CN105118028A - Automatic grafting machine grafted seedling seam identifying apparatus and seam identifying method - Google Patents
Automatic grafting machine grafted seedling seam identifying apparatus and seam identifying method Download PDFInfo
- Publication number
- CN105118028A CN105118028A CN201510459023.3A CN201510459023A CN105118028A CN 105118028 A CN105118028 A CN 105118028A CN 201510459023 A CN201510459023 A CN 201510459023A CN 105118028 A CN105118028 A CN 105118028A
- Authority
- CN
- China
- Prior art keywords
- image
- grafting
- gray
- seam
- scale value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses a method and apparatus suitably used for identifying and detecting vegetable automatic grafting machine grafted seedling seams. The apparatus comprises an industrial camera, an illuminator, an image acquisition card, a computer, a transmission mechanism and a seedling growing pot. The industrial camera is connected with the image acquisition card through a data transmission line. The illuminator is connected with the computer through a wire. The image acquisition card is connected with the computer through a data transmission line. The industrial camera, the illuminator, the image acquisition card and the computer form a machine visual sense system. The identifying and detecting method determined through steps of image acquiring, image processing, inputting, outputting, controlling, executing, etc. achieves grafted seedling sorting. Human eyes are replaced by the machine visual sense system to achieve identification of the vegetable grafted seedling seams. Furthermore, by arranging an execution mechanism with a relatively high mechanization degree, the apparatus which is convenient, rapid and highly objective achieves a high intelligentized level. The apparatus which has strong practicality has reliable performance.
Description
Technical field
The present invention relates to agriculture field, particularly a kind of grafting seam recognition device of full-automatic grafting machine and seam recognition methods.
Background technology
Along with the development of science and technology; Vegetable produce robotization and intelligent level improve constantly; reading intelligent agriculture robot application gets more and more; it is for raising resource utilization and Agricultural Output rate; increase economic efficiency; realizing the scale of industrialized agriculture, mechanization and robotization significant, is the main trend of Future Development.Current grafting machine is the plant equipment that grafting speed is significantly promoted, it adopts different engrafting methods, needing the stock of the plant of grafting, fringe wood is connected as a single entity fast, substantially increase throughput rate and grafting quality, be called as a revolution (Li Zhongqiu, 2007) of graft seedling growth technology.
The eighties, graft technology is just generally applied in by countries in the world among the cultivation of crop.Agricultural chemicals is generally used to ensure the Quality and yield of grafting plant growth both at home and abroad.Adopt the method for grafting very time-consuming because artificial, so some areas just abandon grafting cultivation, but take to spray insecticide in a large number pre-preventing disease and pest.Like this, not only waste resource, but also cause environmental pollution, destroy the living environment of human living.The sixty-four dollar question simultaneously manually carrying out grafting is that grafting efficiency is lower and labour intensity large, the survival rate of artificial grafting nursery stock is lower, and artificial grafting can not adapt to the requirement of China's modern agricultural development.Adopt automatic grafting robot to carry out grafting operation, not only can improve operating speed, can improve grafting operation quality simultaneously, what increase grafting heals into motility rate, is applicable to volume facility graft seedling growth and produces.At present, the production of China vegetables, melon and fruit and the development of installation agriculture technology have possessed the basis and condition of greatly developing automatic grafting robot technology, therefore, development robotization graft technology, be conducive to new and high technology and be converted into yield-power rapidly, promote the great-leap-forward development of China's agricultural modernization.In China, development grafting mechanization and automatic technology imperative.At present, succeeded in developing the grafting robot be made up of parts such as mechanical system, control system, power system and vision systems both at home and abroad, but they still can not meet the technical requirement of people to grafting completely.
Liaoning has become the important industrialized agriculture production base of northern China, and the area of industrialized agriculture increases just year by year, and broken through 1,100 ten thousand mu to the end of the year 2012, heliogreenhouse scale breaks through 8,000,000 mu, and position ranks first in the country.The vegetables total production of the whole province based on facilities vegetable reaches 4,013 ten thousand tons, and wherein facilities vegetable output reaches 2,998 ten thousand tons, accounts for 75% of vegetables total production.Therefore develop full-automatic grafting machine in Liaoning Area and there is important practical significance.
The throughput rate of grafting machine is relevant with its automaticity, and full-automatic grafting machine throughput rate is high, and semi-automatic grafting machine throughput rate is low.The grafting success ratio of grafting machine is automatically a little more than semi-automatic, because the accuracy of mechanical work is better than manual work; But, automaticity is higher, the grafting success ratio of grafting machine is closer with the relation of the standardization level of grafting seedling, and namely only under the physical dimension of the grafting stock provided and scion reaches standard, homogeneous condition, grafting machine just can ensure higher grafting success ratio.Therefore machine vision realizes operation to full-automatic grafting machine and plays a part particularly important.The semi-automatic grafting robot of vegetables succeeded in developing both at home and abroad at present still needs hand inspection seam quality, and seam quality quality directly affects graft survival rate and yield of vegetables.Because eye-observation exists certain subjectivity, in the process judging grafting quality good or not, will error be there is, thus reduce the success ratio of grafting.Research vegetable grafting machine grafting seam visual identifying system, can improve the wrong identification produced because of human eye error effectively.In addition, operating efficiency and graft survival rate can also be improved widely.
Japanese Agriculture, Forestry and Fisheries Ministry Biological department specific industry technical research promotion institution in 1986 is worked in coordination with associated companies and to be taken the lead in the world the grafting machine that begins one's study, through 3 generation experimental prototype exploration, within 1993, start successively to develop polytype grafting robot.Wherein, comprise the R800B/T type automanual system grafting machine of Japanese Jing Guan machinery company's development, productive capacity be 800 strains/hour, be applicable to the melon and solanaceous vegetables of grafting; Yang Ma machinery company's development AG1000 type full rotation type and 600T type automanual system grafting machine, the former productive capacity be 1000 strains/hour, be applicable to Solanaceae, the latter be 600 strains/hour, for melon grafting; The full-automatic grafting machine of entire column type of TGR Research Institute, adopts bonding agent to substitute Grafting clip and fix stock and scion, productive capacity be 1000 strains/hour.
Early 1990s, Korea S also develop automanual system machinery grafting machine, maximum productivity can reach 310 strains/hour, this body amass little, price is low, adopt inarching method.Because price is low, there is very large sales volume in Korea S, Japan and China.
1998, China Agricultural University develops 2JSZ-600 type automatic vegetable grafting robot, this machine uses and amplexiforms the grafting of formula method, what achieve stock and scion send the automated jobs such as seedling, cutting, upper Grafting clip, automaticity is higher, be applicable to melon vegetables grafting, productive capacity reach 600 strains/hour.
At present, the grafting robot that Japan develops, automaticity is high, and price is also higher, and generally between 350,000 yuan ~ 2,000,000 yuans, non-general peasant household and middle-size and small-size nursery center can bear.The grafting robot of Korea S's exploitation due to structure simple, volume is little, only about 3000 yuans, but this grafting machine mainly adopts and is abutted against formula grafting, and grafting job procedure is more numerous and diverse, is only suitable for nursery on a small scale.
The research part commercialization of domestic semi-automatic grafting robot, but grafting robot can not realize the production automation completely, still need manually to carry out quality testing to the seam after grafting, thus have impact on the manufacture and exploit of full-automatic grafting robot.Due to the subjectivity of individual and the difference of skills involved in the labour, grafting quality and grafting efficiency can be affected, thus affect graft survival rate.Therefore, after grafting, need automatic identification and the detection of grafting being carried out to seam quality.The present invention by automatically identifying grafting robot grafting seedling seam, the research of detection technique, for the manufacture and exploit of the full-automatic grafting robot of the vision system that puts together machines successfully is provided fundamental basis and technical support, therefore for the demand of the full-automatic grafting machine of vegetables, develop the full-automatic grafting machine grafting visual identity of a kind of vegetables and pick-up unit and algorithm, realize the efficient grafting of vegetables.
Light source plays vital effect in Vision Builder for Automated Inspection, and it is not only illuminating objects simply, is more information characteristics required in outstanding photographic subjects.Prominent feature target in background environment, needs very strong contrast.Contrast is larger, and target signature is more obvious.The mode strengthening contrast the best adjusts light source exactly.After have employed suitable light source and illumination scheme, greatly reduce graphical analysis difficulty, make algorithm easier, improve operating efficiency simultaneously.Therefore, the selection of light source and the design of illumination scheme play vital effect in the process of image acquisition.
Summary of the invention
The object of the invention is, a kind of grafting seam recognition device and seam recognition methods of full-automatic grafting machine are provided, reasonable in design, cost is lower, usability is reliable, and intelligent level is high, can easily identify and detect the seam of vegetable grafted seedling, establishing criteria judges the quality of grafting, and is completed by system and device and perform an action.
The technical scheme adopted is:
A kind of grafting seam recognition device of full-automatic grafting machine, comprise industrial camera, luminaire, image pick-up card, computing machine, connecting gear and pot for growing seedlings, industrial camera is connected with image pick-up card by data line, luminaire is connected with computing machine by electric wire, image pick-up card is connected with computing machine by data line, industrial camera, luminaire, image pick-up card and computing machine form Vision Builder for Automated Inspection, described connecting gear, comprise driving-belt, first power wheel, second power wheel, motor and frame, the wheel shaft of the first power wheel and the second power wheel is all propping up in frame by bearing assemble, travelling belt twines and is located on the first power wheel and the second power wheel, motor is fixed in frame, and the output shaft of motor is connected by the wheel shaft of reductor with the first power wheel, motor is enable to drive the first power wheel to rotate, and then drive conveyer belt, it is characterized in that: described luminaire, comprise LED and the lamp socket of multiple greens of arrangement formation one ring plain structure, multiple LED is fixed on lamp socket, luminaire is fixed by screws on the support of travelling belt one end, described pot for growing seedlings is placed on a moving belt, the other end of travelling belt is fixed with electronic moisture meter and electronic thermometer, under specific temperature and humidity, and the most applicable collection drawing information of biological aspect of graft, improve photographic quality, and then improve the accuracy rate identified, described luminaire is arranged on the support of travelling belt one end, the annular LED lamp group outward flange place plane orthogonal of luminaire is in surface level, light source chooses green LED light source, plurality of LEDs lamp is combined to form annular plane structure, the light beam that light source sends is under the condition vertical with the jointed graft of band, the picture contrast collected is strong, characteristic information is obvious, convenient process, the side end face of the opposite side of travelling belt is fixed with illuminance sensor by expansion link, one end of expansion link is fixedly connected on the side end face of travelling belt, the other end of expansion link is fixedly connected with illuminance sensor, facilitate the height of adjusting intensity of illumination sensor, make illuminance sensor more close to seaming position, carry out the illuminance of more Accurate Determining seam crossing, under specific intensity of illumination, seam discrimination is the highest, the camera lens of described industrial camera passes the center pit of ring lighting, and is arranged at the rear of luminaire.
Mercury series MER-125-30UC colorful digital CCD camera that described industrial camera can be produced for company of Daheng, interface is USB2.0, and resolution is 1292 (H) × 964 (V); Image sensor chip, dimension scale 1/3inch (4.8 × 3.6mm); Computer series M2514-MP2 selected by camera lens, and focal length parameter is 25mm, and diameter of lens is 46mm.
A grafting seam recognition methods for full-automatic grafting machine, is characterized in that comprising the following steps:
Step one: preliminary work, the grafting seam recognition device of full-automatic grafting machine is placed in a space closed, control travelling belt transmission, pot for growing seedlings is made to stop at positive control funerary objects and industrial camera position, the camera lens 12cm of the grafting distance industrial camera on seedling culture hole plate, the annular LED lamp group of luminaire is placed on camera lens; Regulate lamp brilliance, the illuminance that illuminance sensor is shown is 130 ~ 150Lux; Regulate the temperature and humidity in enclosure space, making electronic moisture meter show humidity is 65 ~ 70%, and electronic thermometer displays temperature is 28 ~ 30 DEG C;
Step 2: image acquisition: maintain above-mentioned state after 10 ~ 15 minutes, industrial camera starts to gather image, then sends image information to image pick-up card, and last image pick-up card sends information to computing machine;
Step 3: image procossing: according to the feature of seam, according to the principle of visual identity, by Matlab software, first the vegetable grafted seedling stem part (comprising Grafting clip) that Grafting clip clamps is found out, and then remove clip according to Iamge Segmentation principle, obtain the image comprising seam portion like this, the more cumulative summation of horizontal direction pixel gray-scale value is carried out to this image, obtain figure characteristic curve.
Wherein, the algorithm of image procossing comprises the following steps:
A. coloured image greyscale transformation: convert the coloured image of acquisition to gray level image, converts single pass gray-scale map to by R, G, B triple channel image.In gray level image, the gray-scale value of each pixel only uses a byte representation, gray scale span is 0 ~ 255, this point of the less expression of numerical value is more black also darker, 0 represents the most black, this point of the larger expression of numerical value is whiter also brighter, 255 represent the brightest namely complete white, and the gray level image after conversion does not contain color information, only containing monochrome information.Coloured image converts the formula of gray level image to such as formula (3-1), Gray (i in formula, j) after representing conversion, gray level image is at (i, j) gray-scale value put, R (i, j), G (i, j), B (i, j) represent the component information of color red, green, blue in coloured image respectively, 0.30,0.59,0.11 is respectively the coefficient that coloured image is weighted 3 variablees when average gray is changed.
Gray(i,j)=0.30R(i,j)+0.59G(i,j)+0.11B(i,j)(3-1)
B. gray level image threshold transformation: use adaptive threshold or dynamic threshold according to feature of image, formula (3-2)
In formula, f (x, y) is the gray-scale value after (x, y) some place pixel transform, and T is the threshold value of setting.
C. image binaryzation: according to the gray-value variation scope of image, choose a certain gray-scale value as threshold value T, image is divided into two parts, gray-scale value G (I, j) by any point (i, j) in image contrasts with T, if be more than or equal to T, the gray-scale value of this point will be set to 1, if be less than T, be just set to 0, shown in (3-3), otherwise Ru shown in (3-4).
Or
In formula: the gray-scale value that G (I, j) is original image, F (I, j) is the gray-scale value of image after binaryzation.
D. picture smooth treatment: the smoothing process of method using mean filter.Mean filter is also called smooth linear wave filter, adopt neighborhood averaging, the mean value of all pixels value of pixel each in neighborhood in image determined in order to the method for filtering mask replaces, namely the gray-scale value of the pending current pixel point (x, y) of assignment is carried out with the gray average of all pixels selected in template.For suppression Gaussian noise, mean filter has good effect, be mainly used in removing the incoherent details of pixel region little compared with filtering mask dimensions in image, formula (3-5) is shown in the calculating using the image of m × n (m and n is odd number) weighted mean to a width M × N to carry out filtering, a=(m-1)/2 in formula and b=(n-1)/2, the mask coefficient that w (s, t) is g (x+s, y+t).
E. morphology closed operation: first carry out expansion process (Dilation) to bianry image and carry out corrosion treatment (Erosion) to the image after process again, its definition is such as formula shown in (3-6).Closed operation process usually can narrow interruption or tiny hole in fused images, and eliminates and the incoherent fine edge of target, makes the outline line of target in bianry image become more smooth.Make up crack in bianry image usually through to image expansion process, to eliminate in bianry image incoherent details usually through to Image erosion process, the definition of dilation and erosion is respectively such as formula shown in (3-7) and (3-8).
Under normal circumstances, A is set to the set of bianry image, and B is set to the structural element for process, and structural element B is also 1 image collection, will operate in morphology operations with structural element to set.Namely expansion process arrives B after the first primitive translation z unit in structural element B
zif include B in A
zelement (not necessarily all comprising), then this point is retained, otherwise does not process this point, and the set that the final z point retained forms is exactly the result that B expands to A.
Step 4: computing machine judges the figure characteristic curve obtained according to the program set, detect qualified grafting and defective grafting, and then computing machine sends instruction to control system, carries out down-stream.
Its advantage is:
Replace the seam of eye recognition vegetable grafted seedling with machine vision, complete vegetable grafted seedling automatic grafting process according to canonical parameter, reasonable in design, dependable performance, practical and convenient, intelligent level is high, can realize grafting job requirements well.
Principle of work:
The image acquisition part in visual identifying system is utilized to carry out image acquisition to vegetable grafted seedling seam, select suitable light illuminating testee, its information characteristics outstanding, recycling camera converts light signal to electric signal, export to image pick-up card, convert thereof into digital image information by image pick-up card again, flow to image processing software.The coloured image collected is carried out gray-scale map conversion by image processing software, gray level image after conversion is not containing color information, only containing monochrome information, the gray level of enhance operation to image slices vegetarian refreshments is used to modify, because exposing some defect of causing such as uneven and unifying to change to the gray level of pixel in entire image during to compensate shooting image.Then adopt the method for binaryzation to carry out threshold transformation to gray level image, target object is extracted from multivalue image.Smoothing processing is used to remove the incoherent details of pixel region little compared with filtering mask dimensions in image afterwards.And then use morphology closed operation to carry out dilation and erosion process, interruption narrow in image and tiny cavity are merged, eliminates and the incoherent fine edge of target, make the outline line of target in bianry image become more smooth.Thus just more clear to the description of grafting seam feature, software program can easily extract characteristic information and be further analyzed process, draw a gray-scale value summation curve, if grafting seam quality is better, then scion seedling contacts closely with the grafting faying face of rootstock seedling, thus seam areas gray-scale value is less, curve obtained is the line that middle part falls in the left, if this grafting seedling seam has gap, the gray-scale value of the curve medium position pixel then obtained is comparatively large, and thus curve is the shape protruded to the right.
Figure of description
Fig. 1 is the structural representation of the device of invention.
Fig. 2 is seam system Software for Design process flow diagram.
Fig. 3 is gray-scale value extraction process process flow diagram.
Fig. 4 is binary conversion treatment process flow diagram.
Fig. 5 is seam recognizer process flow diagram.
Fig. 6 is with width 16 linearly gray-scale value summation curve.
Fig. 7 is the horizontal pixel point gray-scale value cumulative sum curve of seamless image.
Fig. 8 is the horizontal pixel point gray-scale value cumulative sum curve having slot image.
Embodiment
1. display 2. input and output and control gear 3. computing machine (image processing system) 4. light source controller 5. industrial camera 6. annular light source 7. camera lens 8. topworks 9. travelling belt 10. background board in FIG.
Software processes
The object of software process is the characteristic information digitizing grafting seedling seam, finally required characteristic information is fed back in the form of data.In the present invention, machine vision partial software program mainly comprises image acquisition, Image semantic classification, extraction grafting seedling seam feature 3 kinds of functions, and the image pre-processing method of application comprises greyscale transformation, thresholded image, image denoising, carries out closed operation computing to image.Grafting seedling seaming machines visual identity detection system design main-process stream as shown in Figure 2, after program starts, grafting seedling is processed successively, and the information of each strain grafting seedling process gained is finally shown on a display screen, after one strain grafting seedling is disposed, whether judge to also have in the dish of cave grafting seedling etc. pending, if no longer include seedling, program determination in the dish of cave, this serial procedures all will be completed by instruction automatically.
The image acquisition procedure MER-USBDevice that in use vision hardware system, selected camera is supporting gathers the image of vegetable grafting seedling seam.During by this programmed acquisition to the image of vegetable grafting seedling seam, the display of seedling image on a display screen, then opens image exchange software platform, the Image Acquisition in display screen in the internal memory of computing machine.
Image capture module is responsible for collection and the input of image, and it is connected with image acquisition release, and after computing machine receives the steering order of the image acquisition passage that control system sends, this module is started working.
According to the feature of grafting seedling seam, according to the principle of visual identity, carry out algorithm research by Matlab image processing software.First carry out coloured image greyscale transformation, then gray level image done thresholding process, according to gather and grey value characteristics in the image of binary conversion treatment, utilize the difference of grafting seedling seam and cane part gray-scale value to distinguish.For the image through binary conversion treatment, R, G, B value is all equal, is called gray-scale value, and each pixel has a gray-scale value.For the gray level image of 8, have 256, individual gray level, its intensity value ranges is 0 ~ 255.
Seam image after binary conversion treatment, grafting seedling cane part (comprising Grafting clip) that Grafting clip clamps is found out according to the lower edges of red Grafting clip, and then remove Grafting clip according to Iamge Segmentation principle, and to border shaping state closed operation calculation process, so just obtain the image comprising seam portion, again the longest straight line in this image is extended, extend to the border of zone line, then along straight line, cumulative summation is carried out to pixel gray-scale value with width 16, obtain curve shown in Fig. 6.Consider issuable inclination angle in grafting procedures, zone line segmentation time have employed the dividing method being parallel to clip, namely zone line if tilt, the clip zone line of inclination can be partitioned into.Analytic curve is known, if the cane part of one section of conventional scion seedling or rootstock seedling, then curve is close to a vertical straight line, and in the middle of Grafting clip, be one section of diameter that scion seedling and rootstock seedling combine, this curved intermediate part divides the pixel cumulative sum curve being seam position, if the seam quality of scion seedling and rootstock seedling is better, then scion seedling contacts closely with the grafting faying face of rootstock seedling, thus seam areas gray-scale value is less, therefore whole piece curve is the line that middle part falls in the left.If this grafting seedling seam has gap, then the gray-scale value of the curve medium position pixel obtained is comparatively large, and thus curve is the shape protruded to the right.
According to the feature of seam, according to the principle of visual identity, carry out algorithm research by Matlab image processing software.First the tomato diameter parts (comprising Grafting clip) that Grafting clip clamps is found out, and then remove Grafting clip according to Iamge Segmentation principle, obtain the image comprising seam portion like this, the more cumulative summation of horizontal direction pixel gray-scale value is carried out to this image, obtain curve shown in Fig. 7.Consider issuable inclination angle in grafting procedures, zone line segmentation time have employed the dividing method being parallel to clip, namely zone line if tilt, the clip zone line of inclination can be partitioned into.Analytic curve is known, if the stem portion of one section of conventional stock or scion, then curve is close to a vertical straight line, and in the middle of Grafting clip, be one section of diameter that stock and scion combine, this curved intermediate part divides the pixel cumulative sum curve being seam position, if the seam quality of stock and scion is better, then stock contacts closely with the grafting faying face of scion, thus seam areas gray-scale value is less, therefore whole piece curve is the line that middle part falls in the left.If this grafting seam has gap, then the gray-scale value of the curve medium position pixel obtained is comparatively large, and thus curve is the shape of giving prominence to the right, as shown in Figure 8.
Main code of program is as follows:
Claims (3)
1. the grafting seam recognition device of a full-automatic grafting machine, comprise industrial camera, luminaire, image pick-up card, computing machine, connecting gear and pot for growing seedlings, industrial camera is connected with image pick-up card by data line, luminaire is connected with computing machine by electric wire, image pick-up card is connected with computing machine by data line, industrial camera, luminaire, image pick-up card and computing machine form Vision Builder for Automated Inspection, described connecting gear, comprise driving-belt, first power wheel, second power wheel, motor and frame, the wheel shaft of the first power wheel and the second power wheel is all propping up in frame by bearing assemble, travelling belt twines and is located on the first power wheel and the second power wheel, motor is fixed in frame, and the output shaft of motor is connected by the wheel shaft of reductor with the first power wheel, motor is enable to drive the first power wheel to rotate, and then drive conveyer belt, it is characterized in that: described luminaire, comprise LED and the lamp socket of multiple greens of arrangement formation one ring plain structure, annular LED lamp to be enclosed within camera lens and to be fixed on lamp socket, lamp socket is fixed in lighting unit rack, support bottom is fixed by screws on the bracing frame of travelling belt, described pot for growing seedlings is placed on a moving belt, and the side end face of travelling belt is fixed with electronic moisture meter and electronic thermometer, described luminaire is arranged on the side of travelling belt, the annular LED lamp group outmost turns plane of luminaire and surface level perpendicular, the another side of travelling belt is fixed with illuminance sensor by expansion link, one end of expansion link is fixedly connected on the side end face of travelling belt, and the other end of expansion link is fixedly connected with illuminance sensor, described industrial camera is arranged on the rear of luminaire.
2. the grafting seam recognition device of a kind of full-automatic grafting machine according to claim 1, it is characterized in that: described industrial camera is Mercury series MER-125-30UC colorful digital CCD camera that company of Daheng produces, interface is USB2.0, and resolution is 1292 (H) × 964 (V); Image sensor chip, dimension scale 1/3inch (4.8 × 3.6mm); Computer series M2514-MP2 selected by camera lens, and focal length parameter is 25mm, and diameter of lens is 46mm.
3. a grafting seam recognition methods for full-automatic grafting machine, is characterized in that comprising the following steps:
Step one: preliminary work: the grafting seam recognition device of full-automatic grafting machine is placed in a space closed, control travelling belt transmission, seedling culture hole plate is made to stop at positive control funerary objects and industrial camera position, the camera lens 12cm of the grafting distance industrial camera on seedling culture hole plate, the annular LED lamp group of luminaire is placed on camera lens; Regulate lamp brilliance, the illuminance that illuminance sensor is shown is 130 ~ 150Lux; Regulate the temperature and humidity in enclosure space, making electronic moisture meter show humidity is 65 ~ 70%, and electronic thermometer displays temperature is 28 ~ 30 DEG C;
Step 2: image acquisition: maintain above-mentioned state after 10 ~ 15 minutes, industrial camera starts to gather image, then sends image information to image pick-up card, and last image pick-up card sends information to computing machine;
Step 3: image procossing: according to the feature of seam, according to the principle of visual identity, by Matlab software, first the tomato diameter parts (comprising Grafting clip) that Grafting clip clamps is found out, and then remove clip according to Iamge Segmentation principle, obtain the image comprising seam portion like this, the more cumulative summation of horizontal direction pixel gray-scale value is carried out to this image, obtain figure characteristic curve.
Wherein, the algorithm of image procossing comprises the following steps:
A. coloured image greyscale transformation: convert the coloured image of acquisition to gray level image, converts single pass gray-scale map to by R, G, B triple channel image.In gray level image, the gray-scale value of each pixel only uses a byte representation, gray scale span is 0 ~ 255, this point of the less expression of numerical value is more black also darker, 0 represents the most black, this point of the larger expression of numerical value is whiter also brighter, 255 represent the brightest namely complete white, and the gray level image after conversion does not contain color information, only containing monochrome information.Coloured image converts the formula of gray level image to such as formula (3-1), Gray (i in formula, j) after representing conversion, gray level image is at (i, j) gray-scale value put, R (i, j), G (i, j), B (i, j) represent the component information of color red, green, blue in coloured image respectively, 0.30,0.59,0.11 is respectively the coefficient that coloured image is weighted 3 variablees when average gray is changed.
Gray(i,j)=0.30R(i,j)+0.59G(i,j)+0.11B(i,j)(3-1)
B. gray level image threshold transformation: use adaptive threshold or dynamic threshold according to feature of image, formula (3-2)
In formula, f (x, y) is the gray-scale value after (x, y) some place pixel transform, and T is the threshold value of setting.
C. image binaryzation: according to the gray-value variation scope of image, choose a certain gray-scale value as threshold value T, image is divided into two parts, gray-scale value G (I, j) by any point (i, j) in image contrasts with T, if be more than or equal to T, the gray-scale value of this point will be set to 1, if be less than T, be just set to 0, shown in (3-3), otherwise Ru shown in (3-4).
Or
In formula: the gray-scale value that G (I, j) is original image, F (I, j) is the gray-scale value of image after binaryzation.
D. picture smooth treatment: the smoothing process of method using mean filter.Mean filter is also called smooth linear wave filter, adopt neighborhood averaging, the mean value of all pixels value of pixel each in neighborhood in image determined in order to the method for filtering mask replaces, namely the gray-scale value of the pending current pixel point (x, y) of assignment is carried out with the gray average of all pixels selected in template.For suppression Gaussian noise, mean filter has good effect, be mainly used in removing the incoherent details of pixel region little compared with filtering mask dimensions in image, formula (3-5) is shown in the calculating using the image of m × n (m and n is odd number) weighted mean to a width M × N to carry out filtering, a=(m-1)/2 in formula and b=(n-1)/2, the mask coefficient that w (s, t) is g (x+s, y+t).
E. morphology closed operation: first carry out expansion process (Dilation) to bianry image and carry out corrosion treatment (Erosion) to the image after process again, its definition is such as formula shown in (3-6).Closed operation process usually can narrow interruption or tiny hole in fused images, and eliminates and the incoherent fine edge of target, makes the outline line of target in bianry image become more smooth.Make up crack in bianry image usually through to image expansion process, to eliminate in bianry image incoherent details usually through to Image erosion process, the definition of dilation and erosion is respectively such as formula shown in (3-7) and (3-8).
Under normal circumstances, A is set to the set of bianry image, and B is set to the structural element for process, and structural element B is also 1 image collection, will operate in morphology operations with structural element to set.Namely expansion process arrives B after the first primitive translation z unit in structural element B
zif include B in A
zelement (not necessarily all comprising), then this point is retained, otherwise does not process this point, and the set that the final z point retained forms is exactly the result that B expands to A.
Step 4: computing machine judges the figure characteristic curve obtained according to the program set, detect qualified grafting and defective grafting, and then computing machine sends instruction to control system, carries out down-stream.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510459023.3A CN105118028B (en) | 2015-07-30 | 2015-07-30 | A kind of grafting seam identification device of full-automatic grafting machine and seam recognition methods |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510459023.3A CN105118028B (en) | 2015-07-30 | 2015-07-30 | A kind of grafting seam identification device of full-automatic grafting machine and seam recognition methods |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105118028A true CN105118028A (en) | 2015-12-02 |
CN105118028B CN105118028B (en) | 2018-07-06 |
Family
ID=54666003
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510459023.3A Expired - Fee Related CN105118028B (en) | 2015-07-30 | 2015-07-30 | A kind of grafting seam identification device of full-automatic grafting machine and seam recognition methods |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105118028B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106990045A (en) * | 2017-04-19 | 2017-07-28 | 浙江理工大学 | A kind of polishing identifying device of auxiliary machinery vision plug seedlings quality testing |
CN109654997A (en) * | 2019-02-01 | 2019-04-19 | 河南科技大学 | A kind of hole tray hole accurate-location device and method based on machine vision |
CN112130256A (en) * | 2020-11-06 | 2020-12-25 | 南京天兴通电子科技有限公司 | Novel optical fiber type identification system |
CN113295109A (en) * | 2021-06-25 | 2021-08-24 | 中国林业科学研究院林业新技术研究所 | System and method for measuring curvature of nursery stock suitable for mechanical grafting based on machine vision |
CN113826496A (en) * | 2021-11-01 | 2021-12-24 | 合肥佳富特机器人科技有限责任公司 | Automatic grafting method and device in vegetable grafting |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20050001281A1 (en) * | 2003-07-03 | 2005-01-06 | Hung-Jen Hsu | Process to improve image sensor sensitivity |
CN102682286A (en) * | 2012-04-16 | 2012-09-19 | 中国农业大学 | Fruit identification method of picking robots based on laser vision systems |
-
2015
- 2015-07-30 CN CN201510459023.3A patent/CN105118028B/en not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20050001281A1 (en) * | 2003-07-03 | 2005-01-06 | Hung-Jen Hsu | Process to improve image sensor sensitivity |
CN102682286A (en) * | 2012-04-16 | 2012-09-19 | 中国农业大学 | Fruit identification method of picking robots based on laser vision systems |
Non-Patent Citations (2)
Title |
---|
XU GUOYONG等: "Grafting of thermoresponsive polymer from the surface of functionalized multiwalled carbon nanotubes via atom transfer radical polymerization", 《CHINESE SCIENCE BULLETIN》 * |
贺磊盈等: "基于机器视觉的幼苗自动嫁接参数提取", 《农业工程学报》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106990045A (en) * | 2017-04-19 | 2017-07-28 | 浙江理工大学 | A kind of polishing identifying device of auxiliary machinery vision plug seedlings quality testing |
CN109654997A (en) * | 2019-02-01 | 2019-04-19 | 河南科技大学 | A kind of hole tray hole accurate-location device and method based on machine vision |
CN112130256A (en) * | 2020-11-06 | 2020-12-25 | 南京天兴通电子科技有限公司 | Novel optical fiber type identification system |
CN113295109A (en) * | 2021-06-25 | 2021-08-24 | 中国林业科学研究院林业新技术研究所 | System and method for measuring curvature of nursery stock suitable for mechanical grafting based on machine vision |
CN113826496A (en) * | 2021-11-01 | 2021-12-24 | 合肥佳富特机器人科技有限责任公司 | Automatic grafting method and device in vegetable grafting |
Also Published As
Publication number | Publication date |
---|---|
CN105118028B (en) | 2018-07-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105118028A (en) | Automatic grafting machine grafted seedling seam identifying apparatus and seam identifying method | |
CN108596880A (en) | Weld defect feature extraction based on image procossing and welding quality analysis method | |
CN106295789B (en) | Crop seed counting method based on image processing | |
CN106370667A (en) | Visual detection apparatus and method for quality of corn kernel | |
CN105389581B (en) | A kind of rice germ plumule integrity degree intelligent identifying system and its recognition methods | |
CN105930854A (en) | Manipulator visual system | |
CN108020556A (en) | A kind of online damage testing sorting technique of corn seed based on machine vision | |
CN109635806A (en) | Ammeter technique for partitioning based on residual error network | |
CN207238542U (en) | A kind of thin bamboo strip defect on-line detecting system based on machine vision | |
CN108318494B (en) | The red online vision detection and classification devices and methods therefor for proposing fruit powder | |
CN113674226A (en) | Tea leaf picking machine tea leaf bud tip detection method based on deep learning | |
CN110349125A (en) | A kind of LED chip open defect detection method and system based on machine vision | |
CN102495067B (en) | System for identifying impurities of edible funguses on line | |
CN109241948A (en) | A kind of NC cutting tool visual identity method and device | |
CN114199880A (en) | Citrus disease and insect pest real-time detection method based on edge calculation | |
CN105701812A (en) | Visual identification system suitable for cotton picking robot | |
CN103528967A (en) | Hyperspectral image based overripe Lonicera edulis fruit identification method | |
CN108665450A (en) | A kind of corn ear mechanical damage area recognizing method | |
CN115862004A (en) | Corn ear surface defect detection method and device | |
CN116559111A (en) | Sorghum variety identification method based on hyperspectral imaging technology | |
CN102519971A (en) | On-line identification apparatus and method for impurities in edible fungi | |
CN116258664A (en) | Deep learning-based intelligent defect detection method for photovoltaic cell | |
CN102680488B (en) | Device and method for identifying massive agricultural product on line on basis of PCA (Principal Component Analysis) | |
Firdous et al. | Deep convolutional neural network-based framework for apple leaves disease detection | |
CN110987957A (en) | Intelligent defect removing method based on machine vision and laser processing |
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 | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180706 Termination date: 20190730 |
|
CF01 | Termination of patent right due to non-payment of annual fee |