JP2502981Y2 - Image processing system for crop row detection in rice transplanter - Google Patents

Description

[Detailed Description of the Invention] [Industrial field of application] The present invention is an automatic device that enables a rice transplanter to follow a so-called planting seedling row arranged in a row that has already been planted in a field. The present invention relates to the structure of a crop line detection image processing device for steering control.

[Conventional technology]

Conventionally, for example, when planting seedlings in a field with a rice transplanter,
The rice transplanter is provided with a planting mechanism at appropriate intervals on the left and right in the direction of travel, and the planting mechanism that rotates up and down as the rice transplanter progresses is used to plant the seedling mats on the seedling mounting table in the field scene while appropriately dividing the seedling mat by the number of plants. Therefore, in the field scene, at the appropriate seedling planting interval along the traveling direction of the rice transplanter, the planting seedling locations are lined up at the same time, it is possible to be planted in multiple rows at appropriate intervals in the left and right direction to the traveling direction. It is well known.

Then, as a prior art of the automatic steering device that allows the rice transplanter to run substantially parallel to the planting seedling row already planted in the field, in JP-A-62-61509, in the lateral projection from the body of the rice transplanter. A video camera (two-dimensional imaging means) is mounted on the tip of the arm, and the imaging direction of the video camera is forward in the traveling direction of the rice transplanter, and an appropriate range of the adjacent planting seedlings is imaged. Then, this image information is binarized to extract the area corresponding to each planted seedling portion, and then a straight line is approximately calculated from the column composed of the plurality of areas by processing such as Hough transform. It is proposed to execute the steering control of the aircraft so that the deviation between the lateral deviation and the inclination deviation with respect to an arbitrary reference line such as the vertical and horizontal centerlines of the imaging screen and the reference point within the calculation is within a certain allowable range. ing.

[Problems to be solved by the device]

By the way, in order to increase the yield of rice,
The number of seedlings to be planted per unit area (for example, 1 tsubo) of a field (so-called seedling planting density) is changed according to various conditions such as the type of seedlings and the planting area.

There are generally two types of changes in the planting density of this seedling. That is, the interval of planting seedlings along the traveling direction of the rice transplanter (hereinafter referred to as the plant interval) is changed, and the amount of seedlings at one planting seedling site, that is, from the seedling mat on the seedling table in the rice planting machine to the seedling planting mechanism. The amount of seedlings to be changed when splitting.

On the other hand, as described above, when the image of the field scene is picked up by the image pickup means and the image information thereof is image-processed by the image processing apparatus, the following difference is caused due to the fact that the planting density of the seedlings is sparse or dense. Inconvenience occurs.

That is, since the portion corresponding to the planted seedling portion has a different brightness and color from the mud surface of the field, on the imaging screen obtained by imaging these portions with the two-dimensional imaging means, they are arranged two-dimensionally. Quantize the difference in the color component and brightness of the image for each pixel, binarize the planting seedling part to "1" and the other parts to "0", and plant from the binarized image information. Image processing of extracting a portion corresponding to a seedling portion as a characteristic region is executed, and in this image processing, it is assumed that the image pickup screen is composed of, for example, 256 × 256 pixels in length and width. Hereinafter, if the vertical direction (Y axis) of the image pickup screen is taken in the advancing direction of the rice transplanter and the horizontal direction (X axis) is taken in the left and right direction of the rice transplanter, the pixel arrangement in the X axis direction is called the row, and The arrangement of pixels is called a column.

In this extracted image processing, scanning is performed in the right direction sequentially from the pixel portion in the uppermost row on the left end column of the imaging screen, and when it comes to the right end column on the imaging screen, the second row from the top and the right end portion When the location of the value of “1” is found in the binarization by repeating the scanning of “2”, the pixels at all the locations are scanned as notable observation points, and the seedlings are planted. If the density is high, for example if the stock spacing is short,
There are places with a value of "1" at short intervals in the Y-axis direction,
When the amount of seedlings is large, the area of the value of "1" is gathered and exists in a wide range. Therefore, in the former, the number of coordinates as sample points is used in the calculation of the Hough transform for specifying the virtual straight line. In the latter case, an additional calculation is required to determine the barycentric coordinates of the area of the value "1" over a wide range at one location. For example, the instruction amount for the final steering is calculated from the image processing. When the rice transplanter is moving forward at a high speed, the number of times the steering instruction amount per unit forward distance appears decreases, so-called hunting phenomenon easily occurs in the steering control, and the rice transplanter may meander. Many.

In order to eliminate this inconvenience, if the total number of pixels on the imaging screen is reduced, it is possible to find the number of planting seedlings necessary for calculating the virtual straight line when the above-mentioned plant interval is long (large). There was a risk of disappearing.

The present invention aims to solve these problems.

[Means for solving the problem]

Therefore, in order to achieve this object, the present invention relates to a crop row detection image processing device used for automatically steering control of a rice transplanter so as to follow the seedling row already planted in the field and beside the row. , An image processing device for extracting a characteristic region of a planting seedling part from an imaging screen after binarizing image information of a field scene obtained by imaging with a two-dimensional imaging means in the rice transplanter; A pre-planting density detection that detects the planting interval and the seedling take-up amount while changing the planting seedling stock interval along the traveling direction and providing seedling-planting density changing means for changing the seedling take-up amount of each planted seedling Means is provided to set the target pixels as observation points on the imaging screen at appropriate intervals during the extraction image processing of the characteristic region, and the number of target pixels is inversely proportional to the size of the seedling planting density. Set the calculation means to set Those were.

[Function of device]

As described above, in the configuration of the present invention, in the first image pickup screen imaged by the two-dimensional image pickup means, a wide range of the field is photographed and captured as image information, and the binarization process is executed. Regardless of whether it is sparse or dense, it is possible to obtain the number of samples required for the subsequent determination of the virtual straight line, and thus for the steering correction control.

On the other hand, means for changing the spacing between planted seedlings along the traveling direction in the rice transplanter, and one or both of the means for changing the seedling removal amount of each planted seedling, the means for changing seedling planting density,
The operator can arbitrarily set it depending on the field conditions and seedling type.

And, the detection of the density of the seedling planting density is to be detected by the seedling planting density detecting means associated with the seedling planting density changing means. The size of the planting density can be automatically detected, and the result is sent to a computing means in the image processing equipment, and the image of the planting seedling portion is imaged and binarized.
Pixels of interest serving as observation points on the imaging screen are set at appropriate intervals, and the number of pixels of interest is the size of the seedling planting density, that is, both the plant interval and the seedling picking amount, or either one of them. Set to be inversely proportional to the size of.

This inverse proportional setting is, for example, the total number of pixels of the imaging screen is 256
Of 256 pixels, the rows and columns are set at appropriate intervals (scatteringly) to set the pixels at the target pixels as the observation points, and attention is paid as the planting density increases. It may be inversely proportional to the linear isolinearity so that the number of pixels is small, and the number of pixels of interest is large in a place where the seedling planting density is small, and as the seedling planting density increases,
A stepwise inverse proportional relationship in which the number of pixels of interest is reduced stepwise may be used.

In this way, when the seedling planting density is low, increasing the number of pixels of interest will result in a feature of one of the pixels of interest even if the plant spacing is sparse, or the amount of seedling at one location is small and the characteristic region is narrow. The area becomes caught and can be detected.

On the other hand, when the planting density is high, the number of pixels of interest is reduced. It can be avoided.

[Effect of device]

In this way, during image processing, the number of pixels of interest can be automatically changed according to the planting conditions, and the change does not deteriorate the detection accuracy of planted seedlings when the planting density is small. However, even if the seedling planting density is high and the rice transplanter is operated at high speed, the distance between the rice transplanter and the lateral planting row can be calculated very easily. This has an effect that the steering control can be quickly executed without causing a malfunction.

〔Example〕

An embodiment applied to a riding type rice transplanter will be described below. In FIG. 1, reference numeral 1 denotes a traveling machine body supported by front wheels 3,3 on the front left and right sides and rear wheels 4,4 on the rear left and right sides. A multi-row planting type seedling planting device 7 composed of a float 7a, a seedling mount 5, and a plurality of planting mechanisms 6 is mounted on a rear portion of the machine body 1 so as to be vertically movable via a link mechanism 8.

The power of the engine 9 mounted on the upper surface of the traveling body 1 is transmitted via the clutch 10 and the transmission case 11 to the front and rear wheels 3,
4 while transmitting power to the seedling plant 7 via a PTO shaft 12 protruding from the transmission case 11.

Reference numeral 13 is an actuator for turning on / off the clutch 10, 14 is an actuator for traveling speed change, and 15 is an actuator for PTO axis speed change as one of the seedling planting density changing means. And let
The vertical rotation speed of the seedling planting mechanism 6 in the seedling planting device 7 and the horizontal lateral movement speed of the seedling mounting table associated therewith increase or decrease, and the combination of this with the forward speed of the traveling machine body 1 enables the seedling planting along the traveling direction. You can increase or decrease the interval, that is, the stock interval. Sign
Reference numeral 42 denotes a seedling planting density detecting means for detecting the size of the above-mentioned stock spacing, and by detecting the rotation speed of the PTO shaft 12, the size of the seedling planting density is detected by conversion from this.

The seedling planting density changing means 46 shown in FIG. 4 is provided with a seedling take-out plate 43 that moves up and down with respect to the lower end of the seedling placing table 5 and a lower end attached to the seedling taking plate 43 on the back surface of the seedling placing table 5. And a rotatable operation lever 45 connected to the upper end of the elevator shaft 44 and extending in the rear direction of the traveling machine body 1. The seedling take-out plate 43 that is raised and lowered is rotated vertically. A seedling take-out port (not shown) through which the planting claw 6a of the seedling planting mechanism 6 can pass is formed by notching, and the lower end of the seedling mat 47 on the seedling mounting table 5 is divided by the planting claw 6a. The vertical amount (l) of the seedlings determines the amount of seedlings, and thus the density of seedling planting.

Reference numeral 48 is a potentiometer for detecting the turning angle of the operating lever 45, and by detecting the vertical amount (l) by the seedling take-out plate 43, it is possible to detect the so-called seedling planting density detection. It becomes a means.

The steering mechanism 18 is pivotally operated via a steering handle 17 provided in front of the steering seat 16 in the traveling machine body 1 to change the direction of the front wheels 3, 3 to the left or right. An L-shaped steering arm 20 in a plan view, which is horizontally rotatably mounted on a rotation fulcrum shaft 19 of the steering mechanism 18, a pair of left and right tie rods 21 and 21 connected to the steering arm 20, a hydraulic cylinder 22, and a manual steering. And a steering gear box 24 for operating the steering control valve 23. The pitman arm 25 is capable of swinging back and forth.

The control valve 23 is connected to the steering arm 20 backward through a ball joint, and a spool at the rear end of the control valve 23 is connected to the pitman arm 25.

Further, a rear end of a hydraulic cylinder 22 having a front end rotatably connected to the traveling machine body 1 is rotatably connected to the steering arm 20, and swings corresponding to a turning angle of the steering handle 17. By the pitman arm 25, the spool of the control valve 23 is moved forward and backward, the hydraulic pressure from the hydraulic pump 26 driven by the engine 9 is sent, the piston rod in the hydraulic cylinder 22 is retracted, and the steering arm 20 is rotated. Change the direction of the left and right front wheels 3,3.

This hydraulic cylinder 22 is also driven by an electromagnetic solenoid type automatic steering control valve 27 operated by a central controller 31 for automatic steering / traveling, which outputs an output signal in response to a signal from a crop row detecting device 30 described later. At that time, the steering angle of the front wheels 3 can be known by detecting the turning angle of the steering arm 20 with the potentiometer 28 attached to the turning fulcrum shaft 23.

Incidentally, an actuator 13 for ON / OFF of the clutch 10,
Actuator 14 for traveling speed change, actuator 1 for PTO shaft speed change
5 also operates in the central control unit 31.

The crop row detection device 30 is an image pickup means 32 for picking up an image of an object,
An image processing device 33 for processing a captured image and exchanging necessary information (data) with a central control device 31.

The image pickup means 32 is a two-dimensional so-called area sensor in which the image pickup screen 40, such as a so-called video camera, has pixels S of 256 × 256 image pickup elements arranged in the xy plane, for example, when detecting an object. Yes, for example, two-dimensional MOS
There are image pickup devices and two-dimensional CCD image pickup devices.

According to this two-dimensional image pickup means 32, one image pickup screen
A plurality of planting seedling locations 41 can be simultaneously detected in 40 along with their geometrical relationship, and can be obtained as image information.

 Next, the control operation of this configuration will be described.

First, so that the center line Y along the traveling body traveling direction on the image pickup screen 40 of the image pickup means 32 corresponds to the reference line,
It is stored in the central control device 31 and the image processing device 33.

Next, the rice transplanter is put into an actual field, and the rice transplanter is made to run straight along the ridges and the like to create a plurality of rows of planting seedlings in advance.

At this time, the operator manually switches and changes both or any one of the seedling planting density changing means 15 and 46 according to the conditions of the field, the type of seedlings, and the like. The density of the seedling planting density (K) is detected by the seedling planting density detecting means 42 and 48 corresponding to, and the result thereof is stored in the computing means 49. K), the following number of pixels of interest (M × n) is calculated so as to be inversely proportional, and the result of this calculation is sent to the image processing device 33, and when the image processing device 33 performs image extraction processing of a characteristic region described later. The target pixel ST, which is an observation point on the imaging screen 40, can be set at appropriate intervals.

Next, as shown in FIG. 1, the traveling machine body 1 of the rice transplanter is advanced along the side of the existing multiple row planting seedlings, and while performing the rice planting work, the planting seedlings are imaged by the imaging means 32, Incorporated into the image processing device 33 as image information, a binarization process is performed to distinguish the planted crop portion 41 from other farm scenes,
Next, a calculation for determining the coordinates of the position of the binarized planting plant part 41 is executed, and a virtual line is created from the center or centroid position data of each of the plurality of specified characteristic regions as the planting seedling site. Assuming that the line is a straight line, the calculation for calculating the virtual straight line is performed by a virtual line calculating means (a huff (houg) which is built in the image processing device 33 or the central control device 31.
h) Includes a straight line function generation circuit for conversion and an arithmetic circuit).

At the time of these image processes, the calculation means 49 sets the target pixels ST as the observation points on the imaging screen 40 at appropriate intervals, and the number of the target pixels is inversely proportional to the size of the seedling planting density. To set.

FIG. 6 shows that, out of the 256 × 256 pixels S of the imaging screen 40, two pixels S are skipped in a row to set the pixel of interest ST,
An example in which five pixels S are skipped in the vertical row and the target pixel ST is set is shown.

The vertical interval along the Y axis is large because the degree of change in the plant interval is, for example, 10 cm to 20 cm, which is changed at 2 cm intervals. The vertical spacing of 41 is originally large, so it is necessary to skip unnecessary pixel parts that can be sure that the planting seedling part will not be imaged.On the other hand, if the row spacing is increased, one planting seedling part will be displayed. This is because it is not possible to determine the size of the seedling collection amount in 41.

FIG. 7 shows an example of inverse proportional calculation in the calculating means 49. In the first embodiment, as shown by the solid line in FIG. 6, the number of pixels of interest (M ×
N) is large and the seeding planting density (K) is increased by a predetermined step. A stepwise inverse proportional map is set in advance in the read-only memory (ROM) so that the number of pixels of interest (M × N) is gradually decreased. However, in that case, it may be inversely proportional to a linear function as shown by a chain line.

In this embodiment, the number of vertical pixels (N = 32) and the number of horizontal pixels (M = 84) are set in a portion where the number of target pixels (M × N) is large, and the number of vertical pixels (N = 32) and a horizontal number (M =) are set in a small portion. 42) Set to about.

In the second embodiment, using the inverse proportional calculation result,
Software for skipping a predetermined number of pixels may be executed during the noise removal processing of the image information after the binarization processing.

In the imaging screen 40 shown in FIG. 8, the stock interval is (1),
Characteristic area 50 corresponding to planted seedling site 41 with small seedling yield
It shows the case of observing at the target pixel ST (black coating part) for (white part). Since the seedling collection amount is small, it is captured by one target pixel ST, and that part is the position coordinates of the small feature region 50. Even if it is represented by, a large error does not occur.

FIG. 9 is an image capturing screen showing an image of the planting seedling site 41 having the same plant spacing as that of FIG. 8 but having a relatively large seedling collection amount. The characteristic region 50 (white portion) corresponding to one planting seedling site 41. )
However, it can be captured by 3 to 4 target pixels ST (black-painted portion). Then, the barycentric position coordinates of one feature region 50 are
It may be calculated as the barycentric position of the portion surrounded by the 3 to 4 target pixels ST.

Next, the tilt angle ρ of the virtual straight line with respect to the reference line Y of the imaging screen 40 and the deviation (δy) from the reference line Y on the X axis are calculated, and the tilt angle ρ and the deviation (δy) are calculated.
Is compared with a preset allowable value, and if it is within the allowable range, the process proceeds as it is, and if it is outside the allowable range, the inclination angle ρ or the deviation (δy) is set to the value. In response to this, an output signal is output to the central control device 31, the electromagnetic solenoid of the steering control valve 27 is operated, and the hydraulic cylinder 22 that changes the turning angle of the steering arm 20 in the steering mechanism is driven to perform the corrective steering operation. The deviation is set within the allowable range, and the automatic steering control is executed so that the reference line and the virtual straight line substantially coincide with each other.

In this case, the potentiometer 28 for detecting the turning angle of the steering arm 20 can detect whether or not the front wheels 3 have a steering angle that matches the inclination angle of the virtual straight line, and the result is automatically steered. Can be used for control feedback.

[Brief description of drawings]

The drawings show an embodiment of the present invention. Fig. 1 is a plan view of a riding type rice transplanter, Fig. 2 is a side view, and Fig. 3 is a block diagram of an automatic steering / traveling control device and an operation explanation including a hydraulic circuit. Fig. 4 is an enlarged side view of the main part of the seedling planting device, Fig. 5 is an explanatory diagram of the image capturing screen, Fig. 6 is an explanatory diagram of the target pixel in the image capturing screen, and Fig. 7 is a seedling planting on the horizontal axis. FIG. 8 is a diagram showing the relationship between the density and the number of pixels of interest on the vertical axis, FIG. 8 is an explanatory view showing the observation state of the characteristic region when the seedling planting density is small, and FIG. 9 is a medium planting planting density. It is explanatory drawing which shows the observation state of the characteristic area in a case. 1 …… Traveling machine, 3,4 …… Wheels, 5 …… Nursery stand, 6 ……
Planting mechanism, 7 ... Seedling planting device, 9 ... Engine, 11 ... Mission case, 12 ... PTO axis, 17 ... Steering handle, 1
9 …… Rotation fulcrum shaft, 20 …… Steering arm, 22 ……
Hydraulic cylinder, 27 ...... Automatic steering control valve, 28 ...... Potentiometer, 30 ...... Crop row detection device, 31 ...... Central control device, 32 ...... Imaging means, 33 ...... Image processing device, 40 ...... Imaging screen, 41 …… Planting seedling location, 15, 46 …… Seedling planting density changing means, 42, 48 …… Seedling planting density detecting means, 47 …… Seedling mat, 49 …… Computing means, 50 …… Characteristic area.

Claims (1)

(57) [Scope of utility model registration request] 1. A crop row detection image processing device used for automatically steering control of a rice transplanter so that the rice transplanter follows a seedling row already planted in a field so that the rice seedling row is two-dimensionally moved. An image processing device for extracting a characteristic region of a planting seedling portion from an imaged screen after binarizing image information of a field scene obtained by imaging by an image capturing means, and a planting seedling plant along a traveling direction of a rice transplanter. Providing seedling planting density changing means for changing the interval and changing the seedling taking amount of each planted seedling, and providing seedling planting density detecting means for detecting the plant interval and the seedling taking amount, respectively, the characteristic region At the time of the extraction image processing, there is provided an arithmetic means for setting the attention pixels as observation points on the imaging screen at appropriate intervals and setting the number of attention pixels so as to be inversely proportional to the size of the seedling planting density. Rice transplanter that is characterized by Kicking crop row detection image processing apparatus.