CN115365162A - Potato grading device based on machine vision and shape detection method - Google Patents
Potato grading device based on machine vision and shape detection method Download PDFInfo
- Publication number
- CN115365162A CN115365162A CN202211004584.0A CN202211004584A CN115365162A CN 115365162 A CN115365162 A CN 115365162A CN 202211004584 A CN202211004584 A CN 202211004584A CN 115365162 A CN115365162 A CN 115365162A
- Authority
- CN
- China
- Prior art keywords
- potato
- discharging
- kinect camera
- conveyer belt
- mounting plate
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 244000061456 Solanum tuberosum Species 0.000 title claims abstract description 211
- 235000002595 Solanum tuberosum Nutrition 0.000 title claims abstract description 211
- 238000001514 detection method Methods 0.000 title claims abstract description 77
- 230000007246 mechanism Effects 0.000 claims abstract description 154
- 238000007599 discharging Methods 0.000 claims abstract description 96
- 235000012015 potatoes Nutrition 0.000 claims abstract description 73
- 210000000078 claw Anatomy 0.000 claims abstract description 42
- 230000005540 biological transmission Effects 0.000 claims abstract description 10
- 238000000034 method Methods 0.000 claims description 30
- 230000007306 turnover Effects 0.000 claims description 19
- 230000008569 process Effects 0.000 claims description 17
- 230000000007 visual effect Effects 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 9
- 230000008859 change Effects 0.000 claims description 7
- 239000000463 material Substances 0.000 claims description 7
- 238000001914 filtration Methods 0.000 claims description 6
- 238000005192 partition Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 5
- 238000007781 pre-processing Methods 0.000 claims description 4
- 238000003709 image segmentation Methods 0.000 claims description 3
- 238000009434 installation Methods 0.000 claims 2
- 241000196324 Embryophyta Species 0.000 claims 1
- 238000005096 rolling process Methods 0.000 abstract description 7
- 238000012216 screening Methods 0.000 abstract description 5
- 239000012535 impurity Substances 0.000 abstract description 3
- 230000011218 segmentation Effects 0.000 description 9
- 230000008901 benefit Effects 0.000 description 4
- 230000007797 corrosion Effects 0.000 description 4
- 238000005260 corrosion Methods 0.000 description 4
- 238000003708 edge detection Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 210000004556 brain Anatomy 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000004064 recycling Methods 0.000 description 2
- 244000291564 Allium cepa Species 0.000 description 1
- 235000002732 Allium cepa var. cepa Nutrition 0.000 description 1
- 235000007688 Lycopersicon esculentum Nutrition 0.000 description 1
- 244000141359 Malus pumila Species 0.000 description 1
- 239000006002 Pepper Substances 0.000 description 1
- 240000003768 Solanum lycopersicum Species 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 235000021016 apples Nutrition 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000034303 cell budding Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 235000012055 fruits and vegetables Nutrition 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000002347 injection Methods 0.000 description 1
- 239000007924 injection Substances 0.000 description 1
- 238000003780 insertion Methods 0.000 description 1
- 230000037431 insertion Effects 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 235000016709 nutrition Nutrition 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 239000002689 soil Substances 0.000 description 1
- 230000002194 synthesizing effect Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
- 238000004148 unit process Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/34—Sorting according to other particular properties
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/02—Measures preceding sorting, e.g. arranging articles in a stream orientating
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/36—Sorting apparatus characterised by the means used for distribution
- B07C5/361—Processing or control devices therefor, e.g. escort memory
- B07C5/362—Separating or distributor mechanisms
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/36—Sorting apparatus characterised by the means used for distribution
- B07C5/38—Collecting or arranging articles in groups
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C2501/00—Sorting according to a characteristic or feature of the articles or material to be sorted
- B07C2501/009—Sorting of fruit
Abstract
The invention discloses a potato grading device based on machine vision. The device comprises a feeding mechanism, a shape and quality detection mechanism and a distribution mechanism; the shape detection mechanism is mainly composed of a transmission rolling shaft and a secondary conveyer belt which are arranged up and down, and two layers of clapboards are arranged above the rolling shaft to enable potatoes to be conveyed in three longitudinal rows; a first Kinect camera and a second Kinect camera are mounted on the rolling shaft positioning mounting plate; a turning claw is arranged between the cameras and is used for adjusting the posture of the potatoes below; the distribution mechanism mainly comprises a discharging conveyer belt and a distributor, and long baffles are arranged on two opposite sides of the discharging conveyer belt; the baffle of the discharging conveyer belt is provided with a discharging groove; and a distributor taking the air cylinder as power is arranged on the baffle plate of the discharging conveying belt. The device can realize shape detection and quality estimation of large-batch potatoes, combines images, has high screening accuracy and high efficiency, can recover fallen impurities, and has small damage to the potatoes.
Description
Technical Field
The invention relates to the field of agricultural machinery, in particular to a potato grading device based on machine vision and a shape detection method.
Background
The potato is one of the main grain crops in China, the planting history is hundreds of years, and compared with other grain crops, the potato has the advantages of high yield, short growth cycle, rich nutritional value and the like. Since the 21 st century, china gradually establishes potato cultivation demonstration culture bases in different areas, not only obtains good economic benefits, but also promotes sustainable development of agricultural economy, and plays an extremely important role in adjusting rural industrial structures and ensuring yield increase. Based on the above, the research of potato-related industries is also of great importance to the development of agriculture in China in the future.
The first step in the potato processing industry is to screen out potatoes that meet certain criteria, such as setting a certain shape requirement, setting a certain quality interval, etc. At present, the method adopted for screening the potatoes is mostly a manual method or a traditional rolling-bar drum type machine. The screening method with high efficiency, high intelligence and high automation level is urgently needed to be put into the market. The machine vision simulates biological vision by using an acquisition device and a computer, and the main task is to acquire image information of a corresponding scene by processing acquired pictures or videos, like human eyes can acquire visual information and transmit the visual information to a brain, the brain is used as a control center to process and process the information, and the machine vision method is utilized to perform the work.
Disclosure of Invention
The invention aims to provide a potato grading device based on machine vision, which has the advantages of high automation degree and simple structure; the invention also aims to provide a potato shape and quality detection method based on machine vision, which combines a machine vision technology, can grade the shape or quality of potatoes as a standard and has the advantage of high grading efficiency.
In order to achieve the purpose, the potato grading device based on machine vision comprises a feeding mechanism, a shape quality detection mechanism and a distribution mechanism, wherein the feeding mechanism, the shape quality detection mechanism and the distribution mechanism are sequentially connected;
the feeding mechanism mainly comprises a feeding conveying belt;
the shape quality detection mechanism mainly comprises a transmission rolling shaft and a secondary conveyer belt, wherein the transmission rolling shaft and the secondary conveyer belt are arranged up and down, and the transmission rolling shaft is arranged above the secondary conveyer belt;
the conveying roller is formed by installing a plurality of rollers between a first roller positioning mounting plate and a second roller positioning mounting plate, and the rollers are connected with a conveying roller motor through belts;
a first Kinect camera and a second Kinect camera are mounted on the first roller positioning mounting plate; the shape quality detection mechanism is provided with a first Kinect camera and a second Kinect camera in sequence along the direction from a potato feeding conveyor belt to a first-stage unloading conveyor belt and a second-stage unloading conveyor belt, and a turnover claw mechanism is arranged between the first Kinect camera and the second Kinect camera;
the overturning claw mechanism is provided with overturning claws, each overturning claw is composed of 3 rows of claw-shaped structures which are closely arranged in an axial direction, and the included angle between every two rows of claws is 120 degrees;
the distribution mechanism consists of a conveying module and a discharging module, and the conveying module is provided with the discharging module; the conveying module is formed by connecting a primary unloading conveying belt and a secondary unloading conveying belt; the discharging module consists of an air cylinder and a discharging groove; the discharge chute comprises a first discharge chute mechanism and a second discharge chute mechanism;
the two opposite sides of the primary discharging conveyer belt are respectively provided with a first discharging conveyer belt baffle, a first discharging groove mechanism is arranged on the first discharging conveyer belt baffle, the first discharging groove mechanism is correspondingly provided with an air cylinder module, the air cylinder module comprises an air cylinder and an air cylinder connecting rod, the tail end of the rod is provided with a distributing device, and the air cylinder module contracts to drive the distributing device to move towards the first discharging groove mechanism;
second discharging conveyer belt baffles are arranged on two opposite sides of the second-stage discharging conveyer belt, a second discharging groove mechanism is arranged on each second discharging conveyer belt baffle, a cylinder module is correspondingly arranged on each second discharging groove mechanism, each cylinder module comprises a cylinder and a cylinder connecting rod, a distributing device is arranged at the tail end of each rod, and the cylinder module contracts to drive the distributing device to move towards the second discharging groove mechanism;
the feeding conveyer belt is connected with the conveying roller, the conveying roller is connected with the first-level discharging conveyer belt, and the first-level discharging conveyer belt is connected with the second-level discharging conveyer belt.
The baffle mechanism is all installed to the long limit both sides of material loading conveyer belt, the material loading conveyer belt is connected with the material loading groove.
The shape quality detection mechanism is an image detection component and further comprises an industrial camera height adjusting support; the industrial camera height adjusting bracket comprises a first Kinect camera bracket and a second Kinect camera bracket;
a first Kinect camera support and a first Kinect camera support base are sequentially connected between the first Kinect camera and the first roller positioning mounting plate, the first Kinect camera support is of a telescopic rod structure, and the first Kinect camera support base is fixedly connected with the first roller positioning mounting plate;
second Kinect camera support, second Kinect camera support base have connected gradually between second Kinect camera and the first roller bearing location mounting panel, second Kinect camera support is telescopic rod structure, second Kinect camera support base and first roller bearing location mounting panel fixed connection.
The potato grading device based on machine vision further comprises an integral support, wherein the integral support comprises a feeding mechanism support, a shape quality detection mechanism support, a first-stage unloading conveying belt support and a second-stage unloading conveying belt support; the bottom of the feeding mechanism is provided with a feeding mechanism support, the bottom of the shape quality detection mechanism is provided with a shape quality detection mechanism support, and the bottom of the distribution mechanism is provided with a first-stage unloading conveying belt support and a second-stage unloading conveying belt support.
The roller is arranged between the first roller positioning mounting plate and the second roller positioning mounting plate through a bearing, and positioning hinges are arranged between the shape quality detection mechanism support and the first roller positioning mounting plate as well as between the shape quality detection mechanism support and the second roller positioning mounting plate.
The turnover claw is arranged between the first roller positioning mounting plate and the second roller positioning mounting plate;
one end of the turning claw is arranged on the first roller positioning mounting plate through a first turning claw bracket,
the other end of the turning claw is arranged on the second roller positioning mounting plate through a second turning claw support.
And the distribution mechanisms are respectively provided with an infrared generator and a corresponding receiver for detecting the potatoes passing through the material guide chute.
The shape quality detection mechanism is divided into three paths by arranging a partition plate 1, and the first-stage discharging conveying belt is divided into three paths by arranging a partition plate 2.
The two sides of the first-stage discharging conveying belt are provided with a plurality of first discharging groove mechanisms, and the two sides of the second-stage discharging conveying belt are provided with a plurality of second discharging groove mechanisms.
The invention relates to a potato shape and quality detection method based on machine vision, which comprises the following steps:
(1) the first Kinect camera acquires overhead images of the lower three rows of potatoes in the state;
(2) the control end receives the first Kinect camera to acquire a potato overlook image and processes information;
(3) the control end executes the preprocessing operation of the original potato picture and runs a watershed algorithm to mark different potatoes;
performing LOG operator to detect the edge of the potato, judging whether the edge of the potato is missing or not, performing image repairing operation if the edge of the potato is missing, and finally performing shape detection by combining the potato rectangular degree parameter, the minimum external rectangle length-width ratio and the circularity parameter;
combining the collected two-position images of the potatoes with the depth image to generate a point cloud image, and performing image segmentation and filtering processes to obtain a point cloud set of the upper half part of the potatoes; performing surface triangular mesh division on the point cloud set of the upper half part of the potato, combining an internal central point to form a tetrahedral pyramid model of the upper half part of the potato, and multiplying the tetrahedral pyramid model by the density of the potato according to a volume calculation method of a three-dimensional geometric body in a space to obtain the mass of half of the potato;
(4) the turning claw turns the potatoes on the rolling shaft, so that the potatoes change the upward body state;
(5) after the potatoes are adjusted to be in an upward state by the turnover mechanism, the second Kinect camera collects three rows of downward potato overhead images, and collects the same batch of potato images at different visual angles;
(6) the control end receives an overhead view image of the lower potatoes in a changed state acquired by the second Kinect camera and processes information; performing shape detection and quality estimation as in the third step;
(7) analyzing image processing results of the first Kinect camera and the second Kinect camera, and judging the final shape and the final quality of the potato;
(8) and the control end sends an execution command to the allocation mechanism for execution.
The poking device is a square poking sheet.
The first Kinect camera shoots a potato target which is an overlook image before turning, and the second Kinect camera shoots an image after turning the potato; the first Kinect camera and the second Kinect camera are fixed, the target is turned over, and different faces of the same batch of potatoes are shot at different visual angles.
In the step (3), the shape detection of the batch of potatoes is performed, and meanwhile, missing images are repaired, so that the judgment accuracy is improved; the quality detection method based on machine vision comprises the steps of constructing a three-dimensional model by utilizing a potato depth image, detecting key feature points of the three-dimensional model, dividing a triangular mesh by taking the key feature points as nodes, further obtaining a tetrahedral cone model structure, and quickly calculating the volume.
The invention discloses a potato grading device based on machine vision and a shape detection method, which have the beneficial effects that: the method can realize intelligent shape quality screening of the potatoes, and can improve the accuracy by combining two images with different visual angles of the same batch of potatoes in different body position states; meanwhile, impurities which are easy to fall off can be recovered; the damage to the potatoes is small; in addition, the sorting machine can also be used for sorting different fruits and vegetables such as apples, tomatoes, onions and the like.
Drawings
FIG. 1 is a schematic perspective view of a potato grading apparatus;
FIG. 2 is a top view of a potato grading apparatus;
FIG. 3 is a front view of the potato grader;
FIG. 4 is a right side view of the potato grading apparatus;
FIG. 5 is a partially enlarged view of a camera inspection portion;
FIG. 6 is an enlarged view of a portion of the inverting apparatus;
FIG. 7 is an enlarged partial view of the dispenser;
FIG. 8 is a cross-sectional view of the detection section;
FIG. 9 is an isometric view of the invert jaw;
FIGS. 10.1-10.17 are inspection images and analysis;
in the figure: in the figure: 1-feeding groove, 2-feeding conveyer belt, 3-first downward sliding guide piece, 4-first Kinect camera, 5-turnover claw mechanism, 6-second Kinect camera, 7-clapboard 1, 8-second downward sliding guide piece, 9-clapboard 2, 10-first unloading conveyer belt baffle, 11-first-level unloading conveyer belt, 12-second-level unloading conveyer belt, 13-feeding mechanism bracket, 14-shape quality detection mechanism bracket, 15-control panel, 16-control box, 17-second-level conveyer belt, 18-recycling bin, 19-first unloading groove mechanism, 20-first-level unloading conveyer belt bracket, 21-second-level unloading conveyer belt bracket, 22-a-first roller positioning mounting plate, 22-b-second roller positioning mounting plate, 23-first Kinect camera bracket base, 24-first Kinect camera bracket, 26-a-first turnover claw bracket, 26-b-second turnover claw bracket, 27-turnover claw, 28-turnover part, 29-positioning part, 30-second turnover part, 31-roller shifting device, 32-cylinder, 35-second unloading groove conveying mechanism, and connecting rod conveying mechanism.
Detailed Description
Example 1
The invention relates to a potato grading device based on machine vision, which is shown in figures 1-9; including feed mechanism, shape quality detection mechanism, branch dial mechanism, its characterized in that: the feeding mechanism, the shape quality detection mechanism and the distribution mechanism are sequentially connected;
the feeding mechanism mainly comprises a feeding conveying belt 2;
the shape quality detection mechanism mainly comprises a transmission roller 35 and a secondary conveyer belt 17, wherein the transmission roller 35 and the secondary conveyer belt 17 are arranged up and down, and the transmission roller 35 is arranged above the secondary conveyer belt 17;
the conveying roller 35 is formed by installing a plurality of rollers 35-1 between a first roller positioning mounting plate 22-a and a second roller positioning mounting plate 22-b, and the rollers 35-1 are connected with a conveying roller motor through belts;
a first Kinect camera 4 and a second Kinect camera 6 are mounted on the first roller positioning mounting plate 22-a; the shape quality detection mechanism is sequentially provided with a first Kinect camera 4 and a second Kinect camera 6 along the direction from a potato feeding conveyer belt 2 to a first-stage discharging conveyer belt 11 and a second-stage discharging conveyer belt 12, and a turnover claw mechanism 5 is arranged between the first Kinect camera 4 and the second Kinect camera 6;
the overturning claw mechanism 5 is provided with overturning claws 27, the overturning claws 27 are axially formed by 3 rows of claw-shaped structures which are closely arranged, and the included angle between every two rows of claws is 120 degrees;
the distribution mechanism consists of a conveying module and a discharging module, and the conveying module is provided with the discharging module; the conveying module is formed by connecting a primary unloading conveying belt 11 and a secondary unloading conveying belt 12; the discharging module consists of an air cylinder 33 and a discharging groove; the discharge chute comprises a first discharge chute mechanism 19 and a second discharge chute mechanism 34;
the two opposite sides of the first-stage discharging conveyer belt 11 are provided with first discharging conveyer belt baffles 10, a first discharging groove mechanism 19 is arranged on each first discharging conveyer belt baffle 10, the first discharging groove mechanism 19 is correspondingly provided with a cylinder module, each cylinder module comprises a cylinder and a cylinder connecting rod, the tail end of each rod is provided with a distributor, and the cylinder module contracts to drive the distributor to move towards the first discharging groove mechanism 19;
second discharging conveyor belt baffles 30 are arranged on two opposite sides of the second-stage discharging conveyor belt 12, a second discharging groove mechanism 34 is arranged on each second discharging conveyor belt baffle 30, a cylinder module is correspondingly arranged on each second discharging groove mechanism 34, each cylinder module comprises a cylinder and a cylinder connecting rod, a distributor is arranged at the tail end of each rod, and the cylinder module contracts to drive the distributor to move towards the second discharging groove mechanisms 34;
the feeding conveyer belt 2 is connected with a conveying roller 35, the conveying roller 35 is connected with a first-level discharging conveyer belt 11, and the first-level discharging conveyer belt 11 is connected with a second-level discharging conveyer belt 12.
Baffle mechanisms are installed on two sides of the long edge of the feeding conveyer belt 2, and the feeding conveyer belt 2 is connected with the feeding groove 1.
The shape quality detection mechanism is an image detection component and further comprises an industrial camera height adjusting support; the height adjusting bracket for the industrial camera comprises a first Kinect camera bracket 24 and a second Kinect camera bracket;
a first Kinect camera bracket 24 and a first Kinect camera bracket base 23 are sequentially connected between the first Kinect camera 4 and the first roller positioning mounting plate 22-a, the first Kinect camera bracket 24 is of a telescopic rod structure, and the first Kinect camera bracket base 23 is fixedly connected with the first roller positioning mounting plate 22-a;
and a second Kinect camera bracket base are sequentially connected between the second Kinect camera 6 and the first roller positioning mounting plate 22-a, the second Kinect camera bracket is of a telescopic rod structure, and the second Kinect camera bracket base is fixedly connected with the first roller positioning mounting plate 22-a.
The potato grading device based on machine vision further comprises an integral support, wherein the integral support comprises a feeding mechanism support 13, a shape quality detection mechanism support 14, a primary discharging conveying belt support 20 and a secondary discharging conveying belt support 21; the bottom of the feeding mechanism is provided with a feeding mechanism support 13, the bottom of the shape quality detection mechanism is provided with a shape quality detection mechanism support 14, and the bottom of the distribution mechanism is provided with a first-stage discharging conveyer belt support 20 and a second-stage discharging conveyer belt support 21.
The roller 35 is mounted between the first roller positioning mounting plate 22-a and the second roller positioning mounting plate 22-b through a bearing, and the shape quality detection mechanism support 14 and the first roller positioning mounting plate 22-a and the second roller positioning mounting plate 22-b are both provided with positioning hinges 29.
The overturning claw 27 is arranged between the first roller positioning mounting plate 22-a and the second roller positioning mounting plate 22-b;
one end of the flipping claw 27 is mounted to the first roller positioning mounting plate 22-a through the first flipping claw bracket 26-a,
the other end of the flipping claw 27 is mounted to the second roller positioning mounting plate 22-b by a second flipping claw bracket 26-b.
And the distribution mechanisms are respectively provided with an infrared generator and a corresponding receiver for detecting the potatoes passing through the material guide chute.
The shape quality detection mechanism is divided into three paths by arranging a partition plate 17, and the primary discharging conveyor belt 11 is divided into three paths by arranging a partition plate 29.
The two sides of the first-stage discharging conveyer belt 11 are provided with a plurality of first discharging groove mechanisms 19, and the two sides of the second-stage discharging conveyer belt 12 are provided with a plurality of second discharging groove mechanisms 34.
The invention relates to a potato shape and quality detection method based on machine vision, which comprises the following steps:
(1) the first Kinect camera 4 acquires overhead images of the lower three rows of potatoes in the state;
(2) the control end receives the first Kinect camera 4 to acquire the overhead images of the potatoes and processes information;
(3) the control end executes the preprocessing operation of the original potato picture and operates the watershed algorithm to mark different potatoes;
performing potato edge detection by using an LOG operator, judging whether the edge of the potato is missing or not, performing image repairing operation if the edge of the potato is missing, and finally performing shape detection by combining the potato rectangularity parameter, the minimum external rectangle length-width ratio and the circularity parameter;
combining the collected two-position potato image and the depth image to generate a point cloud image, and performing image segmentation and filtering processes to obtain a point cloud set of the upper half part of the potato; performing surface triangular mesh division on the point cloud set of the upper half part of the potato, combining an internal central point to form a tetrahedral pyramid model of the upper half part of the potato, and multiplying the volume by the density of the potato according to a volume calculation method of a three-dimensional geometrical body in space to obtain the mass of half of the potato;
(4) the turning claw 27 turns the potatoes on the roller to change the upward body state of the potatoes;
(5) after the potatoes are adjusted to be in an upward state by the turnover mechanism 5, the second Kinect camera 6 collects three rows of downward potato overhead images, and collects the same batch of potato images at different visual angles;
(6) the control end receives an overhead image of the lower potatoes in a changed state acquired by the second Kinect camera 6 and processes information; performing shape detection and quality estimation as in the third step;
(7) analyzing the image processing results of the first Kinect camera 4 and the second Kinect camera 6 to judge the final shape and the final quality of the potato;
(8) and the control end sends an execution command to the distribution mechanism for execution.
The first Kinect camera 4 shoots a potato target which is an overhead image before turning, and the second Kinect camera 6 shoots an image after turning the potato; the first Kinect camera 4 and the second Kinect camera 6 are fixed, the target is turned over, and different faces of the same batch of potatoes are shot at different visual angles.
In the step (3), the shape detection of the batch of potatoes is performed, and meanwhile, missing images are repaired, so that the judgment accuracy is improved; the quality detection method based on machine vision comprises the steps of constructing a three-dimensional model by utilizing a potato depth image, detecting key feature points of the three-dimensional model, dividing a triangular mesh by taking the key feature points as nodes, further obtaining a tetrahedral cone model structure, and quickly calculating the volume.
Example 2
As shown in FIG. 1, the potato grading device based on machine vision comprises a feeding mechanism, a shape quality detection mechanism and a distribution mechanism, wherein the feeding mechanism, the shape quality detection mechanism and the distribution mechanism are sequentially connected;
the feeding mechanism mainly comprises a feeding conveying belt 2; feeding trough 1 is arranged above feeding conveyer belt 2, feeding trough 1 is installed on the complete machine support, feeding trough 1 has seted up the quad slit, be convenient for the potato fall down on feeding conveyer belt 2, it gets into shape detection mechanism to transport the potato by feeding conveyer belt 2, first gliding guide 3 is installed in the exit of feeding conveyer belt 2, make the potato fall at conveying roller 35, feeding conveyer belt 2 is the low-speed conveyer belt, feeding conveyer belt 2 is driven by the low-speed motor, be provided with control box 16 in the complete machine support, the controller is installed in control box 16, install control panel 15 on shape and the quality detection mechanism, control panel 15 is connected with the low-speed motor, control panel 15 can adjust the low-speed motor rotational speed, control panel 15 and 16 electric connection of control box.
As shown in fig. 1, 5 and 6, potatoes sequentially enter the detection mechanism, a conveying roller 35 consists of a plurality of rollers 35-1, a first roller positioning mounting plate 22-a and a second roller positioning mounting plate 22-b which are opposite to each other, the rollers 35-1 are fixed between the first roller positioning mounting plate 22-a and the second roller positioning mounting plate 22-b through bearings, positioning hinges 29 are respectively arranged between the first roller positioning mounting plate 22-a and the second roller positioning mounting plate 22-b and the shape and quality detection mechanism support 14, the conveying roller 35 is driven by a belt, the belt is driven by a conveying roller motor, and the motor is controlled by a control panel 15;
the detection mechanism can be used as a shape detection component and a quality detection component, and also comprises an industrial camera height adjusting bracket; the industrial camera height adjusting bracket comprises a first Kinect camera bracket 24 and a second Kinect camera bracket;
a first Kinect camera bracket 24 and a first Kinect camera bracket base 23 are sequentially connected between the first Kinect camera 4 and the first roller positioning mounting plate 22-a, the first Kinect camera bracket 24 is of a telescopic rod structure, the first Kinect camera bracket base 23 is fixedly connected with the first roller positioning mounting plate 22-a, and a turning claw 27 is arranged between the first Kinect camera 4 and the second Kinect camera 6;
a second Kinect camera support and a second Kinect camera support base are sequentially connected between the second Kinect camera 6 and the first roller positioning mounting plate 22-a, the second Kinect camera support is of a telescopic rod structure, and the second Kinect camera support base is fixedly connected with the first roller positioning mounting plate 22-a;
the potatoes fall on the conveying roller 35, firstly pass through the first Kinect camera 4, the first Kinect camera 4 collects multi-frame images per second in real time, the first Kinect camera 4 transmits a plurality of pieces of data to the control box 16, the processing unit processes the data according to a programmed program, and the results are output and delivered to the distribution mechanism to be executed; the potatoes pass through a turnover mechanism 5; the turnover mechanism 5 is composed of 3 rows of claw-shaped structures which are closely arranged, the included angle between every two rows of claws is 120 degrees, the rotation speed of the turnover mechanism 5 is matched with the rotation speed of the roller 35, and potatoes passing through the turnover mechanism 5 can be turned over; the turnover mechanism 5 is arranged between the first roller positioning mounting plate 22-a and the second roller positioning mounting plate 22-b, and turnover mechanism positioning parts 28 are arranged between the first roller positioning mounting plate 22-a, the second roller positioning mounting plate 22-b and the turnover mechanism 5; the distance between the first overturning claw support 26-a and the first roller positioning mounting plate 22-a is adjustable, and the distance between the second overturning claw support 26-b and the second roller positioning mounting plate 22-b is adjustable, so that the mounting height of the overturning mechanism 5 is adjustable, and effective overturning force is exerted on three passing rows of potatoes;
finally, the potato passes through the second Kinect camera 6 again, the image of the potato at another visual angle is transmitted to the control box 16, and the result is output; comparing the two results, and judging the final shape of the potato according to the data set.
As shown in fig. 1, 3 and 8, a secondary conveyer belt 17 is installed below the conveying passage of the conveying roller 35, the secondary conveyer belt 17 is connected by the shape quality detecting mechanism support 14, and the secondary conveyer belt 17 conveys impurities such as soil dropped from the conveying roller 35 into the recycling bin 18;
as shown in fig. 1 and 5, a first Kinect camera bracket 24 and a first Kinect camera bracket base 23 are sequentially connected between the first Kinect camera 4 and the first roller positioning mounting plate 22-a, the first Kinect camera bracket 24 is a telescopic rod structure, and the first Kinect camera bracket base 23 is fixedly connected with the first roller positioning mounting plate 22-a;
as shown in fig. 1 and 7, the distribution mechanism mainly comprises two parts of conveying belts, wherein the first part of conveying belts can be used for feeding two rows of potatoes at the outermost layer, and the potatoes at the innermost layer are treated by the second part of conveying belts. The two opposite sides of the primary discharging conveyer belt 11 are provided with first discharging conveyer belt baffles 10; a first discharge chute mechanism 19 is arranged on the first discharge conveyer belt baffle 10, a distributor 31 is arranged on the first discharge conveyer belt baffle 10, the power of the distributor 31 is transmitted by a cylinder 33, and the first discharge chute mechanism 19 is matched with the distributor 31; the structure of the secondary discharging conveyer belt 12 is similar to that of the secondary discharging conveyer belt;
the distributor 31 is connected with an air cylinder 33 through a connecting rod 32, the air cylinder is fixed on the baffle plates 10 and 30 of the blanking mechanism, the distributor 31 is arranged on the baffle plates 10 and 30 on the two opposite sides of the one-level discharging conveying belt 11, the distributor 31 is connected with an air cylinder mechanism, the air cylinder mechanism is connected with a control box 16, and the air cylinder mechanism drives the distributor 31 to move. When the infrared detection device detects potatoes passing through each discharge chute, signals are transmitted to the control box 16, the control box 16 processes the potatoes, specific distributor execution commands are output according to detection results, the distributors 31 are enabled to carry out rapid linear reciprocating motion, the potatoes conforming to the shapes or the qualities of the flow guides enter the corresponding discharge chutes, and the shape or the quality grading of the potatoes is completed.
The invention relates to a potato shape detection method based on machine vision, which comprises the following specific implementation steps of:
(1) The first Kinect camera 4 collects a top view image of one side of the potato; the first Kinect camera 4 and the second Kinect camera 6 are provided with light source generators, and color images of potatoes in different body states in a conveying channel of the conveying roller 35 are acquired;
(2) The control end firstly receives the potato overlook image collected by the first Kinect camera 4 and processes information;
(1) graying the color image by using a weighted average method, carrying out weighted average on three components of R/G/B according to the importance and other indexes of elements by different weights, integrating, carrying out weighted average on the three components of RGB according to the following formula to obtain a more reasonable grayscale image, as shown in figure 10.1, and further obtaining a corresponding grayscale histogram so as to be convenient for filtering;
Gray(i，j)＝0.229*R(i，j)+0.578G(i,j)+0.114*B(i,j) (1.1)
wherein Gray (i, j) represents a Gray value of the coordinate point (i, j);
r (i, j) represents the R luminance of the (i, j) point;
g (i, j) represents the G luminance of the (i, j) point;
b (i, j) represents B luminance of the (i, j) point;
(2) by using the median filter, not only the most frequently occurring salt-pepper noise can be well removed, but also the complex gaussian noise and multiplication noise can be well filtered, as shown in fig. 10.2 below;
(3) performing threshold segmentation on the potato image by using an iteration method to segment areas containing the potatoes and areas not containing the potatoes, wherein the areas cannot be communicated;
first, the average gray level T of the image is calculated 0 (ii) a The image can be divided into two parts by taking the average gray level as a boundary; the average gray levels of the two parts are respectively calculated, and the average gray levels are larger than T 0 T of A And is less than T 0 T of B (ii) a Then obtain T A And T B Is set as T 1 Let T 1 Instead of T 0 Repeating the above process for k times until T k Converging;
when a code experiment is written specifically, a potato image can be divided into a foreground scene and a background scene according to an original iteration function f (x), after the data of the image for the first time is read, the average value of the threshold values of the two scenes is used as a new threshold value, and the threshold value control function f (x) is used for dividing the potato image into the foreground scene and the background scene again and is used as a new iteration function f' (x); iteration is carried out until a new iteration function is not generated, and the foreground and the background of the segmented image are final results; as shown in fig. 10.3;
(4) generally, due to the noise of the image, the edges of the potato image obtained after threshold segmentation are not smooth, the inner area has some noise holes, and the background area has some disordered noise points; therefore, multiple operations of combining expansion and corrosion can be carried out, and the image can achieve the ideal effect of processing; the potato is corroded and then expanded, so that small noise can be filtered, the potatoes are divided at the boundary, the edge of the potatoes is smoothed, and meanwhile, the potato area cannot be greatly changed; by using the expansion and corrosion, the tiny holes in the potatoes can be filled, and meanwhile, the pixels are repaired by smoothing the edges, but the area of the area is not changed. The effect is shown in fig. 10.4, and it can be seen that the operation effect is better when the expansion is used firstly and then the corrosion is used secondly;
(5) after the position of the target object is highlighted by the potato image through the operation, the edge of the potato is extracted through comparison of different operators, and accurate potato edge position information can be identified to make a basis for subsequent potato shape detection and judgment; the LOG operator combining Laplacian with Gaussian operator Gauss is used for potato edge detection, and a smooth filtering mode exists in the LOG operator, so that noise can be better filtered, and the detection effect is better; the Laplacian operator is used as a second derivative, and is mainly used for positioning the position of an image point in a segmented image, namely on a target object or a background; the edge line position is searched by using the characteristic, and the convolution kernel thereof is a 4 x 4 matrix:
the Laplacian operator formula for function f (x, y) is:
the LOG operator of the output function h (x, y) operates as:
that is to say that the first and second electrodes,
wherein, the first and the second end of the pipe are connected with each other,
the detection result of the detection operator is shown in fig. 10.5;
(6) because the potatoes have different sizes, when the camera is positioned right above the potato target and collects a plurality of potato images on the roller, the following situations may occur, as shown in fig. 10.6: the edge of the individual potato is shielded or the edge of the potato is damaged such as sprouting. At the moment, the camera cannot accurately and comprehensively shoot all edges of the potatoes, the edges are easy to lose at the budding position during subsequent edge detection, and partial shielding potato edge prediction and filling processing is carried out before the shape of the potatoes is detected so as to accurately detect the shapes of the potatoes in batches subsequently;
firstly, a watershed algorithm is applied to the segmentation of a single target image of multiple potatoes, and a change region is searched for and a segmentation region is obtained by combining the watershed algorithm with the pixel change on the image; generally, the difference between the pixel values of a target and the adjacent area of the target on an image is large, and gradient change and mutation exist with high probability, and the edge position of the target can be searched by utilizing the principle;
the specific implementation process of the independent region segmentation of the batch potato images by using the watershed algorithm comprises the following steps:
1) Carrying out preprocessing processes such as graying, binaryzation, filtering operation, corrosion or expansion, edge detection and the like on the target image;
2) Solving an image gradient map, finding a contour and marking, and drawing the contour and the marking sequence on a reference target image of a watershed to be equivalent to marking a water injection point;
3) Executing Watershed Watershed operation;
4) Drawing the regions to be separated, each individual target being marked with a different color to distinguish between different targets;
5) Separating the independent target areas; the results are shown in FIG. 10.7;
and adopting a method for synthesizing the missing edge texture according to the known region. The process is that the curvature of the edge end point of the nearest potato from the image missing position is calculated firstly, and attention is paid to the curvature of the end point of the circular arc, because the curvatures of all points on a certain arc are different, only the curvature of the point on the circular arc can be said; calculating the curvatures of two end points of the missing part, then obtaining an average curvature, and drawing an arc according to the average curvature, the end point 1 and the end point 2 so as to obtain a missing edge line; and finally, sampling the pixels in the known area, and transplanting the pixels to the missing part by taking the edge line as a boundary. The results are shown in FIG. 10.8;
(7) calculating and simultaneously displaying the squareness, the length-width ratio of the minimum external rectangle and the circularity parameters of the potato target, and judging the shape; the minimum circumscribed rectangle method for the potatoes comprises the following steps of firstly establishing a two-dimensional rectangular coordinate system by taking the centroid of the potatoes as an origin, then rotating the boundary of an object within a 90-degree range by a certain angle increment (for example, by taking 4 degrees as a fixed increment) every time, and recording the maximum value and the minimum value of circumscribed rectangle boundary points in the coordinate system direction once every time the boundary rotates; after rotating to a certain angle, the area (or perimeter) of the circumscribed rectangle reaches the minimum; taking the parameters of the circumscribed rectangle with the smallest area in all the rectangles as the length and the width under the main shaft meaning; calculating the value of a rectangle fitting factor by the width to length ratio, wherein the value of a thinner potato is smaller;
for a round potato, the squareness parameter is maintained to be near 0.8, the length-width ratio of the minimum circumscribed rectangle is close to 1, and the circularity parameter is generally more than 0.9. The squareness parameter of the oval potato is lower than 0.8, the length-width ratio of the circumscribed rectangle is lower than 1, but the values of the squareness parameter and the circularity parameter are not very low, and the circularity parameter is also lower than and far from 0.9, so that the lower limit of the values can be set for screening the oval potato;
the results of the computer display parameters are shown in FIG. 10.9, and the shapes of the potatoes are comprehensively judged by combining the data sets;
(3) The potatoes on the conveying roller pass through the turnover mechanism, the upward surface of the potatoes is changed, and the second Kinect camera also collects images to be processed as above;
(4) Through two-step image processing of three rows of potatoes arranged in a fixed sequence, related characteristic parameters of the same batch of potatoes under different viewing angles can be obtained. Analyzing the relevant parameters of the two visual angles, and comprehensively judging the shape;
(5) Three rows of potatoes enter the distribution mechanism in a fixed sequence, the infrared detection device can detect the potatoes passing through each material guide chute in sequence, and the control box controls the distribution device to execute commands;
the invention relates to a potato quality detection and estimation method based on machine vision, which comprises the following specific implementation steps of:
(1) A first Kinect camera collects a depth image of one side of the potato; the first Kinect camera and the second Kinect camera are provided with light source generators, and depth images of the potatoes in the conveying roller conveying channel in different body states are collected, as shown in fig. 10.10;
(2) The image is transmitted to a control end to be processed as follows:
(1) converting the depth image into a three-dimensional point cloud image; the principle is that the distance depth of each pixel point of the image is calculated by utilizing the parameter matrix and the coordinate system change of the camera, and the image is mapped to the point; the conversion of the world coordinate system to the pixel coordinate system is expressed by the following formula:
wherein z is c Is a scale factor, also known as effective focal length;
u and v are respectively the abscissa and the ordinate of the corresponding point in the pixel coordinate system;
u 0 ，v 0 referred to as the image center;
f x ，f y referred to as normalized focal length on the x, y axes, respectively;
r is a rotation matrix;
t is a translation matrix;
x w ，y w ，z w three coordinate values respectively of points in a world coordinate system;
if the origin of the world coordinate system is located at the center of the lens, i.e. coinciding with the camera coordinate system, and the image point does not rotate or translate, then,
and because of the coincidence, the length of the target from the origin under the two coordinate systems is consistent, namely,
according to the formula, a point (uv) on the image is obtained T And the conversion relation of corresponding points on the world coordinate system:
from this formula, a point cloud can be constructed, the result is shown in fig. 10.11.
Due to the distance characteristic of the Kinect camera to the depth map, the potato overlook point cloud result is shown in FIG. 10.11 (original point cloud set), and the image also includes point cloud display of objects around the potato in addition to the target potato. The point cloud removal of other objects can use a point cloud segmentation algorithm based on curvature constraint, a desired target is taken as a center, a proper X/Y/Z coordinate range is set, and a point cloud image only containing a potato target and a single background is obtained by segmentation as shown in FIG. 10.12;
most of the point clouds of the obstacle objects are removed from the image, and the background table point cloud set is removed from the image, so that the background point cloud set is a plane area. Removing the point cloud set below the background Zmax from the point cloud shown in FIG. 10.12 by taking the maximum height Z of the potato background as a segmentation condition to obtain an image shown in FIG. 10.13;
if the error of the upper half point cloud picture of the potato obtained by segmentation is smaller relative to the error of the potato, removing external sparse outliers by taking the average distance value of the outermost edge point of the point cloud set and the adjacent inner circle point of the point cloud set as a standard;
(1) let the cloud set of target points be P = { P i |p i ∈R 3 I =1,2,3 \8230;, n }, and any point p included is detected i Adjacent points of (a);
(2) calculate any point p i The distances to all the adjacent points are calculated, then the average operation is carried out, and the standard deviation and the variance of the included points are calculated according to the distance and the average distance; the characteristics of the distance distribution diagram accord with Gaussian distribution;
(3) the judgment condition of the outlier is that a range interval is set according to the variance and the average distance of the point to the point, and the points outside the range are judged as the outlier;
(4) removing the mark points;
carrying out triangular mesh division on the processed point cloud set of the upper half part of the potato; using Lawson algorithm based on optimization criterion to carry out point-by-point insertion to generate Delaunay triangulation network, and the steps are as follows:
(1) a plurality of adjacent triangles are randomly selected from the set to form a polygon;
(2) inserting a new point p at any position in a local triangle in the polygon, and connecting three vertexes of the triangle to generate three new triangles;
(3) checking three new triangles and adjacent triangles by using an empty circle condition, a minimum angle maximization condition and a LOP optimization criterion, and deleting a certain non-compliant side or exchanging diagonals of a quadrangle to enable the three new triangles and the adjacent triangles to meet the condition; the algorithm process is shown in a schematic diagram in FIG. 10.14;
the method comprises the steps of establishing a potato surface triangular net according to key feature points, establishing a small number of large triangular nets at flat positions of the potato surface, and enabling the triangular nets to be smaller in size and higher in density at important feature positions of the surface, so that distortion of important areas of the surface is avoided, data redundancy in flat areas is avoided, and the characteristics of uneven and fluctuant potato surface are reflected well; the results are shown in FIG. 10.15;
connecting the vertices of the triangular mesh on the surface of the potato with the point cloud internal center point (bottom plane position) of the upper half model of the potato to form a plurality of connected four-sided cones in space, wherein each side of each cone is a triangle, as shown in the schematic structural diagram of the potato cone model in fig. 10.16. For calculating the potato quality, the mass formula m = ρ V of the object can be used for calculation;
to calculate the potato volume, the volumes of the four sided pyramids that make up the three dimensional model are calculated, as shown in FIG. 10.17, and then summed; the volume calculation of any polyhedron in the three-dimensional space can be calculated by using a determinant and a vector theory method;
setting a certain four-side cone V in the potato cone model 1 Has four vertex coordinates of O (x) 0 ，y 0 ，z 0 )、A(x 1 ，y 1 ，z 1 )、B(x 2 ，y 2 ，z 2 )、C(x 3 ，y 3 ，z 3 ) Then, the volume of the tetrahedral cone and the total volume of the upper half part of the potato are respectively calculated by the following formulas:
V upper part of ＝V 1 +V 2 +…+V n (1.12)
(3) The potatoes are adjusted by the turnover mechanism to face upwards, are shot by the second Kinect camera and are processed as above to obtain the volume V of the other half part of the potatoes Lower part (ii) a The total mass is found to be approximately equal to:
M＝ρ(V on the upper part +V Lower part ) (I.13)
(4) The control end concludes which discharge chute each of the three rows of potatoes should be introduced into in which quality interval; when the infrared detection device of the distribution mechanism detects passing potatoes, the distribution device is started to distribute specific potato targets.
Claims (10)
1. The utility model provides a potato grading plant based on machine vision, includes feed mechanism, shape quality detection mechanism, divides the mechanism of allocating, its characterized in that: the feeding mechanism, the shape quality detection mechanism and the distribution mechanism are sequentially connected;
the feeding mechanism mainly comprises a feeding conveying belt (2);
the shape quality detection mechanism mainly comprises a transmission roller (35) and a secondary conveyer belt (17), wherein the transmission roller (35) and the secondary conveyer belt (17) are arranged up and down, and the transmission roller (35) is arranged above the secondary conveyer belt (17);
the conveying roller (35) is formed by installing a plurality of rollers (35-1) between a first roller positioning installation plate (22-a) and a second roller positioning installation plate (22-b), and the rollers (35-1) are connected with a conveying roller motor through belts;
a first Kinect camera (4) and a second Kinect camera (6) are mounted on the first roller positioning mounting plate (22-a); the shape quality detection mechanism is sequentially provided with a first Kinect camera (4) and a second Kinect camera (6) along the direction from a potato feeding conveyor belt (2) to a first-stage discharging conveyor belt (11) and a second-stage discharging conveyor belt (12), and a turnover claw mechanism (5) is arranged between the first Kinect camera (4) and the second Kinect camera (6);
the overturning claw mechanism (5) is provided with overturning claws (27), the overturning claws (27) are axially formed by 3 rows of claw-shaped structures which are closely arranged, and the included angle between every two rows of claws is 120 degrees;
the distribution mechanism consists of a conveying module and a discharging module, and the conveying module is provided with the discharging module; the conveying module is formed by connecting a primary discharging conveying belt (11) and a secondary discharging conveying belt (12); the discharging module consists of a cylinder (33) and a discharging groove; the discharge chute comprises a first discharge chute mechanism (19) and a second discharge chute mechanism (34);
first discharging conveyor belt baffles (10) are arranged on two opposite sides of the first-stage discharging conveyor belt (11), a first discharging groove mechanism (19) is arranged on each first discharging conveyor belt baffle (10), a cylinder module is correspondingly arranged on each first discharging groove mechanism (19), each cylinder module comprises a cylinder and a cylinder connecting rod, a distributor is arranged at the tail end of each rod, and the cylinder module contracts to drive the distributor to move towards the first discharging groove mechanism (19);
second discharging conveyor belt baffles (30) are arranged on two opposite sides of the second-stage discharging conveyor belt (12), a second discharging groove mechanism (34) is arranged on each second discharging conveyor belt baffle (30), a cylinder module is correspondingly arranged on each second discharging groove mechanism (34), each cylinder module comprises a cylinder and a cylinder connecting rod, a distributor is arranged at the tail end of each rod, and the cylinder module contracts to drive the distributor to move towards the second discharging groove mechanisms (34);
the feeding conveyer belt (2) is connected with a conveying roller (35), the conveying roller (35) is connected with a first-level discharging conveyer belt (11), and the first-level discharging conveyer belt (11) is connected with a second-level discharging conveyer belt (12).
2. The machine vision-based potato grading device of claim 1, characterized in that: baffle mechanisms are installed on two sides of the long edge of the feeding conveyor belt (2), and the feeding conveyor belt (2) is connected with the feeding groove (1).
3. The machine vision-based potato grading device of claim 1, characterized in that: the shape quality detection mechanism is an image detection component and further comprises an industrial camera height adjusting support; the industrial camera height adjusting bracket comprises a first Kinect camera bracket (24) and a second Kinect camera bracket;
a first Kinect camera bracket (24) and a first Kinect camera bracket base (23) are sequentially connected between the first Kinect camera (4) and the first roller positioning mounting plate (22-a), the first Kinect camera bracket (24) is of a telescopic rod structure, and the first Kinect camera bracket base (23) is fixedly connected with the first roller positioning mounting plate (22-a);
a second Kinect camera support and a second Kinect camera support base are sequentially connected between the second Kinect camera (6) and the first roller positioning mounting plate (22-a), the second Kinect camera support is of a telescopic rod structure, and the second Kinect camera support base is fixedly connected with the first roller positioning mounting plate (22-a).
4. The machine vision-based potato grading device of claim 1, characterized in that: the device also comprises an integral support, wherein the integral support comprises a feeding mechanism support (13), a shape quality detection mechanism support (14), a primary discharging conveying belt support (20) and a secondary discharging conveying belt support (21); the bottom of the feeding mechanism is provided with a feeding mechanism support (13), the bottom of the shape quality detection mechanism is provided with a shape quality detection mechanism support (14), and the bottom of the distribution mechanism is provided with a first-stage discharging conveyer belt support (20) and a second-stage discharging conveyer belt support (21).
5. The machine vision-based potato grading device of claim 1, characterized in that: the roller (35) is arranged between the first roller positioning mounting plate (22-a) and the second roller positioning mounting plate (22-b) through a bearing, and the positioning hinges (29) are arranged between the shape quality detection mechanism support (14) and the first roller positioning mounting plate (22-a) and between the shape quality detection mechanism support and the second roller positioning mounting plate (22-b).
6. The machine vision-based potato grading device of claim 1, wherein: the overturning claw (27) is arranged between the first roller positioning mounting plate (22-a) and the second roller positioning mounting plate (22-b);
one end of the turning claw (27) is arranged on the first roller positioning mounting plate (22-a) through a first turning claw bracket (26-a),
the other end of the overturning claw (27) is arranged on a second roller positioning mounting plate (22-b) through a second overturning claw bracket (26-b).
7. The machine vision-based potato grading device of claim 1, characterized in that: the distribution mechanism is provided with an infrared generator and a corresponding receiver for detecting the potatoes passing through the material guide chute.
8. The machine vision-based potato grading device of claim 1, wherein: the shape quality detection mechanism is divided into three paths by arranging a partition plate 1 (7), and the first-stage discharging conveying belt (11) is divided into three paths by arranging a partition plate 2 (9).
9. The machine vision-based potato grading device of claim 1, wherein: the two sides of the first-stage discharging conveyer belt (11) are provided with a plurality of first discharging groove mechanisms (19), and the two sides of the second-stage discharging conveyer belt (12) are provided with a plurality of second discharging groove mechanisms (34).
10. A potato shape and quality detection method based on machine vision is characterized in that: the method comprises the following implementation steps:
(1) a first Kinect camera (4) acquires overhead images of the lower three rows of potatoes in the state;
(2) the control end receives a first Kinect camera (4) to acquire a potato overlook image and processes information;
(3) the control end executes the preprocessing operation of the original potato picture and operates the watershed algorithm to mark different potatoes;
performing LOG operator to detect the edge of the potato, judging whether the edge of the potato is missing or not, performing image repairing operation if the edge of the potato is missing, and finally performing shape detection by combining the potato rectangular degree parameter, the minimum external rectangle length-width ratio and the circularity parameter;
combining the collected two-position potato image and the depth image to generate a point cloud image, and performing image segmentation and filtering processes to obtain a point cloud set of the upper half part of the potato; performing surface triangular mesh division on the point cloud set of the upper half part of the potato, combining an internal central point to form a tetrahedral pyramid model of the upper half part of the potato, and multiplying the tetrahedral pyramid model by the density of the potato according to a volume calculation method of a three-dimensional geometric body in a space to obtain the mass of half of the potato;
(4) the turning claw (27) turns the potatoes on the roller to change the upward body state of the potatoes;
(5) after the potatoes are adjusted to be in an upward state through the turnover mechanism (5), the second Kinect camera (6) collects three rows of potato overhead images below, and collects the same batch of potato images at different visual angles;
(6) the control end receives a overlook image of the lower part of the potato in a changed state acquired by the second Kinect camera (6), and information is processed; performing shape detection and quality estimation as in the third step;
(7) analyzing the image processing results of the first Kinect camera (4) and the second Kinect camera (6) to judge the final shape and the final quality of the potato;
(8) and the control end sends an execution command to the allocation mechanism for execution.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211004584.0A CN115365162A (en) | 2022-08-22 | 2022-08-22 | Potato grading device based on machine vision and shape detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211004584.0A CN115365162A (en) | 2022-08-22 | 2022-08-22 | Potato grading device based on machine vision and shape detection method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115365162A true CN115365162A (en) | 2022-11-22 |
Family
ID=84067544
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211004584.0A Pending CN115365162A (en) | 2022-08-22 | 2022-08-22 | Potato grading device based on machine vision and shape detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115365162A (en) |
Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1982755A (en) * | 2005-12-14 | 2007-06-20 | 现代自动车株式会社 | Manual valve of hydraulic control system for automatic transmission |
| CN101946966A (en) * | 2010-09-02 | 2011-01-19 | 石河子大学 | Automatic impurity removing and sorting technique for processing tomatoes and dedicated equipment |
| CN203877031U (en) * | 2014-04-23 | 2014-10-15 | 青岛莱茵智能装备有限公司 | Tableware sorting and packaging production line |
| CN204914572U (en) * | 2015-08-14 | 2015-12-30 | 东莞快裕达自动化设备有限公司 | Full -automatic upset gauze mask machine |
| CN106896111A (en) * | 2017-03-28 | 2017-06-27 | 华南农业大学 | A kind of potato external sort intelligent checking system based on machine vision |
| CN207275141U (en) * | 2017-09-22 | 2018-04-27 | 宁波中物光电杀菌技术有限公司 | Two-sided disinfection equipment |
| CN108313750A (en) * | 2018-02-06 | 2018-07-24 | 四川东林矿山运输机械有限公司 | Intelligent all standing bulk material discharging material distributing machine |
| CN110087577A (en) * | 2016-11-30 | 2019-08-02 | 锐珂牙科技术顶阔有限公司 | For the method and system from denture grid removal tooth set |
| CN111715553A (en) * | 2020-05-18 | 2020-09-29 | 镇江宇诚智能装备科技有限责任公司 | Automatic linen sorting system and control method thereof |
| CN112474379A (en) * | 2020-11-27 | 2021-03-12 | 广东力生智能有限公司 | Sorting trolley and continuous conveying equipment |
| CN112641158A (en) * | 2020-12-29 | 2021-04-13 | 安徽三众智能装备有限公司 | Quick turn-over device of gauze mask machine |
| CN215068333U (en) * | 2021-03-19 | 2021-12-07 | 广西机电职业技术学院 | E-commerce logistics storage management system |
| CN215314010U (en) * | 2021-06-23 | 2021-12-28 | 海邦(江苏)国际物流有限公司 | Intelligent storage automatic sorting equipment |
| CN215964874U (en) * | 2021-08-20 | 2022-03-08 | 东莞市大研自动化设备有限公司 | Tableware sorting machine based on manipulator |
| CN216225552U (en) * | 2021-11-17 | 2022-04-08 | 山东丸美佳食品有限公司 | Charcoal grilled fish bean curd automatic sorting equipment |
-
2022
- 2022-08-22 CN CN202211004584.0A patent/CN115365162A/en active Pending
Patent Citations (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1982755A (en) * | 2005-12-14 | 2007-06-20 | 现代自动车株式会社 | Manual valve of hydraulic control system for automatic transmission |
| CN101946966A (en) * | 2010-09-02 | 2011-01-19 | 石河子大学 | Automatic impurity removing and sorting technique for processing tomatoes and dedicated equipment |
| CN203877031U (en) * | 2014-04-23 | 2014-10-15 | 青岛莱茵智能装备有限公司 | Tableware sorting and packaging production line |
| CN204914572U (en) * | 2015-08-14 | 2015-12-30 | 东莞快裕达自动化设备有限公司 | Full -automatic upset gauze mask machine |
| CN110087577A (en) * | 2016-11-30 | 2019-08-02 | 锐珂牙科技术顶阔有限公司 | For the method and system from denture grid removal tooth set |
| CN106896111A (en) * | 2017-03-28 | 2017-06-27 | 华南农业大学 | A kind of potato external sort intelligent checking system based on machine vision |
| CN207275141U (en) * | 2017-09-22 | 2018-04-27 | 宁波中物光电杀菌技术有限公司 | Two-sided disinfection equipment |
| CN108313750A (en) * | 2018-02-06 | 2018-07-24 | 四川东林矿山运输机械有限公司 | Intelligent all standing bulk material discharging material distributing machine |
| CN111715553A (en) * | 2020-05-18 | 2020-09-29 | 镇江宇诚智能装备科技有限责任公司 | Automatic linen sorting system and control method thereof |
| CN112474379A (en) * | 2020-11-27 | 2021-03-12 | 广东力生智能有限公司 | Sorting trolley and continuous conveying equipment |
| CN112641158A (en) * | 2020-12-29 | 2021-04-13 | 安徽三众智能装备有限公司 | Quick turn-over device of gauze mask machine |
| CN215068333U (en) * | 2021-03-19 | 2021-12-07 | 广西机电职业技术学院 | E-commerce logistics storage management system |
| CN215314010U (en) * | 2021-06-23 | 2021-12-28 | 海邦(江苏)国际物流有限公司 | Intelligent storage automatic sorting equipment |
| CN215964874U (en) * | 2021-08-20 | 2022-03-08 | 东莞市大研自动化设备有限公司 | Tableware sorting machine based on manipulator |
| CN216225552U (en) * | 2021-11-17 | 2022-04-08 | 山东丸美佳食品有限公司 | Charcoal grilled fish bean curd automatic sorting equipment |
Non-Patent Citations (1)
| Title |
|---|
| 刘艳君: "基于机器视觉的马铃薯等级分类研究", 中国优秀硕士学位论文全文数据库农业科技辑, no. 1, 15 January 2022 (2022-01-15), pages 024 - 512 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111062915A (en) | Real-time steel pipe defect detection method based on improved YOLOv3 model | |
| WO2018232518A1 (en) | Determining positions and orientations of objects | |
| CN111814711A (en) | Image feature fast matching method and system applied to mine machine vision | |
| CN111982910B (en) | Weak supervision machine vision detection method and system based on artificial defect simulation | |
| CN107230203A (en) | Casting defect recognition methods based on human eye vision attention mechanism | |
| Lou et al. | Accurate multi-view stereo 3D reconstruction for cost-effective plant phenotyping | |
| CN113643280B (en) | Computer vision-based plate sorting system and method | |
| CN112233107B (en) | Sunflower seed grade classification method based on image processing technology | |
| CN113191174B (en) | Article positioning method and device, robot and computer readable storage medium | |
| CN115205319B (en) | Seed feature extraction and classification method used in seed selection process | |
| CN113902812A (en) | Laser radar and camera external parameter automatic calibration method based on multiple calibration plates | |
| CN116363505A (en) | Target picking method based on picking robot vision system | |
| Su et al. | Potato quality grading based on depth imaging and convolutional neural network | |
| CN111583193A (en) | Pistachio nut framework extraction device based on geometric contour template matching and algorithm thereof | |
| Dolata et al. | Instance segmentation of root crops and simulation-based learning to estimate their physical dimensions for on-line machine vision yield monitoring | |
| CN113313692B (en) | Automatic banana young plant identification and counting method based on aerial visible light image | |
| CN108108678A (en) | A kind of tungsten ore ore identifies method for separating | |
| CN110689059A (en) | Automatic garbage sorting method | |
| CN111291686A (en) | Method and system for extracting crop root phenotype parameters and judging root phenotype | |
| Zhang et al. | A novel image detection method for internal cracks in corn seeds in an industrial inspection line | |
| CN115365162A (en) | Potato grading device based on machine vision and shape detection method | |
| CN114820619B (en) | Tuber plant sorting method, system, computer device and storage medium | |
| CN114187269B (en) | Rapid detection method for surface defect edge of small component | |
| Imou et al. | Three-dimensional shape measurement of strawberries by volume intersection method | |
| CN115900586A (en) | Reclaimed sand particle morphology real-time monitoring device and real-time monitoring method |
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 |