 1, overlapping Apple image collection：Using colored CCD camera continuous acquisition image；
Step 2, objective fruit is split：The image collected is split, removes background, and use mathematical morphology side Method carries out perfect, removal noise and hole to the image after segmentation；
Step 3, the center of circle and the radius of overlapping apple are determined：The very big of contour edge minimum range is arrived by finding point in circle Value finds the position in the center of circle；After the center of circle determines, radius is determined according to the distance in the center of circle to contour edge；
Step 4, objective fruit template is extracted：According to the center of circle tried to achieve with radius along with certain reserved value interception is follow-up The template of matching；
Step 5, robot motion path prejudges：To fruit in the newest 10 width image of continuous acquisition in robot motion The motion path in the center of circle is fitted and prejudged, and overlapping fruit region is intercepted according to the radius of the center of circle of anticipation and fruit；
Step 6, match cognization：Match cognization is carried out to overlapping fruit using the matching of quick normalized crosscorrelation.
Further, used in the step 2 Mathematical Morphology Method the image after segmentation is carried out perfect process for：
Step 2.1, dilation operation is carried out to image for the disc-shaped structure element of a pixel with radius, expands border Point, fills up some perforations；
Step 2.2, holes filling is carried out with floodfill algorithms, fills up the aperture of calyx part, try to achieve image afterwards Largest connected region, isolated burr is removed；
Step 2.3, erosion operation is carried out to image, eliminates the noise of boundary member.
Further, the step 3 is specially：
Step 3.1, the step of determining the center of circle：Define four scanning direction A (x+,y+)、B(x-,y+)、C(x-,y-)、D(x+, y-), in A directions, so that from left to right, mode from top to bottom is scanned；In B directions, to be turned left from the right side, side from top to bottom Formula is scanned；In C directions, to be turned left from the right side, mode from top to bottom is scanned；In D directions, with from left to right, under It is scanned to upper mode；
Step 3.2, the step of determining radius：The above-mentioned center of circle A (a for obtaining overlapping applex,ay)、B(bx,by)；Obtain again By the center of circle A, B linear equation y；Obtain the intersection point C (c of the straight line and fruit profilex,cy)、D(dx,dy)；Radius
Further, the concrete processing procedure of the step 5 is as follows：
Step 5.1, overlapping two centers of circle of fruit in the image of the preceding 9 width continuous acquisition determined according to the step 3.1 Position, fitting of a polynomial is carried out to the midpoint in two centers of circle, fits the path of robot motion, in conjunction with robot motion's speed Degree and sampling time are prejudged, and estimate the position that next frame Circle in Digital Images is put in the heart；
Step 5.2, radius is determined according to the step 3.2, obtains the maximum r of two fruit radiusesmax, with step Centered on the midpoint in two centers of circle of 5.1 anticipations, the interception length of side is 4*rmaxThe region that is handled as successive image of square.
Further, the algorithm that normalized crosscorrelation matches in the step 6 is reduced to：
In formula, I is image to be matched (pixel is M × N)；T is that (x, y) is template image (pixel is m × n)；(x, y) is Subgraph Ix,yCoordinate of the upper left corner in image I；(u, γ) is the coordinate of pixel in a template；For subgraph Ix,yPixel Average value；
The present invention technique effect be：For the near-spherical such as apple fruit overlapping feelings of fruit caused by growing naturally Condition, the inventive method can be accurately positioned out overlapping fruit under robot motion's state, and according to front and rear image information pair Robot motion path is prejudged, and reduces the workload of successive image processing, therefore run time is short, and the real-time of harvesting obtains To effectively improving.
Brief description of the drawings
Fig. 1 is the quick Tracking Recognition flow chart of overlapping apple；
Fig. 2 is the image after the segmentation of overlapping Apple image and morphology operations, wherein, Fig. 2 a are original image, Fig. 2 b are to segment the image dealt with problems arising from an accident；
Fig. 3 is the identification positioning figure of overlapping apple original image shown in Fig. 2, wherein, Fig. 3 a are scanning when determining the center of circle Direction schematic diagram, Fig. 3 b are minimum range three-dimensional function figure of the point in circle to contour edge, and Fig. 3 c are the method for determining radius Schematic diagram, Fig. 3 d are the recognition result figure of overlapping fruit image；
Fig. 4 is the matching template figure of extraction；
Fig. 5 is robot motion's path curve fitted figure；
Fig. 6 is overlapping apple extracted region result figure, and wherein Fig. 6 a are the schematic diagram of overlapping apple extracted region, and Fig. 6 b are Result figure；
Fig. 7 is quick normalized crosscorrelation matching result figure.
Embodiment
The embodiment of the present invention is further described below in conjunction with the accompanying drawings, idiographic flow of the invention is as schemed Shown in 1.
1st, overlapping Apple image collection
The present invention uses colored CCD camera continuous acquisition image, and frequency acquisition is 10 frames/second, robot kinematics Middle continuous acquisition image, utilize 10 newest overlapping Apple images.
2nd, objective fruit is split
The present embodiment is split using the OTSU split plot designs based on color characteristic to the image collected, that is, uses R component under RGB color model subtracts G components.There is hole, burr, noise etc. in the image after over-segmentation, therefore use number Morphological method carries out perfect to the image after segmentation.Specific method is first with the disc-shaped structure that radius is 1 pixel Element carries out dilation operation to image, expands boundary point, fills up some perforations；Then hole is carried out with Floodfill algorithms to fill out Fill, fill up the aperture of calyx part, try to achieve the largest connected region of image afterwards, isolated burr is removed；Finally, then to figure As carrying out erosion operation, the noise of boundary member is eliminated.It is as shown in Figure 2 that effect is improved in image segmentation.
3rd, the center of circle and the radius of overlapping apple are determined
3.1 determine the center of circle
The position in the center of circle just can be found to the maximum of contour edge minimum range by point in searching circle.But if The distance of all point to contour edge inherently takes substantial amounts of internal memory in calculating circle and real-time is poor.The present embodiment uses Improved method is scanned to the point in circle, defines four scanning direction A (x+,y+)、B(x-,y+)、C(x-,y-)、D(x+, y-).In A directions, so that from left to right, mode from top to bottom is scanned；In B directions, to be turned left from the right side, side from top to bottom Formula is scanned；In C directions, to be turned left from the right side, mode from top to bottom is scanned；In D directions, with from left to right, under It is scanned to upper mode, schematic diagram is as shown in Figure 3 a.
The point of point and its four neighborhood in profile relatively takes minimum value afterwards, can obtain minimum range function, it is three-dimensional Curved surface design sketch is as shown in Figure 3 b.What wherein red circle marked is maximum at the two of minimum range, that is, two weights The home position of folded apple.
3.2 determine radius
After the center of circle determines, radius can be determined according to the center of circle, but can not be merely by the center of circle to contour edge distance Maximum determines radius, because in the case of overlapping fruit, is probably the distance to another apple profile apart from maximum.For Avoid such case, take in the following manner herein：According to the 3.1 center of circle A (a for obtaining overlapping applex,ay)、B(bx,by)；Again Obtain the linear equation y by the center of circle A, B；Obtain the intersection point C (c of the straight line and fruit profilex,cy)、D(dx,dy)；RadiusDetermine the schematic diagram of radius as shown in Figure 3 c.It is overlapping The design sketch of fruit positioning is as shown in Figure 3 d.
4th, objective fruit template is extracted
Determine the center of circle with radius along with the template of certain reserved value interception subsequent match according to 3.1,3.2.Its effect Figure is as shown in Figure 4.
5th, robot motion path anticipation step：According to fruit in the newest 10 width image of continuous acquisition in robot motion The center of circle its motion path is fitted and prejudged.Specific processing step is as follows：
Overlapping two centers of circle of fruit in the image for the preceding 9 width continuous acquisition that Step1 is determined before by 3.1 method Position, fitting of a polynomial is carried out to the midpoint in two centers of circle, fitting precision is less than or equal to 0.5, as shown in figure 5, fitting machine The path of device people motion, is prejudged in conjunction with robot movement velocity and sampling time, estimates next frame Circle in Digital Images The position put in the heart.
Step2 determines radius by 3.2 method, obtains the maximum r of two fruit radiusesmax, with Step1 anticipations Centered on the midpoint in two centers of circle, the interception length of side is 4*rmaxThe region that is handled as successive image of square.Its schematic diagram is such as Shown in Fig. 6 a.
Overlapping fruit extracted region design sketch after treatment is as shown in Figure 6 b.
6th, match cognization
Match cognization is carried out to overlapping fruit using the matching of quick normalized crosscorrelation.Fig. 7 is quick normalized crosscorrelation Matching result figure.
The algorithm steps of normalized crosscorrelation matching are as follows：
If image I (pixel is M × N) to be matched and template image T (pixel is m × n), the definition of normalizated correlation coefficient For：
In formula：(x, y) is subgraph Ix,yCoordinate of the upper left corner in image I；(u, γ) is the seat of pixel in a template Mark；
For subgraph Ix,yPixel average；
For template T pixel average.For R (x, y) scope between (0,1), coefficient is bigger, illustrates two matching templates Between similitude it is higher.
Normalized crosscorrelation matching primitives amount is excessive, and real-time is poor, therefore, is matched and calculated using quick normalized crosscorrelation Method.Comprise the following steps that：
Step 1 is setThen by simplifying, the molecular moiety of formula (1) can be rewritten as：
In formulaThen molecule is further simplified as：
According to the property of Fourier transformation, molecule can be rewritten as
R(x,y)numerator=F-1{F{I}·F*{T'}} (6)
Step 2 for denominator part, due to template be it is known, thereforeIt is known definite value, Optimal solution is found when not interfering with normalization matching, can not be calculated, so the denominator of formula (1) can be reduced to：
In summary, normalizated correlation coefficient can be reduced to：
It should be understood that above-mentioned example of applying is only illustrative of the invention and is not intended to limit the scope of the invention, the present invention is being read Afterwards, modification of the those skilled in the art to the various equivalent form of values of the present invention falls within the application appended claims and limited Scope.

Claims (5)

1. the quick Tracking Recognition method of the overlapping fruit of a kind of picking robot, it is characterised in that comprise the following steps：
Step 1, overlapping Apple image collection：Using colored CCD camera continuous acquisition image；
Step 2, objective fruit is split：The image collected is split, removes background, and use Mathematical Morphology Method pair Image after segmentation carries out perfect, removal noise and hole；
Step 3, the center of circle and the radius of overlapping apple are determined：Looked for by finding point in circle to the maximum of contour edge minimum range To the position in the center of circle；After the center of circle determines, radius is determined according to the distance in the center of circle to contour edge；
Step 4, objective fruit template is extracted：According to the center of circle tried to achieve with radius along with certain reserved value intercepts subsequent match Template；
Step 5, robot motion path prejudges：To the center of circle of fruit in the newest 10 width image of continuous acquisition in robot motion Motion path be fitted and prejudge, overlapping fruit region is intercepted according to the radius of the center of circle of anticipation and fruit；
Step 6, match cognization：Match cognization is carried out to overlapping fruit using the matching of quick normalized crosscorrelation.
2. the quick Tracking Recognition method of the overlapping fruit of picking robot according to claim 1, it is characterised in that described Used in step 2 Mathematical Morphology Method the image after segmentation is carried out perfect process for：
Step 2.1, dilation operation is carried out to image for the disc-shaped structure element of a pixel with radius, expands boundary point, fill out Mend some perforations；
Step 2.2, holes filling is carried out with floodfill algorithms, fills up the aperture of calyx part, try to achieve the maximum of image afterwards Connected region, isolated burr is removed；
Step 2.3, erosion operation is carried out to image, eliminates the noise of boundary member.
3. the quick Tracking Recognition method of the overlapping fruit of picking robot according to claim 1, it is characterised in that described Step 3 concretely comprises the following steps：
Step 3.1, the step of determining the center of circle：Define four scanning direction A (x+,y+)、B(x-,y+)、C(x-,y-)、D(x+,y-), In A directions, so that from left to right, mode from top to bottom is scanned；In B directions, to be turned left from the right side, mode from top to bottom is entered Row scanning；In C directions, to be turned left from the right side, mode from top to bottom is scanned；In D directions, with from left to right, from top to bottom Mode be scanned；
Step 3.2, the step of determining radius：The above-mentioned center of circle A (a for obtaining overlapping applex,ay)、B(bx,by)；Obtain again by circle The heart A, B linear equation y；Obtain the intersection point C (c of the straight line and fruit profilex,cy)、D(dx,dy)；Radius
4. the quick Tracking Recognition method of the overlapping fruit of picking robot according to claim 3, it is characterised in that described The concrete processing procedure of step 5 is as follows：
Step 5.1, the position in overlapping two centers of circle of fruit in the image of the preceding 9 width continuous acquisition determined according to the step 3.1, Fitting of a polynomial is carried out to the midpoint in two centers of circle, fits the path of robot motion, in conjunction with robot movement velocity with And the sampling time is prejudged, the position that next frame Circle in Digital Images is put in the heart is estimated；
Step 5.2, radius is determined according to the step 3.2, obtains the maximum r of two fruit radiusesmax, it is pre- with step 5.1 Centered on the midpoint in two centers of circle sentenced, the interception length of side is 4*rmaxThe region that is handled as successive image of square.
5. the quick Tracking Recognition method of the overlapping fruit of picking robot according to claim 1, it is characterised in that described The algorithm that normalized crosscorrelation matches in step 6 is reduced to：
In formula, I is image to be matched, and pixel is m × n；T is template image, and pixel is m × n；(x, y) is subgraph Ix,yUpper left Coordinate of the angle in image I；(u, γ) is the coordinate of pixel in a template；For subgraph Ix,yPixel average.
CN201510038828.0A 2015-01-26 2015-01-26 A kind of overlapping fruit quick Tracking Recognition method of picking robot Active CN104636722B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510038828.0A CN104636722B (en) 2015-01-26 2015-01-26 A kind of overlapping fruit quick Tracking Recognition method of picking robot

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510038828.0A CN104636722B (en) 2015-01-26 2015-01-26 A kind of overlapping fruit quick Tracking Recognition method of picking robot

Publications (2)

Publication Number Publication Date
CN104636722A CN104636722A (en) 2015-05-20
CN104636722B true CN104636722B (en) 2018-02-27

Family

ID=53215452

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510038828.0A Active CN104636722B (en) 2015-01-26 2015-01-26 A kind of overlapping fruit quick Tracking Recognition method of picking robot

Country Status (1)

Country Link
CN (1) CN104636722B (en)

Families Citing this family (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106228138A (en) * 2016-07-26 2016-12-14 国网重庆市电力公司电力科学研究院 A kind of Road Detection algorithm of integration region and marginal information
CN106875412B (en) * 2017-02-28 2020-06-23 重庆理工大学 Segmentation positioning method for two overlapped fruits
CN107038446B (en) * 2017-03-23 2020-06-05 中国计量大学 Night double-fruit overlapping tomato identification method based on overlapping edge detection under active illumination
CN108256522B (en) * 2018-01-09 2020-05-19 北京航空航天大学 Two-dimensional identification method for SPR (surface plasmon resonance) and waveguide mixed absorption spectrum characteristic position
CN108470339A (en) * 2018-03-21 2018-08-31 华南理工大学 A kind of visual identity of overlapping apple and localization method based on information fusion
CN109915924B (en) * 2018-05-31 2020-11-24 徐州云创物业服务有限公司 Safety protection type electric heater
CN109255795B (en) * 2018-09-11 2021-04-06 中国计量大学 Tomato plant edge sorting method
CN109544572B (en) * 2018-11-19 2023-07-25 常州大学 Method for acquiring near-large fruit target in orchard image
CN109559299B (en) * 2018-11-19 2023-07-25 常州大学 Method for separating double-fruit overlapped fruits
CN111753577B (en) * 2019-03-28 2024-01-30 天津工业大学 Apple identification and positioning method in automatic picking robot
JP2020195335A (en) * 2019-06-04 2020-12-10 本田技研工業株式会社 Fruit vegetable harvesting device and fruit vegetable harvesting method
CN110766698B (en) * 2019-10-16 2022-04-22 浙江工业大学 Method for tracking and identifying oscillating apples under dynamic background
CN111738146A (en) * 2020-06-22 2020-10-02 哈尔滨理工大学 Rapid separation and identification method for overlapped fruits
CN112369208B (en) * 2020-12-01 2021-07-13 大连理工大学 Method for dynamically planning picking sequence of spheroidal fruits
CN112784735B (en) * 2021-01-21 2021-10-01 重庆文理学院 Method for identifying string-type fruit mother branches based on monocular camera and binocular camera

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101493892A (en) * 2009-02-27 2009-07-29 中国农业大学 Image characteristic extracting method and device
CN104200193A (en) * 2014-08-05 2014-12-10 北京农业信息技术研究中心 Fruit tree yield estimation method and device

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080002878A1 (en) * 2006-06-28 2008-01-03 Somasundaram Meiyappan Method For Fast Stereo Matching Of Images

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101493892A (en) * 2009-02-27 2009-07-29 中国农业大学 Image characteristic extracting method and device
CN104200193A (en) * 2014-08-05 2014-12-10 北京农业信息技术研究中心 Fruit tree yield estimation method and device

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
苹果采摘机器人果实识别与定位方法;司永胜等;《农业机械学报》;20100930;第41卷(第9期);第148-153页 *
苹果采摘机器人目标果实快速跟踪识别方法;吕继东等;《农业机械学报》;20140131;第45卷(第1期);第65-72页 *

Also Published As

Publication number Publication date
CN104636722A (en) 2015-05-20

Similar Documents

Publication Publication Date Title
CN104636722B (en) A kind of overlapping fruit quick Tracking Recognition method of picking robot
Fernandez et al. Automatic detection of field-grown cucumbers for robotic harvesting
Khan et al. An efficient contour based fine-grained algorithm for multi category object detection
CN110232389B (en) Stereoscopic vision navigation method based on invariance of green crop feature extraction
CN110866903B (en) Ping-pong ball identification method based on Hough circle transformation technology
CN104361330B (en) A kind of crop row recognition methods of corn accurate dispenser system
CN106919902B (en) Vehicle identification and track tracking method based on CNN
CN103870808A (en) Finger vein identification method
Gui et al. A new method for soybean leaf disease detection based on modified salient regions
CN106570510A (en) Supermarket commodity identification method
CN107169972A (en) A kind of quick contour tracing method of noncooperative target
CN107316043A (en) A kind of stacking mushroom method for quickly identifying of picking robot
CN109684941A (en) One kind picking region partitioning method based on MATLAB image procossing litchi fruits
CN111445482A (en) Segmentation and identification method of overlapped agaricus bisporus
CN111476804A (en) Method, device and equipment for efficiently segmenting carrier roller image and storage medium
CN113450402B (en) Navigation center line extraction method for vegetable greenhouse inspection robot
Yang et al. Tracking and recognition algorithm for a robot harvesting oscillating apples.
Liu et al. SE-Mask R-CNN: An improved Mask R-CNN for apple detection and segmentation
Moghimi et al. A robo-vision algorithm for automatic harvesting of green bell pepper
CN107093182A (en) A kind of human height's method of estimation of feature based flex point
CN113936305B (en) Middle finger position confirmation method and feature extraction method in palm recognition process
Hu et al. Extraction method for centerlines of crop row based on improved lightweight Yolov4
CN110210342A (en) A kind of Humanface image matching method and its system, readable storage medium storing program for executing
Kanwal et al. Overlapped apple fruit yield estimation using pixel classification and hough transform
CN113781389A (en) Multi-view goods masking method based on depth map

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant