CN110250146A - Fruit tree profiling sprayer and method based on laser detection and image processing technology - Google Patents

Abstract

The invention belongs to the field of orchard profiling spray, in particular to the provision of a fruit tree profiling sprayer and a method based on laser detection and image processing technology. The fruit tree profiling sprayer includes a vehicle chassis, a spraying device, an image acquisition device, a laser detection device and a data processing control system. The present invention selects different profiling spray modes according to the death and disease rates of fruit trees in the orchard. When the death or disease rate of fruit trees in the orchard is low, choose the real-time profiling spray mode; when the death or disease rate of the fruit trees in the orchard is high, choose the centering profiling spray mode. The invention has low imitation cost, realizes variable detection, and can improve the imitation accuracy.

Description

Fruit tree profiling sprayer and method based on laser detection and image processing technology

technical field

The invention belongs to the field of orchard profiling spray, and in particular relates to a fruit tree profiling sprayer and method based on laser detection and image processing technology.

Background technique

The idea of profiling spray is to adjust the application parameters in real time according to the canopy characteristics of the target crops. In order to obtain the target canopy characteristics, sensor-based canopy detection technology is widely used as an important means of canopy information acquisition. Among them, the laser sensor has the advantages of high precision and strong anti-interference ability. It measures the end face of the sensor through a non-contact measurement method. The distance to the surface of the fruit tree is used to detect the contour of the target.

"Research Status and Prospect Analysis of Orchard Variable Spray Technology" pointed out that in the existing technology, the use of multi-functional LIDAR sensors can accurately detect the structure of fruit trees, but the literature points out that LIDAR sensors are expensive, and the amount of data processing during use is relatively high. large, the total cost of use is higher. In addition, according to the flight time of the laser light from the transmitting point to the target reflecting back to the receiving point, the profiling of the fruit tree can also be realized by using the array ultrasonic sensor. However, the ultrasonic sensor measurement method requires that the sensors arranged in an array need to cover the entire object to be measured. This arrangement has the following problems in practical applications: first, the resolution and measurement accuracy of the ultrasonic sensor are low; second, the long strip arrangement During the measurement, the small vibration at the bottom is likely to cause a large measurement error of the top sensor, and the volume is bloated and difficult to install and transport; The degree varies. In the case of relatively sparse canopy, the sensor signal loss rate is large and the measurement accuracy is reduced. In the case of sparseness, the sensor signal loss rate is large, and the accuracy of canopy volume measurement is reduced. Fourth, there are often fruit trees with severe disease and death in the orchard. For this type of fruit trees that need to be replanted, it is not necessary to carry out spraying operations. Environmental pollution.

SUMMARY OF THE INVENTION

One object of the present invention is to provide a fruit tree profiling sprayer based on laser detection and image processing technology, which only uses two laser sensors to reduce the cost of profiling; the laser detection device is small in size and easy to transport; Layer feature acquisition realizes variable detection and improves profiling accuracy.

Another object of the present invention is to provide a fruit tree profiling spray method based on laser detection and image processing technology, which takes into account the characteristics of canopy sparseness and the phenomenon of fruit tree death and disease, and considers the canopy sparseness characteristics to return laser signals and Influence of application rate: According to the sparseness of fruit trees, adaptively adjust detection parameters (time interval of laser sensor movement, speed of laser sensor movement), target parameters (canopy volume, canopy height, air volume), thereby changing spray parameters and improving simulation results. shape application efficiency. In addition, the high disease rate of fruit trees in the orchard is also considered, and the spraying is decided according to the detected target parameters, so as to avoid the increase of the cost of application and the waste of the environment.

In order to achieve the above object, the present invention provides the following technical solutions:

A fruit tree profiling sprayer based on laser detection and image processing technology, comprising a vehicle chassis 28 and a spraying device, the spraying device includes a medicine box 30, a flow pump 17, a fan 18, a left spray head 25 and a right spray head 12 The liquid outlet of the medicine box 30 is sequentially connected to the flow pump 17 and the fan 18 through pipes, and the left and right air outlet pipes of the fan 18 are respectively connected to the left nozzle 25 and the right nozzle 12 through the left fluid pipe 16 and the right fluid pipe 13.

The vehicle body chassis 28 is provided with a speed sensor 29 for collecting the working speed of the implement;

The spraying device further comprises a swing device for moving the spray head up and down; the left spray head 25 and the right spray head 12 are respectively installed on the left and right sides of the rear of the vehicle chassis 28 through the swing device for moving the spray head up and down;

The fruit tree profiling sprayer is further provided with an image acquisition device, a laser detection device 6 and a data processing control system;

The image capture device includes a camera mount 4, a left camera 1 and a right camera 3, the camera mount 4 is vertically fixed to the front of the vehicle body chassis 28, and the left camera 1 and the right camera 3 are respectively mounted On the left and right sides of the camera mount 4;

The laser detection device 6 includes a laser detection mounting frame 7, a base 601, a motor mounting rod 602, a bracket 603, a laser wheel 606, a left laser sensor 614, a right laser sensor 605, a first rotating motor 607, a second rotating motor 611, the first link 613 and the second link 609;

The laser detection mounting bracket 7 is vertically fixed to the middle of the chassis 28 of the vehicle body, and the base 601 is fixed to the rear end surface of the laser detection mounting bracket 7;

The front ends of a pair of horizontal brackets 603 are vertically fixed to the upper and lower parts of the base 601 respectively, and the upper and lower ends of the laser wheel 606 parallel to the vertical plane are respectively fixed to the rear ends of the two brackets 603; the motor mounting rods 602 is fixed between the rear ends of the two brackets 603; the left laser sensor 614 and the right laser sensor 605 are slidably installed on the left and right semi-circles of the laser wheel 606, respectively, to the left and right The lateral fruit tree canopy is scanned;

The first rotary motor 607 and the second rotary motor 611 are fixed on the motor mounting rod 602 and the base 601 concentrically with the laser wheel 606 respectively; the power output shaft of the first rotary motor 607 passes through the first connecting rod 613 Connected with the left laser sensor 614, the power output shaft of the second rotating motor 611 is connected with the right laser sensor 605 through the second connecting rod 609;

A first angle sensor 612 is arranged between the first rotating electrical machine 607 and the first connecting rod 613 , and a second angle sensor 610 is arranged between the second rotating electrical machine 611 and the second connecting rod 609 ;

The data processing control system includes an image data processing control module 2, a laser data processing control module 5 and a spray control module 27;

The image data processing control module 2 is installed on the camera mounting frame 4, and is connected with the left camera 1, the right camera 3, the laser data processing control module 5 and the spray control module 27, and receives and processes the data collected by the left camera 1 and the right camera 3. Fruit tree canopy pictures, and send the processing results to the laser data processing control module 5 and the spray control module 27;

The laser data processing control module 5 is installed on the laser detection mounting frame 7, and is connected with the speed sensor 29, the first rotating motor 607, the second rotating motor 611, the left laser sensor 614, the right laser sensor 605, and the first angle sensor 612. , the second angle sensor 610 is connected to the spray control module 27; the laser data processing control module 5 processes and controls the operating speed of the implement returned by the speed sensor 29, the distance between the camera mounting frame 4 and the laser detection mounting frame 7 and the image data processing control The processing result sent by the module 2 controls the start-stop and movement speed of the first rotating motor 607 and the second rotating motor 611; receives and processes the data collected by the left laser sensor 614, the right laser sensor 605, the first angle sensor 612 and the second angle sensor 610. data, and send the processing results to the spray control module 27;

The spray control module 27 is arranged on the vehicle body chassis 28, and is connected with the speed sensor 29, the flow pump 17 of the spraying device, the fan 18, the left drive motor 21, the right drive motor 14, the left rotation motor 24 and the right rotation motor 11. Connection, according to the processing results sent by the laser data processing control module 5 and the operating speed of the implement returned by the speed sensor 29, the spraying demand and spraying delay time are calculated, and then the spraying device is adjusted according to the spraying demand and starts spraying.

The rear ends of the two brackets 603 are respectively provided with a first limiting piece 604 and a second limiting piece 608 to limit the displacement of the right laser sensor 605 and the left laser sensor 614 .

The distance between the laser detection device 6 and the ground is 30cm-80cm;

The distance between the camera of the image acquisition device and the laser sensor of the laser detection device 6 is 0.5m-1.5m;

The distance between the laser sensor of the laser detection device 6 and the spray head of the spraying device is 0.5m-1.5m.

The left and right air outlet pipes of the fan 18 are respectively provided with a left adjustment piece 20 and a right adjustment piece 19 for adjusting the air volume.

The spray head moves up and down swinging device including left drive motor 21, right drive motor 14, left lead screw 22, right lead screw 10, left bracket 15, right bracket 9, left moving slider 23, right moving slider 8, left rotation motor 24 and right rotation motor 11;

The left screw 22 and the left bracket 15 that are juxtaposed to each other are vertically installed on the rear left side of the chassis 28 of the vehicle body, and the right screw 10 and the right bracket 9 that are parallel to each other are vertically fixed to the chassis of the vehicle body. 28; the left moving slider 23 is sleeved on the left screw 22 and the left bracket 15, and the right moving slider 8 is sleeved on the right screw 10 and the right bracket 9; the left The drive motor 21 and the right drive motor 14 drive the left lead screw 22 and the right lead screw 10 to rotate, respectively, thereby making the left moving slider 23 and the right moving slider 8 perform up and down linear motions;

The left rotating motor 24 and the right rotating motor 11 are respectively fixed on the left moving slider 23 and the right moving slider 8; wherein the rotating shaft of the left rotating motor 24 is connected to the left spray head 25, and the rotating shaft of the right rotating motor 11 is connected to Right nozzle 12.

The left nozzle 25 and the right nozzle 12 are gas-liquid dual-flow nozzles, and the nozzle body is provided with an atomizing chamber; the left camera 1 and the right camera 3 are CCD cameras.

A fruit tree profiling method based on laser detection and image processing technology utilizing the described fruit tree profiling sprayer, the method comprises a real-time profiling spray mode, and the specific steps are as follows:

Step 1: Read the fruit tree position and row spacing data, and collect fruit tree images;

The data processing control system reads the data of fruit tree position and planting row spacing L _line from the fruit tree planting database, and the sprayer travels along the middle of the two rows of fruit trees. When the sprayer travels to the position corresponding to the position of the fruit trees, the image data processing control module 2. Control the left camera 1 and the right camera 3 to collect the fruit tree images on the left and right sides respectively;

Step 2: Calculate the porosity of the fruit tree canopy;

Perform image segmentation and morphological processing on the collected fruit tree images to obtain a binary image that only contains the fruit tree canopy, fill the binary image area, and then calculate the fruit tree canopy porosity K by the following formula. The layer porosity K is used as the standard to reflect the canopy sparsity of fruit trees;

In the formula, K is the porosity of the fruit tree canopy; R ₁ is the total number of pixels with a pixel value of 1 in the binary image; R ₂ is the total number of pixels with a pixel value of 1 in the binary image after region filling;

Step 3: Variable scanning to obtain the unit volume of the fruit tree canopy in real time;

The laser data processing control module 5 calculates the delay time detected by the laser sensor according to the operating speed S of the implement returned by the speed sensor 29 and the distance between the _sprayer and the camera and the laser sensor; after the delay time, the laser data processing control module 5 respectively The first rotary motor 607 and the second rotary motor 611 are controlled to be turned on at a certain speed, and the laser data processing control module 5 sets the left laser sensor 614 and the right laser sensor 605 according to the porosity K of the fruit tree canopy on the left and right sides respectively to scan the angular velocity S _laser The _wheel and different motion intervals detect the fruit trees on the left and right sides; the motion interval is between the completion of one semicircular motion on the laser wheel 606 by the left laser sensor 614 or the right laser sensor 605 and the start of the next semicircular motion The laser data processing control module 5 calculates the fruit tree canopy unit volume on the left and right sides by the following formula according to the data collected by the left laser sensor 614, the right laser sensor 605, the first angle sensor 612 and the second angle sensor 610:

_Li '=L _lines /2;

L ₀ =(L _i +R)cosσ;

L=L _i '-L ₀ ;

H _n =(L _i1 +L _in +2R)×sinσ;

V _n =H _n ×2L×S _sprayer ×T;

In the formula,

σ is the scanning angle in the circular motion of the laser sensor collected by the angle sensor;

R is the radius of the laser wheel;

Li is the length of the laser _beam when the laser reaches the surface of the canopy;

L ₀ is the horizontal projection connecting the center of the laser wheel and the length of the laser beam reaching the canopy surface;

L _behavior planting row spacing;

_Li ' is the distance from the center of the laser wheel to the center of the trunk;

L is the distance from the canopy surface to the center of the trunk;

L _i1 is the laser beam length at the highest point of the canopy surface detected by the laser;

L _in is the laser beam length at the lowest point of the canopy surface detected by the laser;

H _n is the height of the canopy unit detected in real time;

S _sprayer is the operating speed of the machine;

T is the time it takes for the laser sensor to turn half a circle;

V _n is the real-time detection of the fruit tree canopy unit volume;

Step 4: Obtain the required air volume, spray volume, spray height and swing angle range of the nozzle;

The spray control module 27 calculates the spray delay time according to the machine operating speed S returned by the speed sensor 29 and the distance between the _sprayer and the laser sensor and the nozzle; after the delay time, the spray control module 27 controls the applicator to start working; spray The control module 27 calculates the air volume required for spraying, the spray volume, and the spray height and swing angle range of the nozzle according to the volume of the fruit tree canopy on the left and right sides by the following formula:

H _n =(L _i1 +L _in +2R)×sinσ;

W _n =P1+V _n ;

Q= _Vn *q;

Swing angle range: [σ1, σ2];

In the formula,

σ is the scanning angle in the circular motion of the laser sensor collected by the angle sensor;

L _i1 is the laser beam length at the highest point of the canopy surface detected by the laser;

L _in is the laser beam length at the lowest point of the canopy surface detected by the laser;

H _n is the height of the canopy unit detected in real time;

R is the radius of the laser wheel;

L _i ' is the distance from the center of the laser wheel to the canopy surface of the fruit tree;

s _sprayer is the travel speed of the sprayer;

T is the time it takes for the laser sensor to turn half a circle;

P1 is the space volume between the sprayer nozzle and the canopy surface of the fruit tree;

Q is the required spray volume;

q is the dosage required per unit volume;

V _n is the real-time detection of the fruit tree canopy unit volume;

W _n is the required air volume corresponding to the detected canopy;

H is the spray height of the nozzle;

σ1 is the angle at which the output of the angle sensor hits the laser beam at the bottom of the canopy surface;

σ2 is the angle at which the output of the angle sensor hits the topmost laser beam on the canopy surface.

In the step 2, the image data processing control module 2 converts the void ratio calculation result into a digital signal through signal processing, and then performs canopy sparsity classification processing, and sends the processing result to the laser data processing control module 5; void ratio calculation result. The digital signal is divided into three canopy sparsity grades: relatively sparse K1, medium K2 and relatively dense K3, the larger the value of void ratio K, the sparser the canopy, K1>K2>K3;

In the step 3, three different circular motion interval times t1, t2, and t3 are set respectively corresponding to the three canopy sparsity levels of relatively sparse K1, medium K2 and relatively dense K3, t1>t2>t3, so as to adapt to different crowns. Volume detection under layer density.

A fruit tree profiling method based on laser detection and image processing technology utilizing the described fruit tree profiling sprayer, the method comprises a centering profiling spray mode, and the specific steps are as follows:

Step 1: Read the fruit tree position and row spacing data, and collect fruit tree images;

The data processing control system reads the data of fruit tree position and planting row spacing L _line from the fruit tree planting database, and the sprayer travels along the middle of the two rows of fruit trees. When the sprayer travels to the position corresponding to the position of the fruit trees, the image data processing control module 2. Control the left camera 1 and the right camera 3 to collect images of fruit trees on the left and right sides respectively;

Step 2: Adjust the position of the sprayer, collect the image of the fruit tree again, and calculate the porosity of the canopy of the fruit tree;

Perform image and morphological processing on the collected fruit tree images to obtain a binary image containing only the fruit tree canopy, mark the binary image and frame the maximum connected area of the canopy, and calculate the minimum circumscribed rectangle of the connected area. Center line, locate the center line of the canopy layer of the fruit tree, move the sprayer to make the camera of the image acquisition device correspond to the center line of the canopy layer of the fruit tree, and obtain the fruit tree image again;

Perform image segmentation and morphological processing on the re-collected fruit tree image to obtain a binary image that only contains the fruit tree canopy layer, fill the area of the binary image of the re-collected fruit tree image, and then calculate the fruit tree canopy porosity by the following formula: K, and the porosity K of the fruit tree canopy is taken as the standard reflecting the sparseness of the fruit tree canopy;

In the formula, K is the porosity of the fruit tree canopy; R ₁ is the total number of pixels with a pixel value of 1 in the binary image; R ₂ is the total number of pixels with a pixel value of 1 in the binary image after region filling;

Step 3: Variable scanning to obtain the canopy volume of the fruit tree;

The laser data processing control module 5 respectively controls the first rotary motor 607 and the second rotary motor 611 to turn on at a certain speed. The laser data processing control module 5 sets the left laser sensor 614 and the right laser respectively according to the porosity K of the fruit tree canopy on the left and right sides. The sensor 605 detects the fruit trees on the left and right sides with different scanning angular speeds S. The _{laser wheel} detects the fruit trees on the left and right sides; the laser data processing control module 5 collects data according to the left laser sensor 614, the right laser sensor 605, the first angle sensor 612 and the second angle sensor 610. , the canopy volume of the fruit tree on the left and right sides is calculated by the following formula:

y=f(x);

ph=(L _i +R)×sinσ;

In the formula,

x is the position information of the laser boundary point obtained by each scan;

y is the canopy profile curve fitted after each scan;

The length of the laser beam L _i reaches the canopy surface;

ph is the vertical projection with the position of the angle sensor as the starting point and the canopy surface hit by the laser beam as the end point;

R is the radius of the laser wheel;

σ is the scanning angle in the circular motion of the laser sensor collected by the angle sensor;

Vertical projection when H _max is the highest laser point;

H _min is the vertical projection at the lowest laser point;

V is the canopy volume of the fruit tree;

Step 4: Decision spray;

The laser data processing control module compares the fruit tree canopy volume detected on the left and right with the preset standard spray decision value. When the fruit tree canopy volume is less than the standard spray decision value, choose not to spray, and the sprayer starts to move forward. Until the position of the next fruit tree; when the canopy volume of the fruit tree is greater than or equal to the standard spray decision value, choose to spray and continue to the next step;

Step 5: Obtain the required air volume, spray volume, spray height and swing angle range of the nozzle;

The spray control module 27 calculates the spray delay time according to the machine operating speed S returned by the speed sensor 29 and the distance between the _sprayer and the laser sensor and the nozzle; after the delay time, the spray control module 27 controls the applicator to start working; spray The control module 27 calculates the air volume required for spraying, the spray volume, and the spray height and swing angle range of the nozzle according to the volume of the fruit tree canopy on the left and right sides by the following formula:

H=(L _i1 +L _in +2R)×sinσ;

L ₀ =(L _i +R)cosσ;

_Li '=L _lines /2;

L=L _i '-L ₀ ;

W=P2+V;

Q=V*q;

Swing angle range: [σ1, σ2];

In the formula,

σ is the real-time rotation angle of the laser sensor collected by the angle sensor;

R is the radius of the laser wheel;

L _i1 is the laser beam length at the highest point of the canopy surface detected by the laser;

L _in is the laser beam length at the lowest point of the canopy surface detected by the laser;

H is the crown height;

L _behavior planting row spacing;

Li is the length of the laser _beam when the laser reaches the surface of the canopy;

L ₀ is the horizontal projection connecting the center of the laser wheel and the length of the laser beam reaching the canopy surface;

_Li ' is the distance from the center of the laser wheel to the center of the trunk;

L is the distance from the canopy surface to the center of the trunk;

P2 is the space volume between the sprayer and the fruit tree;

V is the canopy volume of the fruit tree;

W is the required air volume;

Q is the required spray volume;

q is the dosage required per unit volume.

In the step 2, the image data processing control module 2 converts the void ratio calculation result into a digital signal through signal processing, and then performs canopy sparsity classification processing, and sends the processing result to the laser data processing control module 5; void ratio calculation result. The digital signal is divided into three canopy sparsity grades: relatively sparse K1, medium K2 and relatively dense K3, the larger the value of void ratio K, the sparser the canopy, K1>K2>K3;

In the step 3, three different laser sensor scanning angular velocities s1, s2, and s3 are set respectively corresponding to the three canopy sparsity levels of relatively sparse K1, medium K2 and relatively dense K3, s1<s2<s3, so as to adapt to different crowns. Volume detection under layer density.

Compared with the prior art, the beneficial effects of the present invention are:

1) the present invention considers the phenomenon of bad fruit trees such as disease and death in the orchard, for the orchard with less diseased, dead, bad fruit trees, adopt the real-time profiling spray pattern, for the orchard with higher disease death of fruit trees, replanting rate, Adopt the centering profiling spray mode that pre-judges whether or not to spray, which improves the applicability of the profiling sprayer;

2) The present invention greatly reduces the cost of the laser sensor in the profiling spray application, and only uses two laser sensor devices to profile the fruit trees on both sides of the sprayer at the same time, and the profiling surface basically covers the left and right sides facing the sprayer. area of fruit trees;

3) The method of using the laser sensor for semicircular motion scanning in the present invention solves the problem of the shape limitation of the causal tree, the need to adjust the installation position of the laser sensor, and improves the profiling efficiency;

4) While fitting the canopy volume by semi-circular scanning with laser sensor, the spray angle and spray height were obtained, and the accuracy of orchard profiling spray was improved by adjusting the spray parameters in real time;

5) The present invention considers canopy density characteristics, combined with image processing technology, considers the influence of target canopy density on spray and laser profiling, and performs variable detection and variable application according to canopy density, which makes profile spraying. more intelligent;

6) The present invention improves the accuracy of spraying by detecting the height of the canopy in real time and adjusting the swinging height and angle of the nozzle in real time.

Description of drawings

Fig. 1 is the structural representation of the fruit tree profiling sprayer based on laser detection and image processing technology of the present invention;

FIG. 2 is a schematic structural diagram of the laser detection device 6;

Figure 3a is a segmented canopy binary image;

Figure 3b is an image of the fully enclosed canopy area of a fruit tree after the area of Figure 3a is filled;

Fig. 4 is a schematic diagram of image processing of the center of the canopy in the centering anti-shape spray;

Figure 5a is a schematic diagram of a real-time profiling spray canopy volume calculation method;

Figure 5b is a schematic diagram of a method for calculating the volume of a centering profiling spray canopy;

Fig. 6a is the schematic diagram of real-time profiling spray method;

Figure 6b is a schematic diagram of a centering profiling spray method;

Figure 7a is a schematic diagram of the calculation of the equivalent volume of air volume during real-time profiling;

Figure 7b is a schematic diagram of the calculation of the equivalent volume of air volume during centering and profiling;

Fig. 8 is a schematic diagram of the calculation relationship of profiling spray;

Fig. 9 is the flow chart of real-time profiling spray method;

Figure 10 is a flow chart of the centering profiling spray method.

The reference numbers are:

1 Left camera 2 Image processing control module

3 Right camera 4 Camera mount

5 Laser data processing control module 6 Laser detection device

601 base 602 motor mounting rod

603 bracket 604 first limit piece

605 Right Laser Sensor 606 Laser Wheel

607 The first rotating motor 608 The second limit piece

609 Second Link 610 Second Angle Sensor

611 Second rotary motor 612 First angle sensor

613 First Link 614 Left Laser Sensor

7 Laser detection mounting bracket 8 Right moving slider

9 Right bracket 10 Right screw

11 Right rotation motor 12 Right nozzle

13 Right fluid line 14 Right drive motor

15 Left bracket 16 Left fluid line

17 flow pump 18 fan

19 Right adjustment tab 20 Left adjustment tab

21 Left drive motor 22 Left screw

23 Left moving slider 24 Left rotating motor

25 Left nozzle 26 Travel motor

27 Spray control module 28 Body chassis

29 Speed Sensor 30 Medicine Box

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

As shown in Figure 1, a fruit tree profiling sprayer based on laser detection and image processing technology includes a vehicle chassis 28, a spraying device, an image acquisition device, a laser detection device 6 and a data processing control system.

The vehicle body chassis 28 is a wheeled structure, and is driven by a traveling motor 26 arranged on the vehicle body chassis 28 ; the vehicle body chassis 28 is provided with a speed sensor 29 for collecting the operating speed of the implement.

The spraying device includes a medicine box 30 , a flow pump 17 , a fan 18 , a swing device for moving the spray head up and down, a left spray head 25 and a right spray head 12 . The medicine box 30 is installed on the chassis 28 of the vehicle body. The liquid outlet of the medicine box 30 is connected to the flow pump 17 and the fan 18 in turn through pipes. The left and right air outlet pipes of the fan 18 pass through the left fluid pipe 16 and the right fluid pipe respectively. 13 is connected to the left nozzle 25 and the right nozzle 12.

Preferably, the left and right air outlet pipes of the fan 18 are respectively provided with a left adjusting sheet 20 and a right adjusting sheet 19 for adjusting the air volume.

The left spray head 25 and the right spray head 12 are respectively installed on the left and right sides of the rear of the vehicle chassis 28 through the spray head moving up and down swinging device. The spray head moving up and down swinging device can adjust the spray head height and spray angle according to the canopy information of the fruit tree, so as to realize the fruit tree imitation. shaped spray.

The spray head moves up and down swinging device including left drive motor 21, right drive motor 14, left lead screw 22, right lead screw 10, left bracket 15, right bracket 9, left moving slider 23, right moving slider 8, left rotation Motor 24 and right rotation motor 11 .

The left screw 22 and the left bracket 15 that are juxtaposed to each other are vertically installed on the rear left side of the chassis 28 of the vehicle body, and the right screw 10 and the right bracket 9 that are parallel to each other are vertically fixed to the chassis of the vehicle body. 28; the left moving slider 23 is sleeved on the left screw 22 and the left bracket 15, and the right moving slider 8 is sleeved on the right screw 10 and the right bracket 9; the left The drive motor 21 and the right drive motor 14 drive the left lead screw 22 and the right lead screw 10 to rotate, respectively, so that the left moving slider 23 and the right moving slider 8 move linearly up and down.

The left rotating motor 24 and the right rotating motor 11 are respectively fixed on the left moving slider 23 and the right moving slider 8; wherein the rotating shaft of the left rotating motor 24 is connected to the left spray head 25, and the rotating shaft of the right rotating motor 11 is connected to Right nozzle 12.

The left nozzle 25 and the right nozzle 12 are gas-liquid dual-flow nozzles, and the nozzle body is provided with an atomizing chamber.

The image capture device includes a camera mount 4, a left camera 1 and a right camera 3, the camera mount 4 is vertically fixed to the front of the vehicle body chassis 28, and the left camera 1 and the right camera 3 are respectively mounted On the left and right sides of the camera mount 4.

The left camera 1 and the right camera 3 are ordinary CCD cameras.

The laser detection device 6 includes a laser detection mounting frame 7, a base 601, a motor mounting rod 602, a bracket 603, a laser wheel 606, a left laser sensor 614, a right laser sensor 605, a first rotating motor 607, a second rotating motor 611, The first link 613 and the second link 609 .

The laser detection mounting bracket 7 is vertically fixed to the middle of the chassis 28 of the vehicle body, and the base 601 is fixed to the rear end surface of the laser detection mounting bracket 7 .

As shown in FIG. 2 , the front ends of a pair of horizontal brackets 603 are vertically fixed to the upper and lower parts of the base 601 respectively, and the upper and lower ends of the laser wheel 606 parallel to the vertical plane are respectively fixed to the rear ends of the two brackets 603 . The motor mounting rod 602 is fixedly connected between the rear ends of the two brackets 603 . The left laser sensor 614 and the right laser sensor 605 are respectively slidably mounted on the left and right semicircles of the laser wheel 606 to scan the fruit tree canopy on the left and right sides respectively.

The first rotating motor 607 and the second rotating motor 611 are fixed on the motor mounting rod 602 and the base 601 concentrically with the laser wheel 606 , respectively. The power output shaft of the first rotating electrical machine 607 is connected to the left laser sensor 614 through the first link 613 , and the power output shaft of the second rotating electrical machine 611 is connected to the right laser sensor 605 through the second link 609 .

A first angle sensor 612 is arranged between the first rotating electrical machine 607 and the first link 613 , and a second angle sensor 610 is arranged between the second rotating electrical machine 611 and the second link 609 .

The rear ends of the two brackets 603 are respectively provided with a first limiting piece 604 and a second limiting piece 608 .

The first rotating motor 607 drives the left laser sensor 614 to make a semicircular motion along the left part of the laser wheel 606 through the first link 613 , and the second rotating motor 611 drives the right laser sensor 605 through the second link 609 to move along the right side of the laser wheel 606 . Do a semi-circular motion. The first limiting piece 604 and the second limiting piece 608 limit the displacement of the right laser sensor 605 and the left laser sensor 614 . The first angle sensor 612 and the second angle sensor 610 are used to detect real-time movement angles of the left laser sensor 614 and the right laser sensor 605, respectively.

The distance between the laser detection device 6 and the ground is 30cm˜80cm.

The data processing control system includes an image data processing control module 2 , a laser data processing control module 5 and a spray control module 27 .

The image data processing control module 2 is installed on the camera mounting frame 4, and is connected with the left camera 1, the right camera 3, the laser data processing control module 5 and the spray control module 27, and receives and processes the data collected by the left camera 1 and the right camera 3. and send the processing results to the laser data processing control module 5 and the spray control module 27.

The laser data processing control module 5 is installed on the laser detection mounting frame 7, and is connected with the speed sensor 29, the first rotating motor 607, the second rotating motor 611, the left laser sensor 614, the right laser sensor 605, and the first angle sensor 612. , the second angle sensor 610 is connected to the spray control module 27 . The laser data processing control module 5 controls the first rotating motor 607 according to the operating speed of the tool returned by the speed sensor 29, the distance between the camera mounting frame 4 and the laser detection mounting frame 7, and the processing result sent by the image data processing control module 2. and the start-stop and movement speed of the second rotating motor 611; receive and process the data collected by the left laser sensor 614, the right laser sensor 605, the first angle sensor 612 and the second angle sensor 610, and send the processing results to the spray control module 27.

The spray control module 27 is arranged on the vehicle body chassis 28, and is connected with the speed sensor 29, the flow pump 17 of the spraying device, the fan 18, the left drive motor 21, the right drive motor 14, the left rotation motor 24 and the right rotation motor 11. Connection, according to the processing results sent by the laser data processing control module 5 and the operating speed of the implement returned by the speed sensor 29, the spraying demand and spraying delay time are calculated, and then the spraying device is adjusted according to the spraying demand and starts spraying.

Preferably, the distance between the camera of the image acquisition device and the laser sensor of the laser detection device 6 is 0.5m˜1.5m.

Preferably, the distance between the laser sensor of the laser detection device 6 and the spray head of the spraying device is 0.5m˜1.5m.

The invention provides a fruit tree profiling method based on laser detection and image processing technology, and different profiling spray modes are selected according to the death and disease rates of fruit trees in the orchard. When the death and disease rates of fruit trees in the orchard are low, select the real-time profiling spray mode; when the death and disease rates of the orchard fruit trees are high, select the centering profiling spray mode.

The real-time profiling spray mode obtains the real-time canopy volume detected by the laser during the traveling process of the sprayer, and performs real-time profiling variable spraying.

First, read the position and row spacing data of the fruit tree, collect the fruit tree image through the image acquisition device, calculate the porosity of the fruit tree canopy, and use the porosity to characterize the sparseness of the fruit tree canopy; secondly, according to the sparseness of the fruit tree canopy, change the laser sensor Then, according to the laser scanning results, the canopy volume of the fruit tree is calculated; finally, according to the canopy volume of the fruit tree, the air volume required for spraying, the spray volume, the spray height and the swing angle range of the nozzle are obtained, and the spraying is carried out. .

As shown in Figure 9, the concrete steps of the real-time profiling spray pattern are as follows:

Step 1: Read the fruit tree position and row spacing data, and collect fruit tree images.

The data processing control system reads the data of fruit tree position and planting row spacing L _line from the fruit tree planting database, and the sprayer travels along the middle of the two rows of fruit trees. When the sprayer travels to the position corresponding to the position of the fruit trees, the image data processing control module 2. Control the left camera 1 and the right camera 3 to capture the fruit tree images on the left and right sides respectively.

Step 2: Calculate the porosity of the fruit tree canopy.

Using MATALB 2018a to perform image segmentation and morphological processing on the collected fruit tree images, a binary image containing only the fruit tree canopy was obtained, and the binary image was filled with regions. The porosity K of the fruit tree canopy is taken as the standard reflecting the sparseness of the fruit tree canopy;

In the formula, K is the porosity of the fruit tree canopy; R ₁ is the total number of pixels with a pixel value of 1 in the binary image; R ₂ is the total number of pixels with a pixel value of 1 in the binary image after region filling;

The step 2 specifically includes the following steps:

1) Use the Retinex image equalization algorithm to perform image segmentation preprocessing on the collected fruit tree images. The Retinex algorithm compresses the gray levels of the R, G and B components of the original image, and the compressed gray level dynamic range is changed from the original 0. -255 is compressed to 50-250, and the overexposed area is compressed from the original 100-255 to 200-255, so as to obtain an image with uniform contrast, so as to improve the definition of the main body of the image and increase the detail quality of the image.

2) The three-channel R, G, and B of the fruit tree image after equalization of the illumination model are used as input features, the fuzzy parameter is selected as 2, the number of clusters is 2, and the FCM clustering algorithm is performed on the cluster center with the largest 2G-R-B value. The canopy image is subjected to image segmentation and morphological processing to obtain a binary image containing only the canopy, as shown in Figure 3a, as the canopy segmentation result map, the background pixel value in the canopy segmentation result image is 0, and the canopy The area pixel value is 1;

3) The total number of pixels with a pixel value of 1 in the statistical canopy segmentation result graph is recorded as R ₁ , which is the canopy area of the fruit tree with gaps. The canopy segmentation result map is then filled with regions to obtain a complete canopy area without voids, as shown in Figure 3b, which is defined as a fully enclosed fruit tree canopy area with 0 voids, and the pixels in the binary image are counted. The total number of pixels with a value of 1 is denoted R ₂ . Then calculate the canopy porosity K of the fruit tree by the following formula:

4) The image data processing control module 2 converts the void ratio calculation result into a digital signal through signal processing, and then performs canopy sparsity classification processing, and sends the processing result to the laser data processing control module 5 . In this embodiment, the digital signal of the void fraction calculation result is divided into three canopy sparsity levels: relatively sparse K1, medium K2 and relatively dense K3. The larger the void fraction K value is, the more sparse the canopy is. K3, the specific value depends on the growth period of the fruit tree.

Step 3: Variable scanning to obtain the unit volume of fruit tree canopy in real time.

The laser data processing control module 5 calculates the delay time detected by the laser sensor according to the operating speed S of the implement returned by the speed sensor 29 and the distance between the _sprayer and the camera and the laser sensor; after the delay time, the laser data processing control module 5 respectively The first rotary motor 607 and the second rotary motor 611 are controlled to be turned on at a certain speed, and the laser data processing control module 5 sets the left laser sensor 614 and the right laser sensor 605 according to the porosity K of the fruit tree canopy on the left and right sides respectively to scan the angular velocity S _laser The fruit trees on the left and right sides are detected by the _wheel and different movement intervals. The movement interval is the interval between the completion of one semicircular movement on the laser wheel 606 by the left laser sensor 614 or the right laser sensor 605 and the start of the next semicircular movement. The angular velocity S of the _{laser wheel} is between 18°/s and 42°/s. According to the data collected by the left laser sensor 614, the right laser sensor 605, the first angle sensor 612 and the second angle sensor 610, the laser data processing control module 5 calculates the unit volume of the fruit tree canopy on the left and right sides by the following formula (part of the calculation parameters are shown) As shown in Figure 8):

_Li '=L _lines /2;

L ₀ =(L _i +R)cosσ;

L=L _i '-L ₀ ;

H _n =(L _i1 +L _in +2R)×sinσ;

V _n =H _n ×2L×S _sprayer ×T;

In the formula,

σ is the scanning angle in the circular motion of the laser sensor collected by the angle sensor;

R is the radius of the laser wheel 606;

Li is the length of the laser _beam when the laser reaches the surface of the canopy;

L ₀ is the horizontal projection connecting the center of the laser wheel and the length of the laser beam reaching the canopy surface;

L _behavior planting row spacing;

_Li ' is the distance from the center of the laser wheel to the center of the trunk;

L is the distance from the canopy surface to the center of the trunk;

L _i1 is the laser beam length at the highest point of the canopy surface detected by the laser;

L _in is the laser beam length at the lowest point of the canopy surface detected by the laser;

H _n is the height of the canopy unit detected in real time;

S _sprayer is the operating speed of the machine;

T is the time it takes for the laser sensor to turn half a circle;

V _n is the real-time detection of the fruit tree canopy unit volume.

In this embodiment, three different circular motion intervals t1, t2, and t3 are respectively set for the three canopy sparsity levels of relatively sparse K1, medium K2, and relatively dense K3, so as to adapt to volume detection under different canopy density levels. Under a certain sprayer movement speed and fruit tree width, the longer the scanning interval is, the fewer times the laser sensor detects. In the real-time profiling spray mode, since the fruit trees with dense canopy require more pesticides, set the is t1>t2>t3. For fruit trees with different canopy sparsity levels on both sides during the traveling process of the sprayer, the laser sensor makes semicircular motions at different time intervals to realize variable detection.

According to the laser information returned by the laser sensor, the volume of the canopy can be obtained to achieve the purpose of profiling spray. When the sprayer moves forward and the laser sensor semicircular scanning is carried out at the same time, the curve of the laser point hitting the canopy surface The shape is a sinusoidal distribution of 1/4 period. In order to reduce the error, the relationship between the scanning angular speed S of the laser sensor and the operating speed S of the _sprayer is: S _{laser wheel >>} S _sprayer , at this time 1 The sinusoidal shape of /4 period is close to a vertical line, as shown in Fig. 5a.

Denote the canopy volume calculated when the laser sensor scans one semicircle as one unit, as shown in Figure 5a. The canopy unit calculated when the laser sensor moves for half a circle is equivalent to a cuboid. As shown in the figure, the length of the cuboid is 2L, and L is the distance from the canopy surface to the center of the trunk; the height of the cuboid is the crown height H; the width is The distance traveled by the sprayer within the time T when the laser sensor completes one half-cycle movement, S _sprayer × T. As the sprayer travels, the canopy volume after each scan is obtained.

As shown in Figure 6a, it is a schematic diagram of the real-time profiling spray method under different canopy densities. The graph reflects the shorter time interval between laser sensor motions as the canopy density increases. Because the higher the canopy density is, the larger the required spraying area is, so the total spraying amount of the fruit tree with the canopy porosity K3 is greater than the canopy spraying amount when the canopy porosity is K1. Laser detection is carried out according to different canopy densities, and spraying is implemented to realize the process of variable detection and variable spraying.

Step 4: Obtain the required air volume, spray volume, spray height and swing angle range of the nozzle;

The spray control module 27 calculates the spray delay time according to the operating speed S of the implement returned by the speed sensor 29 and the distance between the _sprayer and the laser sensor and the spray head; after the delay time, the spray control module 27 controls the spraying device to start working. The spray control module 27 calculates the air volume required for spraying, the spray volume, and the spray height and swing angle range of the nozzle according to the volume of the fruit tree canopy on the left and right sides by the following formula:

H _n =(L _i1 +L _in +2R)×sinσ;

W _n =P1+V _n ;

Q= _Vn *q;

Swing angle range: [σ1, σ2];

In the formula,

σ is the scanning angle in the circular motion of the laser sensor collected by the angle sensor;

L _i1 is the laser beam length at the highest point of the canopy surface detected by the laser;

L _in is the laser beam length at the lowest point of the canopy surface detected by the laser;

H _n is the height of the canopy unit detected in real time;

R is the radius of the laser wheel;

L _i ' is the distance from the center of the laser wheel to the canopy surface of the fruit tree;

s _sprayer is the travel speed of the sprayer;

T is the time it takes for the laser sensor to turn half a circle;

P1 is the space volume between the sprayer nozzle and the canopy surface of the fruit tree;

Q is the required spray volume;

q is the required amount of spraying per unit volume, and the specific value depends on the spraying object;

V _n is the real-time detection of the fruit tree canopy unit volume;

W _n is the required air volume corresponding to the detected canopy;

H is the spray height of the nozzle;

σ1 is the angle at which the output of the angle sensor hits the laser beam at the bottom of the canopy surface;

σ2 is the angle at which the output of the angle sensor hits the topmost laser beam on the canopy surface.

The required air volume is also different under different canopy volumes. On the premise of the replacement principle, the air volume required by the fruit tree is calculated as the sum of the fruit tree canopy volume and the space volume P1 from the front of the fan to the fruit tree, where the fruit tree canopy volume is calculated in step 3. Obtained, the space volume between the sprayer nozzle and the surface of the canopy of the fruit tree is equivalent to the volume of a quadrangular _pyramid , and its equivalent relationship is shown in Figure 7a. distance, the bottom surface of the quadrangular pyramid is equivalent to a rectangle, the length H _n of the rectangle is the height of the canopy unit detected in real time, and the width is the distance traveled by the spray in the time T when the laser sensor completes a semicircular motion, S _sprayer × T.

According to the installation distance of the laser detection device and the speed of the sprayer returned by the sprayer speed sensor 29, the moving distance of the sprayer is calculated to adjust the sprayer to move so that the front of the nozzle is in the center of the canopy of the fruit tree. Calculate the forward time of the sprayer through the canopy height information H, and control the slider to adjust the height of the bracket, so that the nozzle can spray at the height h. At the same time, the angle sensor outputs the angle σ1 of the laser beam hitting the bottom of the canopy surface and the angle σ2 of the topmost laser beam, and controls the nozzle to swing the spray between the angles [σ1, σ2].

In the centering and profiling spray mode, first, the position and row spacing data of the fruit tree are read, the image of the fruit tree is obtained through the image acquisition device, the center line of the canopy layer of the fruit tree is positioned, and the sprayer is moved to make the camera of the image acquisition device and the center of the canopy layer of the fruit tree Corresponding to the line, obtain the fruit tree image again, calculate the porosity of the fruit tree canopy, and use the porosity to characterize the sparseness of the fruit tree canopy; secondly, according to the sparseness of the fruit tree canopy, change the scanning speed of the laser sensor for variable detection; again , According to the laser scanning results, use the captured laser points to fit the canopy contour and calculate the canopy volume of the fruit tree; then decide whether to carry out the spraying operation according to the canopy volume of the fruit tree; The air volume, spray volume, spray height and swing angle range of the nozzle are required to spray. If it is judged that it is not spraying, the sprayer will continue to move to the position of the fruit tree at the next moment.

As shown in Figure 10, the centering profiling spray mode specifically includes the following steps:

Step 1: Read the fruit tree position and row spacing data, and collect fruit tree images.

The data processing control system reads the data of fruit tree position and planting row spacing L _line from the fruit tree planting database, and the sprayer travels along the middle of the two rows of fruit trees. When the sprayer travels to the position corresponding to the position of the fruit trees, the image data processing control module 2. Control the left camera 1 and the right camera 3 to collect images of fruit trees on the left and right sides respectively.

Step 2: Adjust the position of the sprayer, collect the image of the fruit tree again, and calculate the canopy porosity of the fruit tree.

Using MATALB 2018a to perform image and morphological processing on the collected fruit tree images, a binary image containing only the fruit tree canopy is obtained, the binary image is marked and the maximum connected area of the canopy is selected, and the minimum value of the connected area is calculated. Circumscribe the center line of the rectangle, locate the center line of the canopy layer of the fruit tree, move the sprayer so that the camera of the image acquisition device corresponds to the center line of the canopy layer of the fruit tree, and obtain the fruit tree image again;

Using MATALB 2018a to perform image segmentation and morphological processing on the re-collected fruit tree images, a binary image containing only the fruit tree canopy is obtained, and the region is filled with the binary image of the re-collected fruit tree images, and then the fruit tree canopy is calculated by the following formula. layer porosity K, and the canopy porosity K of fruit trees is used as the standard to reflect the canopy sparsity of fruit trees;

In the formula, K is the porosity of the fruit tree canopy; R ₁ is the total number of pixels with a pixel value of 1 in the binary image; R ₂ is the total number of pixels with a pixel value of 1 in the binary image after region filling;

The step 2 specifically includes the following steps:

1) Use the Retinex image equalization algorithm to perform image segmentation preprocessing on the collected fruit tree images. The Retinex algorithm compresses the gray levels of the R, G and B components of the original image, and the compressed gray level dynamic range is changed from the original 0. -255 is compressed to 50-250, and the overexposed area is compressed from the original 100-255 to 200-255, so as to obtain an image with uniform contrast, so as to improve the definition of the main body of the image and increase the detail quality of the image.

2) The three-channel R, G, and B of the fruit tree image after equalization of the illumination model are used as input features, the fuzzy parameter is selected as 2, the number of clusters is 2, and the FCM clustering algorithm is performed on the cluster center with the largest 2G-R-B value. The canopy image is subjected to image segmentation and morphological processing to obtain a binary image containing only the canopy, as shown in Figure 3a, as the canopy segmentation result map, the background pixel value in the canopy segmentation result image is 0, and the canopy The area pixel value is 1;

3) Use the bwconncomp function to mark the canopy segmentation result image, and use the BoundingBox attribute string box in the regionprops function to select the maximum connected area of the canopy for the marked image, and calculate the center line of the minimum circumscribed rectangle of the connected area. It is the centerline of the fruit tree canopy. As shown in Figure 4, the box in the figure is represented as a circumscribed rectangle containing the maximum connected area of the entire canopy, SL ₁ represents the centerline of the fruit tree canopy, and SL ₂ represents the centerline of the entire image. Since the centerline of the fruit tree canopy may be offset from the fruit tree trunk (the position of the fruit tree read by the sprayer), SL ₁ and SL ₂ do not coincide, and the sprayer position needs to be adjusted. Calculate the abscissa pixel position R ₃ of SL ₁ and the abscissa pixel position R ₄ of SL ₂ , read the fruit tree position Q1 at this moment, and calculate the actual position Q2 of the center of the fruit tree canopy as: The image processing control platform returns Q2 to the spray control module 27, and the spray control module 27 outputs a signal to adjust the position of the sprayer, so that the camera of the image acquisition device corresponds to the center line of the canopy of the fruit tree.

4) Obtain the fruit tree image again through the image acquisition device, repeat step 1) and step 2), and the total number of pixels with a pixel value of 1 in the statistical canopy segmentation result graph is recorded as R ₁ , which is the canopy layer of the fruit tree with voids. area. The canopy segmentation result map is then filled with regions to obtain a complete canopy area without voids, as shown in Figure 3b, which is defined as a fully enclosed fruit tree canopy area with 0 voids, and the pixels in the binary image are counted. The total number of pixels with a value of 1 is denoted R ₂ . Then calculate the canopy porosity K of the fruit tree by the following formula:

5) The image data processing control module 2 converts the void ratio calculation result into a digital signal through signal processing, and then performs canopy sparsity classification processing, and sends the processing result to the laser data processing control module 5 . In this embodiment, the digital signal of the void fraction calculation result is divided into three canopy sparsity levels: relatively sparse K1, medium K2 and relatively dense K3. The larger the void fraction K value is, the more sparse the canopy is. K3, the specific value depends on the growth period of the fruit tree.

Step 3: Variable scanning to obtain the canopy volume of fruit trees.

The laser data processing control module 5 respectively controls the first rotating motor 607 and the second rotating motor 611 to turn on at a certain speed, and the laser data processing control module 5 respectively sets the left laser sensor 614 and the right laser according to the porosity K of the fruit tree canopy on the left and right sides. The sensor 605 detects the fruit trees on the left and right sides with different scanning angular speeds S and the _{laser wheel} . The laser data processing control module 5 calculates the canopy volume of the fruit tree on the left and right sides by the following formula according to the data collected by the left laser sensor 614, the right laser sensor 605, the first angle sensor 612 and the second angle sensor 610:

y=f(x);

ph=(L _i +R)×sinσ;

In the formula,

x is the position information of the laser boundary point obtained by each scan;

y is the fitted canopy profile curve after each scan, as shown in Figure 5b;

The length of the laser beam L _i reaches the canopy surface;

ph is the vertical projection with the position of the angle sensor as the starting point and the canopy surface hit by the laser beam as the end point;

R is the radius of the laser wheel;

σ is the scanning angle in the circular motion of the laser sensor collected by the angle sensor;

Vertical projection when H _max is the highest laser point;

H _min is the vertical projection at the lowest laser point;

V is the canopy volume of fruit trees.

Corresponding to the three canopy sparsity levels of relatively sparse K1, medium K2 and relatively dense K3, three different laser sensor scanning angular velocities s1, s2, s3 are set respectively, s1<s2<s3, to adapt to the volume under different canopy density levels probe. As shown in Figure 6b, it is a schematic diagram of the centering profiling spray method under different void ratios. The larger the void fraction, the slower the movement speed of the laser sensor, in order to reduce the problem of laser loss when the canopy is sparse. Therefore, the loss rate of the laser signal is larger. The fruit trees of K2 and K3, corresponding to the laser sensors, move at the speed of s1, s2, and s3 respectively, where s1<s2<s3.

Calculate the canopy profile curve. The laser data processing control module accepts the information of the canopy boundary points collected by the laser sensor, uses the interpolation algorithm to perform curve fitting, programs the algorithm through the QT compiler, and solidifies the program into the microcontroller through the serial port, and the crown can be directly obtained from the data measured by the laser sensor. Layer profile fit curve.

Step 4: Decision Spray

The laser data processing control module compares the fruit tree canopy volume detected on the left and right with the preset standard spray decision value. When the fruit tree canopy volume is less than the standard spray decision value, choose not to spray, and the sprayer starts to move forward. Until the position of the next fruit tree; when the canopy volume of the fruit tree is greater than or equal to the standard spray decision value, choose to spray and continue to the next step;

The selection method of the standard spraying decision value of this embodiment is as follows: the first five fruit trees in the spraying process of the preset sprayer are normal fruit trees, and the volumes v ₁ , v ₂ , v ₃ , v ₄ , and v ₅ of the five fruit trees are stored. , find the standard deviation of the five fruit trees This value is stored as the volume size of normal fruit trees, and compared with the volume of the canopy to be sprayed by the sprayer. When _V≥Vs , the spraying starts.

Step 5: Obtain the air volume required for spraying, the spray volume, and the spray height and swing angle range of the nozzle.

The spray control module 27 calculates the spray delay time according to the operating speed S of the implement returned by the speed sensor 29 and the distance between the _sprayer and the laser sensor and the spray head; after the delay time, the spray control module 27 controls the spraying device to start working. The spray control module 27 calculates the required air volume for spraying (as shown in Figure 7b), the spray volume and the spray height and swing angle range of the nozzle according to the volume of the fruit tree canopy on the left and right sides (part of the calculation parameters are shown in Figure 8). Show):

H=(L _i1 +L _in +2R)×sinσ;

L ₀ =(L _i +R)cosσ;

_Li '=L _lines /2;

L=L _i '-L ₀ ;

W=P2+V;

Q=V*q;

Swing angle range: [σ1, σ2];

In the formula,

σ is the real-time rotation angle of the laser sensor collected by the angle sensor;

R is the radius of the laser wheel;

L _i1 is the laser beam length at the highest point of the canopy surface detected by the laser;

L _in is the laser beam length at the lowest point of the canopy surface detected by the laser;

H is the crown height;

L _behavior planting row spacing;

Li is the length of the laser _beam when the laser reaches the surface of the canopy;

L ₀ is the horizontal projection connecting the center of the laser wheel and the length of the laser beam reaching the canopy surface;

_Li ' is the distance from the center of the laser wheel to the center of the trunk;

L is the distance from the canopy surface to the center of the trunk;

P2 is the space volume between the sprayer and the fruit tree;

V is the canopy volume of the fruit tree;

W is the required air volume;

Q is the required spray volume;

q is the required amount of spray per unit volume, and the specific value depends on the spray object.

Through the canopy height information H, control the slider to adjust the height of the bracket, so that the nozzle sprays at the height h, and the angle sensor outputs the angle σ1 of the laser beam at the bottom of the canopy surface and the angle σ2 of the topmost laser beam to control the nozzle. Swing the spray between [σ1, σ2] angles.

Claims (10)

1. a kind of fruiter profile modeling spray machine based on laser acquisition and image processing techniques, including chassis of vehicle body (28) and application dress It sets, the device for administration of drugs includes medicine-chest (30), flow pump (17), blower (18), left spray head (25) and right spray head (12), medicine-chest (30) liquid outlet is sequentially connected flow pump (17) and blower (18), the discharge pipe difference of left and right two of blower (18) by pipeline It is connect by left fluid line (16) and right fluid line (13) with left spray head (25) and right spray head (12), it is characterised in that:

The chassis of vehicle body (28) is equipped with the velocity sensor (29) for acquiring equipment operating speed；

The device for administration of drugs further comprises that spray head moves up and down pendulous device；The left spray head (25) and right spray head (12) are respectively Pendulous device is moved up and down by spray head to be mounted at left and right sides of the rear portion of chassis of vehicle body (28)；

The further image collecting device of the fruiter profile modeling spray machine, Laser Detecting Set (6) and data processing control system；

Described image acquisition device includes camera mounting bracket (4), Zuo Xiangji (1) and right camera (3), the camera mounting bracket (4) It is vertically fixed in the front of chassis of vehicle body (28), the left camera (1) and right camera (3) are separately mounted to camera mounting bracket (4) the left and right sides；

The Laser Detecting Set (6) includes laser acquisition mounting rack (7), pedestal (601), motor mounting rod (602), bracket (603), laser wheel (606), left laser sensor (614), right laser sensor (605), the first rotating electric machine (607), second Rotating electric machine (611), first connecting rod (613) and second connecting rod (609)；

The laser acquisition mounting rack (7) is vertically fixed in the middle part of chassis of vehicle body (28), and the pedestal (601) is fixed in sharp On the rear end face of optical detection mounting rack (7)；

The front end of a pair of horizontal bracket (603) is respectively perpendicular the upper and lower part for being fixed in pedestal (601), flat with perpendicular The upper and lower ends of capable laser wheel (606) are affixed with the rear end of two brackets (603) respectively；The motor mounting rod (602) is solid It connects between the rear end of two brackets (603)；The left laser sensor (614) and right laser sensor (605) can be slided respectively It is mounted in the left part semi-circumference and right part semi-circumference of laser wheel (606), the fruit tree canopy of the left and right sides is carried out dynamicly respectively Scanning；

It is respectively and fixedly connected to electricity to first rotating electric machine (607) and the second rotating electric machine (611) and laser wheel (606) concentric On machine mounting rod (602) and pedestal (601)；The power output shaft of first rotating electric machine (607) passes through first connecting rod (613) Connect with left laser sensor (614), the power output shaft of second rotating electric machine (611) by second connecting rod (609) with Right laser sensor (605) connection；

It is provided with first angle sensor (612) between first rotating electric machine (607) and first connecting rod (613), the second rotation Second angle sensor (610) are provided between rotating motor (611) and second connecting rod (609)；

The data processing and control system include image real time transfer control module (2), laser data processing and control module (5) and Spraying control module (27)；

Described image data processing and control module (2) is mounted on camera mounting bracket (4), and with Zuo Xiangji (1), right camera (3), Laser data processing and control module (5) and spraying control module (27) connection, receive the left camera (1) of processing and right camera (3) are adopted The fruit tree canopy picture of collection, and processing result is sent to laser data processing and control module (5) and is sprayed control module (27)；

The laser data processing and control module (5) is mounted on laser acquisition mounting rack (7), and with velocity sensor (29), First rotating electric machine (607), the second rotating electric machine (611), left laser sensor (614), right laser sensor (605), first Angular transducer (612), second angle sensor (610) and spraying control module (27) connection；Laser data processing control mould Equipment operating speed that block (5) is returned according to velocity sensor (29), camera mounting bracket (4) and laser acquisition mounting rack (7) it Between spacing distance and image real time transfer control module (2) send processing result, control the first rotating electric machine (607) and The start and stop and movement velocity of second rotating electric machine (611)；It receives and handles left laser sensor (614), right laser sensor (605), the data of first angle sensor (612) and second angle sensor (610) acquisition, and processing result is sent to spray Mist control module (27)；

Described spraying control module (27) are arranged on chassis of vehicle body (28), and the stream with velocity sensor (29), device for administration of drugs Amount pump (17), blower (18), left driving motor (21), right driving motor (14), left-handed rotating motor (24) and dextrorotation rotating motor (11) it connects, the machine of processing result and velocity sensor (29) passback sent according to laser data processing and control module (5) Have operating speed, calculates spraying demand and spraying delay time, and then device for administration of drugs is adjusted according to spraying demand and is started spraying.

2. fruiter profile modeling spray machine according to claim 1, it is characterised in that:

It is respectively arranged with the first limit film (604) and the second limit film (608) on the rear end of described two brackets (603), limits The displacement of right laser sensor (605) and left laser sensor (614).

3. fruiter profile modeling spray machine according to claim 1, it is characterised in that:

The distance between the Laser Detecting Set (6) and ground are 30cm~80cm；

Spacing distance between the laser sensor of the camera and Laser Detecting Set (6) of described image acquisition device be 0.5m~ 1.5m；

Spacing distance between the laser sensor of the Laser Detecting Set (6) and the spray head of device for administration of drugs be 0.5m~ 1.5m。

4. fruiter profile modeling spray machine according to claim 1, it is characterised in that:

The inside of two discharge pipes in left and right of the blower (18) is respectively provided with the left adjustment sheet (20) for adjusting air quantity and the right side Adjustment sheet (19).

5. fruiter profile modeling spray machine according to claim 1, it is characterised in that:

It includes left driving motor (21), right driving motor (14), left lead screw (22), right silk that the spray head, which moves up and down pendulous device, Thick stick (10), right support (9), moves to left movable slider (23), moves to right movable slider (8), left-handed rotating motor (24) and dextrorotation at left support (15) Rotating motor (11)；

The left lead screw (22) being mutually juxtaposed and left support (15) are vertically mounted to the rear left of chassis of vehicle body (28), phase Mutually the right lead screw (10) and right support (9) arranged side by side are vertically fixed in the rear right of chassis of vehicle body (28)；It is described to move to left Movable slider (23) is socketed on left lead screw (22) and left support (15), it is described move to right movable slider (8) be socketed in right lead screw (10) and On right support (9)；The left driving motor (21) and right driving motor (14) respectively drive left lead screw (22) and right lead screw (10) Rotation does linear movement up and down so that moving to left movable slider (23) and moving to right movable slider (8)；

The left-handed rotating motor (24) and dextrorotation rotating motor (11) are respectively and fixedly connected to move to left movable slider (23) and move to right movable slider (8) On；Wherein, the left spray head of rotation axis connection (25) of left-handed rotating motor (24), the right spray head of rotation axis connection of dextrorotation rotating motor (11) (12)。

6. fruiter profile modeling spray machine according to claim 1, it is characterised in that:

The left spray head (25) and right spray head (12) select gas-liquid double fluid spray head, are provided with spray chamber on sprayer body；The left phase Machine (1) and right camera (3) are CCD camera.

7. it is a kind of using fruiter profile modeling spray machine described in any one of claims 1-6 based on laser acquisition and image procossing skill The fruit tree profiling method of art, it is characterised in that: this method includes a kind of real-time profile modeling spray mode, the specific steps are as follows:

Step 1: reading fruit tree position and line-spacing data, acquire fruit tree image；

Data processing and control system reads fruit tree position and Planting Row Distance L from planting fruit trees database_RowData, spraying machine is along two Arrange the traveling operation of fruit tree middle position, when spraying machine is marched to fruit tree position corresponding, image real time transfer control Molding block (2) controls left camera (1) and right camera (3) acquires the fruit tree image of the left and right sides respectively；

Step 2: calculating fruit tree canopy voidage；

Image segmentation and Morphological scale-space are carried out to fruit tree image collected, only included the binary map of fruit tree canopy Picture carries out area filling to bianry image, then calculates fruit tree canopy voidage K by following formula, and fruit tree canopy is empty Standard of the gap rate K as reflection fruit tree canopy degree of rarefication；

In formula, K is fruit tree canopy voidage；R₁The total pixel number for being 1 for the pixel value in bianry image；R₂After area filling Bianry image in pixel value be 1 total pixel number；

Step 3: variable scanning obtains fruit tree canopy unit volume in real time；

The equipment operating speed S that laser data processing and control module (5) is returned according to velocity sensor (29)_{Spraying machine}With camera and swash Spacing distance between optical sensor calculates the delay time of laser sensor detection；After delay time, at laser data Reason control module (5) controls the first rotating electric machine (607) and the second rotating electric machine (611) respectively with the unlatching of certain revolving speed, laser Left laser sensor (614) and the right side is respectively set according to the fruit tree canopy voidage K of the left and right sides in data processing and control module (5) Laser sensor (605) is with angular scanning speed S_{Laser wheel}The fruit tree of the left and right sides is detected with the different movement space time； The movement space time is that left laser sensor (614) or right laser sensor (605) are completed once on laser wheel (606) Half circular motion and start the interval time between half circular motion next time；Laser data processing and control module (5) is according to a left side Laser sensor (614), right laser sensor (605), first angle sensor (612) and second angle sensor (610) are adopted The data of collection calculate the fruit tree canopy unit volume of the left and right sides by following formula:

L_i'=L_Row/2；

L₀=(L_i+R)cosσ；

L=L_i’-L₀；

H_n=(L_i1+L_in+2R)×sinσ；

V_n=H_n×2L×S_{Spraying machine}×T；

In formula,

σ is the scanning angle in the laser sensor circular motion of angular transducer acquisition；

R is laser wheel radius；

L_iLaser beam length when Malabar Pied Hornbill is reached for laser；

L₀For the floor projection at connection laser wheel center and the laser beam length for reaching Malabar Pied Hornbill；

L_RowFor Planting Row Distance；

L_i' it is distance of the laser wheel center away from trunk center；

L is distance of the Malabar Pied Hornbill to trunk center；

L_i1For the laser beam length for the Malabar Pied Hornbill highest point that laser acquisition is arrived；

L_inFor the laser beam length for the Malabar Pied Hornbill lowest part that laser acquisition is arrived；

H_nFor the canopy cell height of real-time detection；

S_{Spraying machine}For equipment operating speed；

T is that laser sensor turns over duration used in half cycle；

V_nFor the fruit tree canopy unit volume of real-time detection；

Step 4: obtaining the spray height and swing angle range of spraying institute's required airflow, spray amount and spray head；

The equipment operating speed S that spraying control module (27) are returned according to velocity sensor (29)_{Spraying machine}With laser sensor and spray Spacing distance between head calculates spraying delay time；After delay time, spraying control module (27) control device for administration of drugs It starts to work；Spraying control module (27) are calculated spraying required according to the fruit tree canopy volume of the left and right sides by following formula Air quantity, the spray height of spray amount and spray head and swing angle range:

H_n=(L_i1+L_in+2R)×sinσ；

W_n=P1+V_n；

Q=V_n*q；

Swing angle range: [σ 1, σ 2]；

In formula,

σ is the scanning angle in the laser sensor circular motion of angular transducer acquisition；

L_i1For the laser beam length for the Malabar Pied Hornbill highest point that laser acquisition is arrived；

L_inFor the laser beam length for the Malabar Pied Hornbill lowest part that laser acquisition is arrived；

H_nFor the canopy cell height of real-time detection；

R is laser wheel radius；

L_i' be laser wheel center to fruit tree fruit tree Malabar Pied Hornbill distance；

s_{Spraying machine}For spraying machine travel speed；

T is that laser sensor turns over duration used in half cycle；

P1 is spatial volume of the sprayer nozzle apart from fruit tree Malabar Pied Hornbill；

Q is required spray amount；

Q is formulation rate needed for unit volume；

V_nFor the fruit tree canopy unit volume of real-time detection；

W_nRequired air quantity is corresponded to for the canopy of detection；

H is the spray height of spray head；

σ 1 is the angle that Malabar Pied Hornbill bottommost laser beam is got in angular transducer output；

σ 2 is the angle that Malabar Pied Hornbill top laser beam is got in angular transducer output.

8. according to the method described in claim 7, it is characterized by:

In the step 2, voidage calculated result is changed into digital letter by signal processing by image real time transfer control module (2) Canopy degree of rarefication classification processing is carried out after number, and processing result is sent to laser data processing and control module (5)；Voidage meter The digital signal for calculating result is divided into more sparse K1, medium K2 and tri- canopy degree of rarefication grades of denser K3, gap The bigger expression canopy of rate K value is more sparse, K1 > K2 > K3；

In the step 3, tri- corresponding more sparse K1, medium K2 and denser K3 canopy degree of rarefication grades are respectively set three A different circular motion interval time t1, t2, t3, t1 > t2 > t3, to adapt to the detection of different canopy layers density degree lower volume.

9. it is a kind of using fruiter profile modeling spray machine described in any one of claims 1-6 based on laser acquisition and image procossing skill The fruit tree profiling method of art, it is characterised in that: this method includes a kind of centering profile modeling spray mode, the specific steps are as follows:

Step 1: reading fruit tree position and line-spacing data, acquire fruit tree image；

Data processing and control system reads fruit tree position and Planting Row Distance L from planting fruit trees database_RowData, spraying machine is along two Arrange the traveling operation of fruit tree middle position, when spraying machine is marched to fruit tree position corresponding, image real time transfer control Molding block (2) controls left camera (1) and right camera (3) acquires left and right sides fruit tree image respectively；

Step 2: the adjustment of spraying machine position acquires fruit tree image again, calculates fruit tree canopy voidage；

Image and Morphological scale-space are carried out to fruit tree image collected, they only included the bianry image of fruit tree canopy, and it is right Bianry image is marked and frame selects the largest connected region of canopy, calculates the center of the minimum circumscribed rectangle of the connected region Line positions fruit tree canopy center line, and mobile spraying machine keeps the camera of image collecting device corresponding with fruit tree canopy center line, then Secondary acquisition fruit tree image；

Image segmentation and Morphological scale-space are carried out to the fruit tree image acquired again, only included the binary map of fruit tree canopy Picture carries out area filling to the bianry image of the fruit tree image acquired again, and it is empty then to calculate fruit tree canopy by following formula Gap rate K, and using fruit tree canopy voidage K as the standard of reflection fruit tree canopy degree of rarefication；

In formula, K is fruit tree canopy voidage；R₁The total pixel number for being 1 for the pixel value in bianry image；R₂After area filling Bianry image in pixel value be 1 total pixel number；

Step 3: variable scanning obtains fruit tree canopy volume；

Laser data processing and control module (5) controls the first rotating electric machine (607) and the second rotating electric machine (611) respectively with certain Revolving speed is opened, and left laser is respectively set according to the fruit tree canopy voidage K of the left and right sides in laser data processing and control module (5) Sensor (614) and right laser sensor (605) are with different angular scanning speed S_{Laser wheel}The fruit tree of the left and right sides is detected；Swash Light data processing and control module (5) is according to left laser sensor (614), right laser sensor (605), first angle sensor (612) and second angle sensor (610) acquisition data, pass through following formula calculate the left and right sides fruit tree canopy volume:

Y=f (x)；

Ph=(L_i+R)×sinσ；

In formula,

X is the location information for scanning resulting laser boundary point every time；

Y is the canopy contour curve that is fitted after scanning every time；

L_iReach the laser beam length of Malabar Pied Hornbill；

Ph is to arrive tree crown surface as the upright projection of terminal using what laser beam was got to using angular transducer position as starting point；

R is laser wheel radius；

σ is the scanning angle in the laser sensor circular motion of angular transducer acquisition；

H_maxUpright projection when at highest laser point；

H_minUpright projection when at minimum laser point；

V is fruit tree canopy volume；

Step 4: decision is spraying；

Laser data processing and control module by the fruit tree canopy volume of left and right detection respectively with preset standard spray decision value into Row compares, and when fruit tree canopy volume is less than standard spray decision value, selects without spraying operation, spraying machine starts preposition straight To next fruit tree position；When fruit tree canopy volume is more than or equal to standard spray decision value, selection carries out spraying operation, under continuing One step；

Step 5: obtaining the spray height and swing angle range of spraying institute's required airflow, spray amount and spray head；

The equipment operating speed S that spraying control module (27) are returned according to velocity sensor (29)_{Spraying machine}With laser sensor and spray Spacing distance between head calculates spraying delay time；After delay time, spraying control module (27) control device for administration of drugs It starts to work；Spraying control module (27) are calculated spraying required according to the fruit tree canopy volume of the left and right sides by following formula Air quantity, the spray height of spray amount and spray head and swing angle range:

H=(L_i1+L_in+2R)×sinσ；

L₀=(L_i+R)cosσ；

L_i'=L_Row/2；

L=L_i’-L₀；

W=P2+V；

Q=V*q；

Swing angle range: [σ 1, σ 2]；

In formula,

σ is the angle that the laser sensor of angular transducer acquisition turns in real time；

R is laser wheel radius；

L_i1For the laser beam length for the Malabar Pied Hornbill highest point that laser acquisition is arrived；

L_inFor the laser beam length for the Malabar Pied Hornbill lowest part that laser acquisition is arrived；

H is height of tree crown；

L_RowFor Planting Row Distance；

L_iLaser beam length when Malabar Pied Hornbill is reached for laser；

L₀For the floor projection at connection laser wheel center and the laser beam length for reaching Malabar Pied Hornbill；

L_i' it is distance of the laser wheel center away from trunk center；

L is distance of the Malabar Pied Hornbill to trunk center；

P2 is spatial volume of the spraying machine before fruit tree；

V is fruit tree canopy volume；

W is institute's required airflow；

Q is required spray amount；

Q is formulation rate needed for unit volume.

10. according to the method described in claim 9, it is characterized by:

In the step 2, voidage calculated result is changed into digital letter by signal processing by image real time transfer control module (2) Canopy degree of rarefication classification processing is carried out after number, and processing result is sent to laser data processing and control module (5)；Voidage meter The digital signal for calculating result is divided into more sparse K1, medium K2 and tri- canopy degree of rarefication grades of denser K3, gap The bigger expression canopy of rate K value is more sparse, K1 > K2 > K3；

In the step 3, tri- corresponding more sparse K1, medium K2 and denser K3 canopy degree of rarefication grades are respectively set three A different laser sensor angular scanning speed s1, s2, s3, s1 < s2 < s3, to adapt to the spy of different canopy layers density degree lower volume It surveys.