JP3680322B2 - Cucumber harvesting machine - Google Patents

Description

[0001]
[Industrial application fields]
The present invention relates to a cucumber harvester.
[0002]
[Background Art and Problems to be Solved by the Invention]
When using a conventional cucumber harvester to detect cucumbers and harvest them, the shape of the cucumber is photographed by the camera while self-propelled along the cultivated vegetation, and the detected image signal obtained by this photography Is binarized to create a binarized image, the ratio of the vertical and horizontal of the contour of the binarized image is calculated, and whether the calculated ratio is a cucumber or a leaf other than cucumber Judgment is made and only cucumbers are harvested.
[0003]
During the above cucumber harvesting work, when taking a cucumber with the camera, if the cucumber and the leaf are taken together, the binarized image becomes an image in which the cucumber and the leaf are integrated. The ratio of length to width has changed, and it was sometimes mistakenly detected that it was not a cucumber. However, this is solved and the cucumber is accurately detected to try to accurately harvest the cucumber.
[0004]
[Means for Solving the Problems]
According to the present invention, a color camera 2 that captures a predetermined range including cucumbers, a control device 20 that processes an image signal obtained by capturing with the color camera 2 and detects cucumbers to be harvested, and the control device 20 in cucumber harvester provided with a hand device 15 which sandwich the detected cucumber cut, the control device 20, the RGB signals of the color camera 2, the red and blue luminance difference image (a), green and red The brightness difference image (b) is obtained, and the brightness difference image (c) is further obtained from the brightness difference image (b) (b), and the brightness difference image (c) is binarized. A process for obtaining a binarized image by processing is performed, and the cucumber harvesting machine is configured to detect a cucumber to be harvested from the binarized image.
[0005]
Operation and effect of the invention
In the cucumber harvester, when cucumbers are detected and harvested, the shape of the cucumber is photographed by the color camera 2 while traveling along the cultivated vegetation, and the detected image signal obtained by this photographing is The visual means 6 extracts the G color (green) system portion from the R color (red) -B color (blue) from the R color (red), G color (green) and B color (blue) signals. Processing is performed to detect a luminance difference image (A), and processing for extracting a G color (green) -R color (red) and a dark G color (green) portion is performed. (B) is detected, and the process of removing the dark G color (green) portion from the G color (green) in the luminance difference image (b) -luminance difference image (b) is performed. Then, the luminance difference image (c) is binarized to create a binarized image, and from this binarized image, it is a cucumber or a leaf Luke is detected, is detected as a cucumber, only the cucumbers are harvested.
[0006]
As described above, the color camera 2 can use a general-purpose camera and is low in cost. Further, the cucumber detection is less affected by the brightness by using the relative value of the brightness, and for this reason, Can accurately detect cucumbers.
[0007]
【Example】
Hereinafter, an embodiment of the present invention will be described with reference to the drawings. The figure shows the fruit vegetable harvesting machine 1. For example, the shape of the cucumber is photographed by the color camera 2, and the detected image signal obtained by this photographing is R color (red), G by the visual means 6. A luminance difference image (A) of R color (red) -B color (blue) is detected from signals of color (green) and B color (blue), and G color (green) -R color (red). The luminance difference image (b) is detected, and the luminance difference image (c) is detected from the luminance difference image (b) -luminance difference image (b). The luminance difference image (c) is binarized, and the 2 A fruit and vegetable detection device that creates a binarized image and detects the cucumber 7 from the binarized image will be described.
[0008]
The fruit and vegetable harvesting machine 1 includes, on an electric traveling unit 3, a fruit cucumber 7 harvesting manipulator 4, a color camera 2 that captures the shape of the cucumber to be harvested, and a position sensor 5 that detects the position of the cucumber. The visual means 6 is provided. The cultivation field using the fruit and vegetable harvesting machine 1 grows by supporting the tree body 8 of the cucumber 7 on the support body 9 provided obliquely.
[0009]
The manipulator 4 includes an inclined frame 10 that is inclined so as to be substantially parallel to the inclination of the tree 8 that faces the base, a base 12 that is mounted so as to be movable up and down along a guide rail 11 of the inclined frame 10, and the base A hand 13 that sandwiches and cuts a main body 13 provided horizontally on a table 12, an arm 14 that forms a joint type provided on the manipulator 4, and a cucumber 7 provided at the tip of the arm 14. This is a configuration comprising the device 15.
[0010]
The inclined frame 10 is attached to the traveling unit 3 using a hinge 16 and supported on the back side by a support link 17. The lower end portion of the support link 17 is attached to an arbitrary position of the long hole 18 of the support plate 19, and the inclination angle of the inclined frame 10 can be arbitrarily adjusted by adjusting the attachment position. By operating each part of the manipulator 4, the cucumber 7 having the hand device 15 as a harvesting target is sandwiched and the handle portion of the cucumber 7 is cut.
[0011]
As shown in FIG. 2, the visual means 6 includes a color camera 2, a position sensor 5 that measures the distance to the cucumber 7 of the object by horizontally and vertically scanning the visual field of the color camera 2, and the color camera 2. The illuminance sensor 19 'detects the illuminance when an image taken with light of a predetermined wavelength is captured. The detection control device 20 is provided in the visual means 6 as shown in FIGS. 1 and 2, and various detections detected by the color camera 2, the position sensor 5 and the illuminance sensor 19 ′ are input to the CPU 22 via the input circuit 21, Analysis processing is performed by the CPU 22, and various commands are commanded through the output circuit 23 to operate the manipulator 4 and the like.
[0012]
The detection control device 20 shoots with the color camera 2 and the detected image signal obtained by this is selected from the R color (red), G color (green) and B color (blue) signals by the visual means 6. , R color (red)-B color (blue) and G color (green) system are extracted to detect a luminance difference image (A), and G color (green)-R color (red) The process of extracting the dark G color (green) is performed to detect the brightness difference image (b), and the brightness difference image (b) -brightness difference image (b) is dark from the G color (green). A process of removing the G color (green) portion is performed to detect a brightness difference image (c), the brightness difference image (c) is binarized, and a binarized image is created. Is input to the CPU 22 of the detection control device 20, and it is detected from the binarized image that it is a cucumber 7 or a leaf and a stem. Is detected by the hand device 15, and when it is detected by the detection that it is a leaf and a stem, it is clamped by the hand device 15. And it is set as the structure which is not cut | disconnected. 3 is an original image, and FIG. 4 is a binarized image obtained by binarizing (R color-B color)-(G color-R color).
[0013]
The detection control of the cucumber 7 by the detection control device 20 is configured as follows. The harvesting operation of the cucumber 7 is started (step 101), the color image taken by the color camera 2 is input (step 102), and the detected image signal R color (red) -B color (blue) is G color (green). ) System portion extraction processing is performed to detect a luminance difference image (A) (step 103), and G color (green) -R color (red) and dark G color (green) portions are extracted. The brightness difference image (b) is detected (step 104), and the process of removing the dark G color (green) portion from the G color (green) in the brightness difference image (b) -brightness difference image (b) The luminance difference image (c) is detected (step 105), the luminance difference image (c) is binarized, and a binarized image is created. From the binarized image, the cucumber 7 is used. Whether it is a leaf or stem (step 106) and RET There (step 107).
[0014]
The operation of the above embodiment will be described below. In the fruit vegetable harvesting machine 1, for example, the fruit vegetables are cucumbers 7. When the cucumber 7 is detected and harvested, the fruit harvesting machine 1 is allowed to self-run along the cultivated vegetation, The shape is photographed with light of a predetermined wavelength by the color camera 2 of the visual means 6.
[0015]
A luminance difference between R color (red) and B color (blue) from the R color (red), G color (green), and B color (blue) signals of the detected image signal obtained by this photographing. The image (A) is detected, and the luminance difference image (B) is detected for G color (green) -R color (Red). The luminance difference image (B) is detected by the luminance difference image (A) -luminance difference image (B). C) is detected and input to the CPU 22 of the detection control device 20 and binarized to create a binarized image. From this binarized image, whether it is a cucumber 7 or a leaf and a stem Is detected, and only the cucumber 7 detected as the cucumber 7 is clamped by the hand device 15 and cut and harvested.
[0016]
FIGS. 6-13 is a figure which shows another Example, is a structure which estimates this fruit length from the diameter of the fruit center part of the cucumber 7, and performs the temporary determination of the fruit of the cucumber 7 which can be harvested. 6 is a diagram showing the fruit, leaves, fruit handle, stem, and the like of the cucumber 7 and is configured to detect the diameter of the central part of the fruit part. FIG. 7 is a diagram showing the ratio of the number of places where the fruit portion of the cucumber 7 (91 pieces) cultivated at an inclined cultivation angle of 70 ° is hidden. FIG. 8 is a diagram showing a hidden state of the cucumber 7 in which only one fruit of the cucumber 7 is hidden. FIG. 9 shows the correlation coefficient among the cultivation method, the measurement direction, and the test result, the correlation with the fruit length of cucumber 7, the difference in correlation coefficient, the 95% confidence interval of the correlation coefficient in the center of the fruit, and the bending correction. It is a figure which shows the change of a correlation coefficient, etc. FIG. 10 is a diagram showing the estimation accuracy of the fruit length of the cucumber 7 by the regression among the test results, without bending correction, with bending correction, and the like. 11 and 12 are estimation diagrams of the 95% confidence interval of the fruit length of cucumber 7. FIG.
[0017]
The fruit length is estimated from the diameter of the fruit central part of the cucumber 7 by the detection control device 20, and the determination control as to whether or not the fruit of the cucumber 7 is harvestable is performed as follows. The cucumber 7 harvesting operation is started (step 201), an image taken by the color camera 2 is input (step 202), the cucumber 7 fruit is identified (step 203), and the number of fruits in the image is determined. Measured and detected n pieces (step 204), n pieces of the measured number are reset (step 205), n pieces are detected as n + 1 (step 206), and the fruit of cucumber 7 is completely exposed. If it is detected as NO (step 207), if it is detected as NO, it is detected whether it is the central part of the fruit or the fruit handle (step 208), and if YES (center part) is detected, the fruit length is estimated (step 209). ), The fruit position data of the cucumber 7 is saved (step 210), and it is detected whether all the cucumbers 7 have been measured (step 211). Back to flop 206. If YES is detected, RET is set (step 212).
[0018]
When YES is detected in step 207, the fruit length of the cucumber 7 is calculated (step 213), and the process proceeds to step 210. When NO is detected in step 208, the process proceeds to step 211. Due to the above configuration, the full exposure of the cucumber 7 fruit is about 60%. Therefore, in order to improve the yield of the cucumber 7 fruit, the estimation of the fruit length hidden in the obstacle is performed. As for the fruit diameter, there is a correlation between the fruit diameter and the fruit diameter in the central part and the fruit length, and among them, the correlation is stronger in the central part. The harvest rate can be improved by estimating the fruit length from the regression equation of the length and the fruit diameter (central part). The estimated accuracy of fruit length is about ± 16-18mm from the standard normal distribution, and about 70% can be harvested. This value is within the range of the cucumber 7 fruit selection standard, so there is no problem. Can harvest 7 cucumbers.
[0019]
14-17 is a figure which shows another Example, FIG. 14 is the whole fruit vegetable harvester 24 front view. FIG. 15 is an overall plan view of the fruit vegetable harvesting machine 24. FIG. 16 is a block diagram. FIG. 17 is a flowchart. The fruit and vegetable harvesting machine 24 is provided with front wheels 26 and 26 rotatably on the left and right sides of the front part below the vehicle body 25, and the rear wheels 27 and 27 are rotatable on the left and right sides of the rear part below the vehicle body 25. The left rear wheel 27 is rotationally driven by a drive motor (A) 28, and the right rear wheel 27 is rotationally driven by a drive motor (B) 29.
[0020]
In front of the induction sensor (left), (right) 30a, 30b are provided in the rear part of the front wheels 26, 26, and after the induction sensor (left), (right) 31a, 31b are in the rear part of the rear wheels 27, 27. Provided. The magnetic sensor 32 is provided at an intermediate portion behind the front wheels 26 and 26. The turn joint encoders 33 and 33 are provided inside the front wheels 26 and 26. Reference numeral 34 denotes a turn joint origin SW.
[0021]
The inclined plate 35 is provided on the upper side of the vehicle body 25, and the inclined plate 35 is configured to be inclined substantially parallel to the inclination of the tree body 8 of the cucumber 7 by the rotation of the turn joint motor 36. The CCD camera 37 is provided above the steering motor 38, and the steering motor 38 is configured to rotate and adjust the entire harvesting section such as the CCD camera 37, the manipulator 39, the arm 40, and the hand device 41.
[0022]
The CPU 42 receives the detection voltage values before (left), (right) 30a, 30b, after the induction sensor (left), (right) 31a, 31b, the detection value of the magnetic sensor 32, and the turn joint encoder. The detection values 33 and 33, the operation of the turn joint origin SW, and the like are input. Various motors 28, 29, 36, and 38 are driven to rotate in response to an output command from the CPU 42.
[0023]
The fruit and vegetable harvesting machine 24 detects a deviation (angle) from the center line of the ridge groove when traveling is stopped, and rotates the harvesting part having the CCD camera 37 above the vehicle body 25 in the horizontal direction by this deviation. The shelf (tree 8) to be harvested and the optical axis of the CCD camera 37 are perpendicular to each other. The control of the harvesting section by the CPU 42 is configured as follows.
[0024]
The harvesting operation of cucumber 7 is started (step 301), the travel of the fruit vegetable harvester 24 is started (step 302), the stop position is detected (step 303), and the moving mechanism (travel) is stopped (step 304). ), The voltage values of each induction sensor before (left), (right) 30a, 30b, after (left), (right) 31a, 31b are read (step 305), and the voltage value of the induction sensor before (left) 30a (GR1)-voltage value (gL1) 30b before the induction sensor (right)> correction presence / absence reference value (a) is detected (step 306), and if NO is detected, the induction sensor (left) 31a Voltage value (gR2) -after the induction sensor (right) 31b voltage value (gL2)> correction presence / absence reference value (a) is detected (step 307), and if YES is detected, the rotation angle of the turn joint (ΘT) is the following formula Calculated by T = tan-1k (step 308), whether the calculated rotation angle of the turn joint (θT) <0 is detected (step 309), and if YES is detected, the left rotation is corrected (step 310). Then, it is detected whether the waist joint angle = the rotation angle (θT) of the turn joint (step 311). If NO is detected, the process returns to step 310. When YES is detected, the harvesting operation is started (step 312).
[0025]
If YES is detected in step 306, the process proceeds to step 308. If NO is detected in step 307, the process proceeds to step 312. If NO is detected in step 309, the rotation is corrected to the right (step 313), and it is detected whether the waist joint angle = the rotation angle (θT) of the turn joint. (Step 314), if NO is detected, the process returns to step 313. If YES is detected, the process proceeds to step 312.
[0026]
As described above, the positional relationship between the harvest object and the CCD camera 37 can be made almost constant at all times, whereby the image input range can be specified and the overlap of the images can be reduced. The image processing range can be reduced and the processing speed can be improved. Also, the variation in the distance to the harvested object was reduced, and the extraction accuracy of the trace amount such as the size was not varied depending on the position of the fruit of the cucumber 7.
[Brief description of the drawings]
The figure shows an embodiment of the present invention.
1 is an overall front view of a fruit harvester in a field for cultivating fruit vegetables. FIG. 2 is a block diagram. FIG. 3 is an original image. FIG. 4 is (R color-B color)-(G color-R color). FIG. 5 is a flowchart showing another embodiment, and FIG. 6 is a name diagram of cucumbers. FIG. 7 is a diagram showing another embodiment, and is a ratio diagram of the number of places where cucumbers are hidden. FIG. 8 is a diagram showing another embodiment, and a cucumber hidden state diagram in which only one portion of the cucumber is hidden. FIG. 9 is a diagram showing another embodiment and a display diagram of a correlation coefficient of a test result. FIG. 10 is a diagram showing another embodiment, and is a diagram showing an estimation accuracy of the fruit length by regression of the test result. FIG. 11 is a diagram showing another embodiment, and a 95% confidence interval estimation diagram of the cucumber fruit length. [Fig. 13] Fig. 13 is a diagram showing another embodiment, and Fig. 13 is a diagram showing an estimation of the 95% confidence interval of the cucumber fruit length. [Fig. 13] FIG. 14 is a diagram showing another embodiment, and is an overall front view of the fruit and vegetable harvesting machine. FIG. 15 is a diagram showing another embodiment and is an overall plan view of the fruit and vegetable harvesting machine. FIG. Fig. 17 is a block diagram. Fig. 17 is a diagram showing another embodiment.
2 color camera 6 visual means 20 detection control device

Claims (1)

A color camera 2 for photographing a predetermined range including cucumbers, a control device 20 for processing image signals obtained by photographing with the color camera 2 to detect cucumbers to be harvested, and cucumbers detected by the control device 20 in cucumber harvester provided with a hand device 15 for clamping and cutting the said control device 20, the RGB signals of the color camera 2, the red and blue luminance difference image (a), green and red luminance difference image (B) is obtained, a luminance difference image (c) is further obtained from these luminance difference images (b) and (b), and the luminance difference image (c) is binarized to obtain a binary image. A cucumber harvesting machine characterized in that it performs processing for obtaining a binarized image and detects a cucumber to be harvested from the binarized image.

Images