CN109570051A - Chinese chestnut small holes caused by worms detection device based on machine vision, laser and acoustics - Google Patents

Abstract

A chestnut worm eye detection device based on machine vision, laser and acoustics, including a frame, a lifting conveyor belt and a horizontal conveyor belt are arranged in sequence from left to right on the frame. The camera darkroom detection module, the laser detection module, the acoustic detection module and the sorting device are arranged on the top of the horizontal conveyor belt in order from left to right. The side of the frame is provided with an electric control module, a chestnut positioning device and a chestnut unlocking device. Through the multi-level joint intelligent information fusion of machine vision, acoustic detection and laser detection, the present invention can realize high-precision chestnut worm eye detection. On the one hand, the intelligent level of detection can be effectively improved. Improve sorting accuracy and reduce device cost. The invention can be used for on-line real-time detection of insect eyes of chestnut agricultural products, has important significance for promoting the development of deep processing of chestnut agricultural products in my country, and has good market application prospects.

Description

Chestnut bug eye detection device based on machine vision, laser and acoustics

technical field

The invention belongs to the technical field of appearance quality detection of agricultural products used in the agricultural field, and in particular relates to a chestnut worm-eye detection device based on machine vision, laser and acoustics. , so as to complete the real-time detection and sorting of chestnut worm eyes.

Background technique

Chestnut is known as the "king of dried fruits". Chestnut is rich in nutrients. The fruit contains 70.1% sugar and starch, and 7% protein. In addition, it also contains fat, calcium, phosphorus, iron, various vitamins and trace elements, especially the content of vitamin C, B1 and carotene is higher than that of ordinary dried fruits. Chestnuts are rich in nutrients. In addition to being rich in starch, they also contain simple sugars. With disaccharide, carotene, thiamine, riboflavin, niacin, ascorbic acid, protein, fat, inorganic salts and other nutrients. However, in the process of chestnut planting, various pests and diseases are often attacked, so the detection of chestnut insect eyes is very important to improve the quality of chestnut.

At present, most of the detection of chestnut worm eyes is still in the stage of identification and judgment by artificial senses, and there are many inconveniences. Relying on manual hand picking to complete the sorting of chestnut worm eyes, firstly, it wastes a lot of labor, and there is a shortage of labor with the harvest season, and secondly, it is prone to missed inspections, the consistency of test results is poor, the efficiency is low, and it is difficult to meet the requirements of high-standard classification. Not conducive to automation. With the accelerated progress of artificial intelligence technology, multi-level joint detection using intelligent fusion of multi-source information can improve the accuracy of insect eye detection.

Although the chestnut grading method and equipment based on computer vision has gradually become a hot spot, it is generally limited to laboratory research or taking pictures of it with a camera, and the technical means used are relatively simple. In the case of small insect eyes, or complex shapes and positions, accurate judgment cannot be made, and there is a risk of missed detection; or hyperspectral imaging technology is used to detect the chestnut to be inspected, but the cost and complexity of the detection device are increased.

In order to solve the defects of the existing chestnut worm-eye sorting device, a multi-level combined chestnut worm-eye intelligent detection method and device based on machine vision, laser and acoustics are proposed.

Through the domestic patent literature search, we found some relevant patent literature reports, mainly including the following:

1. The publication number is CN 108801971 A, the invention patent titled "Detection Method for Mold Infested Chestnut Based on Hyperspectral Imaging Technology", proposes a detection method for mold infested chestnut based on hyperspectral imaging technology, but the A single hyperspectral imaging technical solution cannot achieve high-precision detection of chestnut worm eyes.

2. The announcement number is CN 107350173 A, and the name is "an automatic sorting device and sorting method for chestnuts". The invention patent uses multiple cameras, computers, PLCs and solenoid valves, etc., and uses machine vision algorithms to identify chestnuts And sorting, according to the color photos of the chestnut surface, filtering and contour extraction are performed to obtain the effective proportion of the damaged chestnut area, so as to realize the sorting of the chestnut to be classified, but the classification device is small in insect eyes, or the shape and location are more complex. In this case, an accurate judgment cannot be made.

3. The publication number is CN 108620338 A, and the invention patent is entitled "Multi-information fusion electromagnetic crawler chestnut detection and screening device", a multi-information fusion electromagnetic crawler chestnut detection and screening device is designed, which integrates infrared spectroscopy and machine vision. , to realize the detection and screening of electromagnetic crawler chestnut under multi-information fusion, but the device mainly uses machine vision and infrared spectroscopy to complete the detection of chestnut damage and mildew, but for the detection of tiny insect eyes, there is a possibility of missed detection.

Although the above-mentioned patent proposes a sorting method and grading device for chestnuts, although some equipment can be graded according to the appearance characteristics, because only the machine vision method is used, or the hyperspectral and infrared spectroscopy technologies are combined, the chestnut damage and mildew detection can be realized. . However, the detection of chestnut worm eyes is not comprehensive, especially when the worm eyes are small, or the shape and location are complex, there is a probability of missed detection.

SUMMARY OF THE INVENTION

In order to solve the deficiencies in the prior art, the present invention provides a chestnut worm-eye detection device based on machine vision, laser and acoustics, which further improves the accuracy of worm-eye detection and reduces the missed detection rate.

In order to solve the above-mentioned technical problems, the present invention adopts the following technical solutions: a chestnut worm-eye detection device based on machine vision, laser and acoustics, including a frame 1, and a lifting conveyor belt 2 and a horizontal conveyor belt are sequentially arranged on the frame 1 from left to right. 4. The hoisting conveyor belt 2 is inclined at a low left and a high right. The frame 1 is provided with a camera darkroom detection module 3, a laser detection module 5, an acoustic detection module 6 and a sorting system located above the horizontal conveyor belt 4 in order from left to right. Device 7, the side of the frame 1 is provided with an electric control module 9, a chestnut positioning device and a chestnut unlocking device, the working part of the chestnut positioning device is located above the lifting conveyor belt 2, and the working part of the chestnut unlocking device is located above the horizontal conveyor belt 4 Acoustic between the detection module 6 and the sorting device 7 .

The lifting conveyor belt 2 and the horizontal conveyor belt 4 are both made of metal thermal conductor materials. The upper surfaces of the lifting conveyor belt 2 and the horizontal conveyor belt 4 are evenly spaced 5 to 8 cm in the left-right direction with grooves 54 for storing the chestnuts to be transported. Conveyor belt 2 and horizontal conveyor belt 4 are combined into a complete conveyor belt.

The camera darkroom detection module 3 includes a cavity box 10 , a ring LED light 11 and an industrial camera 12 . The bottom of the cavity box 10 is open and faces the horizontal conveyor belt 4 vertically downward. The left and right sides of the cavity box 10 are screwed through. Fixed on the rack 1, the left and right sides of the bottom edge of the cavity box 10 are left with rectangular gaps for the horizontal transportation of chestnuts; the ring LED light 11 and the industrial camera 12 are installed in the top center of the cavity box 10. Location.

The laser detection module 5 includes a first X-direction stepping motor 13, a first X-direction moving screw 14, a first X-direction bracket 15, a first Y-direction stepping motor 16, a first Y-direction moving screw 17, a first Y-direction guide rail 18, laser detection fixture 19 and laser detection sub-module 20;

The X direction is the left and right horizontal direction, and the Y direction is the front and rear horizontal direction. The frame 1 is provided with a first support guide rail 55 arranged along the X direction and in the shape of a rectangular frame on the front and rear sides of the horizontal conveyor belt 4 respectively. Each of the support guide rails 55 is provided with the first X-direction moving screw 14, one end of the first X-direction moving screw 14 is rotatably connected to the first support guide rail 55, and the other side of the first X-direction moving screw 14 is rotatably connected to the first support guide rail 55. One end is drivingly connected with the first X-direction stepping motor 13, the first X-direction moving screw 14 is threadedly connected with the first X-direction bracket 15 sliding along the first support guide rail 55, the first Y-direction The front and rear ends of the guide rail 18 are fixedly connected to the upper ends of the two first X-direction brackets 15, the first Y-direction guide rail 18 is a hollow rod-shaped structure, and the first Y-direction moving screw 17 is arranged in the front-rear direction on the first Y-direction guide rail. 18, one end of the first Y-direction moving screw 17 is rotatably connected to one end of the first Y-direction guide rail 18, and the other end of the first Y-direction moving screw 17 is drivingly connected to the first Y-direction stepping motor 16. , the laser detection fixture 19 is screwed on the first Y-direction moving screw 17 and slides along the first Y-direction guide rail 18 .

The acoustic detection module 6 includes a second X-direction stepping motor 21, a second X-direction moving screw 22, a second X-direction bracket 23, a second Y-direction stepping motor 24, a second Y-direction moving screw 25, a second Y-direction guide rail 26, sliding frame, C-axis motion motor 27, C-axis reducer 28, high-pressure gas needle 29, high-pressure gas pipe 30, acoustic sensor 31 and first solenoid valve 51; X-direction is the left-right horizontal direction, Y-direction is front and rear horizontal direction;

On the frame 1, a second support guide rail 56 arranged along the X direction and in the shape of a rectangular frame is respectively provided on the front and rear sides of the horizontal conveyor belt 4, and each second support guide rail 56 is provided with a said second X To the moving screw 22, one end of the second X-direction moving screw 22 is rotatably connected to the second support guide rail 56, and the other end of the second X-direction moving screw 22 is driven with the second X-direction stepping motor 21. Connection, the second X-direction moving screw 22 is threadedly connected with the second X-direction bracket 23 sliding along the second support guide rail 56, and the front and rear ends of the second Y-direction guide rail 26 are fixedly connected to the two second X-direction brackets 23. To the upper end of the bracket 23, the second Y-direction guide rail 26 is a hollow rod-shaped structure, the second Y-direction moving screw 25 is arranged inside the second Y-direction guide rail 26 in the front-rear direction, and one end of the second Y-direction moving screw 25 rotates Connected to one end of the second Y-direction guide rail 26, the other end of the second Y-direction moving screw 25 is connected to the second Y-direction stepping motor 24, the C-axis moving motor 27 and the C-axis reducer 28 are arranged in Inside the sliding frame, the sliding frame is threaded on the second Y-direction moving screw 25 and is slidably arranged along the second Y-direction guide rail 26 , and the power output end of the C-axis motion motor 27 is drive-connected with the power input end of the C-axis reducer 28 , the high-pressure gas needle 29 is installed on the power output shaft of the C-axis reducer 28, the power output shaft of the C-axis reducer 28 is arranged horizontally along the left-right direction, and the high-pressure gas needle 29 is perpendicular to the power output shaft of the C-axis reducer 28, The air inlet of the high pressure air needle 29 is connected with the air outlet of the high pressure air pipe 30, the first solenoid valve 51 is arranged on the high pressure air pipe 30, the air compressor is connected to the air inlet of the high pressure air pipe 30, and the acoustic sensor 31 is arranged on the high pressure air needle 29 at the jet port.

The sorting device 7 includes a sorting support plate 32, a sorting cylinder 33, a sorting rod 34, a sorting rod head 35 and a slideway 36; the slideway 36 is installed on the right front side of the frame 1 by bolting; the sorting support plate 32 is vertically installed on the right rear position of the frame 1 through bolt connection, and the sorting cylinder 33 is fixedly installed on the left side of the sorting support plate 32. The rods 34 are arranged horizontally in the left-right direction.

The chestnut positioning device is a cold air nozzle 37 and a water drip pipe arranged side by side above the lifting conveyor belt 2. The water drip pipe is located on the left side of the cold air nozzle 37. the groove 54; the cold air nozzle 37 is provided with a second solenoid valve 52;

The chestnut unlocking device is a hot gas nozzle 38 arranged above the horizontal conveyor belt 4 , the nozzle of the hot gas nozzle 38 is set downward and faces the passing groove 54 , and a third solenoid valve 53 is provided on the hot gas nozzle 38 .

The electronic control module 9 includes an embedded ARM single-chip microcomputer, an input/output sub-module, an industrial Ethernet communication sub-module, a display sub-module, and a light warning sub-module; the embedded ARM single-chip microcomputer is connected to the input/output sub-module through an on-chip bus;

The embedded ARM single-chip microcomputer is respectively connected with the industrial camera 12 and the laser detection sub-module 20 through the industrial Ethernet communication sub-module; The stepper motor 13, the first Y-direction stepper motor 16, the second X-direction stepper motor 21, the C-axis motion motor 27, the second Y-direction stepper motor 24, and the sorting cylinder 33 are connected through a control signal line, according to The control process controls the corresponding stepper motor to act in an orderly manner to realize positioning and sorting; the input/output sub-module is also connected with the first solenoid valve 51, the second solenoid valve 52 and the third solenoid valve 53. According to the action flow, Open or close the solenoid valve to realize the functions of detection, freezing and thawing respectively; the input/output sub-module is also connected to and controls the opening and closing of the ring LED light 11 and its brightness; the input/output sub-module is also connected with the acoustic sensor 31 .

Using the above technical solution, the detection method of the chestnut worm eye detection device based on machine vision, laser and acoustics includes the following steps:

(1) The lifting conveyor belt 2 and the horizontal conveyor belt 4 are driven by the stepping motor to move from left to right, and the chestnuts are placed one by one into the leftmost groove 54 on the top of the lifting conveyor belt 2, and the water in the drip pipe drops directly below. In the groove 54, the chestnuts are placed one by one in the groove 54 with liquid water of the lifting conveyor belt 2 from the left side, and the lifting conveyor belt 2 and the horizontal conveyor belt 4 drive the chestnuts to move to the right;

(2) When the chestnut moves under the spout of the cold air nozzle 37, the electronic control module 9 controls the second solenoid valve 52 to open, and the high-pressure cold air is sprayed to the groove 54 with liquid water, and the liquid water freezes and freezes to fix the chestnut. into the groove 54;

(3) The lifting conveyor belt 2 and the horizontal conveyor belt 4 drive the chestnut to move to the right in the camera detection module 3, and the position of the chestnut is calculated by the machine vision method of the electronic control module 9. The industrial camera 12 of the camera detection module 3 is in the electronic control module. Under the control of the module 9, the chestnut to be detected is photographed, sent to the electronic control module 9 through the industrial Ethernet communication sub-module, and processed by the embedded ARM single-chip microcomputer. The height H of the physical space is a known quantity. Using the monocular vision method, the angle a can be calculated by the ratio of similar triangles, and L1 is the length of the pixel, which can be obtained by the image processing algorithm, so as to calculate the L2 value in the physical space, Obtain the coordinate position of the chestnut to be detected in the photographing darkroom; then obtain the size of the chestnut to be detected and the area information of the suspected bug eye through the feature extraction and analysis of the image; the calculated position and size information of the chestnut to be detected are the follow-up laser The detection module 5 and the acoustic detection module 6 provide positioning information;

(4) The horizontal conveyor belt 4 drives the chestnut to move to the right under the laser detection module 5 , and the first X-direction stepping motor 13 drives the first X-direction moving screw 14 to rotate, thereby making the first X-direction moving screw 14 rotate. The first X-direction bracket 15 moves to the left or right along the first support guide rail 55, and the first Y-direction guide rail 18, the laser detection fixture 19 and the laser detection sub-module 20 also move to the left or right; the first Y-direction steps The motor 16 drives the first Y-direction moving screw 17 to rotate, so that the laser detection fixture 19 and the first Y-direction moving screw 17 move forward or backward along the first Y-direction guide rail 18, and the laser detection sub-module 20 also moves forward. Move forward or backward; move the laser detection fixture 19 to the top of the chestnut to be detected through the X-Y plane positioning, and the laser detection sub-module 20 is aimed at the vertical top of the suspected chestnut worm-eye area; after the laser detection sub-module 20 is detected, the detection After processing the laser reflected signal intensity and position height information, it is handed over to the electronic control module 9 for analysis;

The specific detection process of the laser detection sub-module 20 in step (4) is as follows: the laser detection sub-module 20 is positioned on the XY plane to scan the suspected chestnut worm eye area with an interval of 2 mm in the X direction and the Y direction, and the scanning length and width are both 14 mm. 8×8=64 times, the laser detection sub-module 20 obtains the laser measurement height data set H ₆₄ , and the height is the distance from the laser emitting head in the laser detection sub-module to the chestnut surface to be detected, as shown below:

Further, the average height difference is calculated along the X direction and the Y direction for each data in the height data set H ₆₄ , as shown in the following formula:

In the above formula, when the subscript (i or j) of the height dataset is less than or equal to 0 or greater than 9, the average height difference ( ); if the average height difference at the current measurement location ( ) is greater than the threshold of 0.18mm, indicating that the current measurement position is the edge of the suspected chestnut worm's eye; then the detected edge position of the suspected chestnut worm's eye (point position set) is used to close the surface with a B-spline curve, and the area of the suspected chestnut worm's eye area is calculated, If the area is greater than the threshold value of 15 mm ² , the laser detection module 5 judges that the chestnut to be tested currently has worm-eye defects, and on the contrary, judges that the chestnut to be tested does not have worm-eye defects;

(5) After the horizontal conveyor belt 4 drives the chestnut to move to the right under the acoustic detection module 6, the second X-direction stepping motor 21 drives the second X-direction moving screw 22 to rotate, so that the second X-direction moving screw 22 rotates. The second X-direction bracket 23 moves to the left or right along the second support guide rail 56, and the second Y-direction guide rail 26, the carriage, the C-axis motion motor 27, the C-axis reducer 28 and the high-pressure gas needle 29 also move left or right. Move to the right; the second Y-direction stepping motor 24 drives the second Y-direction moving screw 25 to rotate, so that the carriage with the second Y-direction moving screw 25 moves forward or backward along the second Y-direction guide rail 26 , the C-axis motion motor 27, the C-axis reducer 28 and the high-pressure gas needle 29 also move forward or backward; move the second C-axis reducer 28 to the top of the chestnut to be detected through the X-Y plane positioning, and then pass the second C-axis The rotary motion of the shaft reducer 28 aligns the high-pressure gas needle 29 at the vertical top of the suspected chestnut worm's eye; after the high-pressure gas needle 29 ejects airflow, the acoustic sensor 31 collects the airflow echo signal at the suspected chestnut worm's eye position, and sends it to the electronic control module 9 for analysis.

In step (5), the specific process of the cooperation of the acoustic detection module 6 and the electronic control module 9 for detection is as follows: under the control of the electronic control module 9, the acoustic detection module 6 moves to the top of the chestnut to be detected through the XY plane positioning; decelerates through the C axis The movement of the machine 28 collects the acoustic signal data of three positions, the position 1 is the vertical directly above the suspected chestnut worm's eye area, and the position 2 is the vertical direction of the surface of the suspected chestnut worm's eye area, which is obtained by calculating the HX ₄₉ and HY ₄₉ data. , position 3 is the symmetrical position of position 1 about the vertical line of the suspected chestnut worm's eye area; the electronic control module 9 opens the first solenoid valve 51, the high-pressure gas acts on the suspected chestnut worm's eye area through the high-pressure gas needle 29, and the acoustic sensor 31 collects 2 After seconds of data, the data is sent to the electronic control module 9 through the input/output sub-module. After the success, the electronic control module 9 closes the first solenoid valve 51; similarly, the electronic control module 9 performs acoustic signal data for position 2 and position 3. Acquisition; After the acquisition of the acoustic signal data of all three positions is completed, the embedded ARM single-chip microcomputer of the electronic control module 9 is processed according to the following process:

A. The acoustic signal data collected at position 1, position 2 and position 3 are processed according to the procedures B~E;

B. Perform pre-emphasis, framing and windowing (12 short-term analysis windows) operations on the 4-second acoustic signal data collected at each location;

C. For each short-term analysis window, obtain the corresponding frequency spectrum through FFT, and then obtain the Mel (Mel) frequency spectrum through the Mel (Mel) filter bank;

D. Perform cepstrum analysis on the Mel (Mel) spectrum (take logarithm, after DCT discrete cosine transform processing, take the 2nd to 13th coefficients after DCT discrete cosine transform as MFCC coefficients), obtain 12 Mel frequency cepstral coefficient MFCC;

E. After the 12 Mel-frequency cepstral coefficients MFCC of the above 12 short-term analysis windows are subjected to 256-level grayscale processing, a Mel-frequency cepstral coefficient (MFCC) feature map is obtained, with a size of 12×12 256-level gray further, the 12×12 256-level grayscale image is sent to the residual network for identification, and the identification result is a two-dimensional vector [P ₀ , P ₁ ], when P ₀ is greater than P ₁ , the acoustic detection module Mark the current position as the chestnut to be tested has worm-eye defects, otherwise, judge that the chestnut to be tested does not have worm-eye defects;

F. Summarize and analyze the preliminary results of position 1, position 2 and position 3. If there is a worm-eye defect mark in any of the positions of position 1, position 2 and position 3, the acoustic detection module judges that there is a worm-eye defect in the chestnut to be tested. There is no bug eye defect in the chestnut to be tested;

The final identification result after the combined detection of camera vision, laser and acoustics is displayed through the display sub-module; chestnuts with insect eyes are also displayed in an alarm through the light warning sub-module to remind the staff; the display sub-module can also detect the detected Statistical analysis of data, displaying statistical data by hour or batch, which is convenient for manual viewing;

(6) When the detected chestnut is transported to the right under the nozzle of the hot air nozzle 38, the solid ice inside the groove 54 of the horizontal conveyor belt 4 of the metal heat conductor is rapidly heated by the hot air, and turned into liquid water. , which is convenient for the sorting device 7 to carry out classification;

(7) When the chestnut with worm eyes is transported to the right to the position corresponding to the front and rear of the sorting device 7, the sorting cylinder 33 drives the sorting rod 34 to move, and pushes the sorting rod head 35 to push the chestnut with the worm eye defect out of the chute 36. The sorting cylinder 33 is connected with the electronic control module 9, and under the control of the electronic control module 9, the sorting rod 34 is pushed to move, so that the chestnut with bug eye defect falls into the defective product frame through the slideway 36, and the qualified Chestnuts are transported to the right by the horizontal conveyor belt 4.

The residual network includes 2 residual blocks, the A1 convolutional layer, the A2 Batch Norm layer, the A3 activation layer, the B1 convolutional layer, the B2 Batch Norm layer and the B3 activation layer constitute the first residual block. The A1 volume The accumulation layer, the A2 Batch Norm layer, the A3 activation layer, the B1 convolution layer, the B2 Batch Norm layer and the B3 activation layer are connected in turn. Between layers; the C1 convolutional layer, the C2 Batch Norm layer, the C3 activation layer, the D1 convolutional layer, the D2 Batch Norm layer and the D3 activation layer constitute the second residual block, the C1 convolutional layer, the C2 BatchNorm layer , C3 activation layer, D1 convolution layer, D2 Batch Norm layer and D3 activation layer are connected end to end, further, the data at the entrance of C1 convolution layer can be short-circuit inserted between D2 Batch Norm layer and D3 activation layer; the first residual The difference block and the second residual block are connected end to end in turn. After the MFCC grayscale image is input into the two residual blocks of the residual network, it enters the E1 fully connected layer, and the data flows into the F1 soft regression layer, and finally The identification result is output.

Further, the residual network is trained offline, and a total of 3160 samples of data are collected for training, of which 1510 are data without bug eye defects, and 1650 are data with bug eye defects. Inside the embedded ARM microcontroller of the control module 9.

Obtain the area information of suspected insect eyes from the detection results of machine vision; detect the position information of suspected insect eyes by machine vision, and obtain the laser detection results through laser detection; Acoustic detection, obtain the acoustic detection results; from the above machine vision results: the position and size information of the suspected insect eye, the laser detection result and the acoustic detection result, identify through the SVM model, and output the final identification result: normal and insect eye, and then by the display sub-module and the light warning sub-module for output.

The SVM model is trained offline and shares the same training samples with the residual network, that is, training on data collected from a total of 3160 samples, of which 1510 are data without bug eye defects, and 1650 are data with bug eye defects. The data of the bug eye defect and the trained SVM model parameters are stored in the embedded ARM single-chip microcomputer of the electronic control module 9 .

The present invention can also be set to cascade operation, two devices of the present invention and a robot hand with machine vision are combined to form a set of detection units; the robot hand with machine vision is located in the middle of the two devices of the present invention, and machine vision is used for auxiliary positioning , to realize that the chestnut to be detected is moved from one device of the present invention to another device of the present invention, and the chestnut is turned over during the moving process, and then the detection is performed, which can further improve the accuracy of the detection.

To sum up, the present invention has the following beneficial effects relative to the prior art:

Through the multi-level joint intelligent information fusion of machine vision, acoustic detection and laser detection, the present invention can realize high-precision chestnut worm eye detection. On the one hand, the intelligent level of detection can be effectively improved. Improve sorting accuracy and reduce device cost. The invention can be used for on-line real-time detection of insect eyes of chestnut agricultural products, has important significance for promoting the development of deep processing of chestnut agricultural products in my country, and has good market application prospects.

Description of drawings

Fig. 1 is the three-dimensional structure schematic diagram of the device of the present invention under one viewing angle;

Fig. 2 is the three-dimensional structure schematic diagram of the device of the present invention under another viewing angle;

3 is a schematic top view of the structure of the laser detection module in the device of the present invention;

Fig. 4 is the left side view of Fig. 3;

5 is a schematic structural diagram of an acoustic detection module in the device of the present invention;

Fig. 6 is the structural representation of the sorting device in the device of the present invention;

Fig. 7 is the bottom view schematic diagram of photographing darkroom detection module;

Figure 8 is a schematic diagram of chestnut position calculation in machine vision;

Fig. 9 is the working schematic diagram of the laser detection module;

Fig. 10 is the working schematic diagram of acoustic detection module;

Figure 11 is a structural diagram of a residual network (ResNet) used for acoustic signal data processing;

Figure 12 is a flow chart of multi-level joint intelligent detection of machine vision, laser and acoustics;

FIG. 13 is a structural diagram of an electronic control module.

Detailed ways

As shown in Figures 1 to 6, the chestnut worm eye detection device based on machine vision, laser and acoustics of the present invention includes a frame 1, and the frame 1 is provided with a lifting conveyor belt 2 and a horizontal conveyor belt 4 sequentially from left to right. , the lifting conveyor belt 2 is inclined to the left and right, and the frame 1 is sequentially provided with a camera darkroom detection module 3, a laser detection module 5, an acoustic detection module 6 and a sorting device located above the horizontal conveyor belt 4 from left to right. 7. The side of the frame 1 is provided with an electric control module 9, a chestnut positioning device and a chestnut unlocking device, the working part of the chestnut positioning device is located above the lifting conveyor belt 2, and the working part of the chestnut unlocking device is located above the horizontal conveyor belt 4 Acoustic detection between the module 6 and the sorting device 7 .

The lifting conveyor belt 2 and the horizontal conveyor belt 4 are both made of metal thermal conductor materials. The upper surfaces of the lifting conveyor belt 2 and the horizontal conveyor belt 4 are evenly spaced 5 to 8 cm in the left-right direction with grooves 54 for storing the chestnuts to be transported. The conveyor belt 2 and the horizontal conveyor belt 4 are combined into a complete conveyor belt. The chestnut to be tested can be transported without changing the conveyor belt groove 54 to realize the whole process of transportation from lifting to grading. The groove 54 stores a small amount of liquid water inside for freezing and grading. Fixed chestnuts.

The camera darkroom detection module 3 includes a cavity box 10 , a ring LED light 11 and an industrial camera 12 . The bottom of the cavity box 10 is open and faces the horizontal conveyor belt 4 vertically downward. The left and right sides of the cavity box 10 are screwed through. Fixed on the rack 1, the left and right sides of the bottom edge of the cavity box 10 are left with rectangular gaps for the horizontal transportation of chestnuts; the ring LED light 11 and the industrial camera 12 are installed in the top center of the cavity box 10. Location.

The laser detection module 5 includes a first X-direction stepping motor 13, a first X-direction moving screw 14, a first X-direction bracket 15, a first Y-direction stepping motor 16, a first Y-direction moving screw 17, a first Y-direction guide rail 18, laser detection fixture 19 and laser detection sub-module 20;

The X direction is the left and right horizontal direction, and the Y direction is the front and rear horizontal direction. The frame 1 is provided with a first support guide rail 55 arranged along the X direction and in the shape of a rectangular frame on the front and rear sides of the horizontal conveyor belt 4 respectively. Each of the support guide rails 55 is provided with the first X-direction moving screw 14, one end of the first X-direction moving screw 14 is rotatably connected to the first support guide rail 55, and the other side of the first X-direction moving screw 14 is rotatably connected to the first support guide rail 55. One end is drivingly connected with the first X-direction stepping motor 13, the first X-direction moving screw 14 is threadedly connected with the first X-direction bracket 15 sliding along the first support guide rail 55, the first Y-direction The front and rear ends of the guide rail 18 are fixedly connected to the upper ends of the two first X-direction brackets 15, the first Y-direction guide rail 18 is a hollow rod-shaped structure, and the first Y-direction moving screw 17 is arranged in the front-rear direction on the first Y-direction guide rail. 18, one end of the first Y-direction moving screw 17 is rotatably connected to one end of the first Y-direction guide rail 18, and the other end of the first Y-direction moving screw 17 is drivingly connected to the first Y-direction stepping motor 16. , the laser detection fixture 19 is screwed on the first Y-direction moving screw 17 and slides along the first Y-direction guide rail 18 .

The acoustic detection module 6 includes a second X-direction stepping motor 21, a second X-direction moving screw 22, a second X-direction bracket 23, a second Y-direction stepping motor 24, a second Y-direction moving screw 25, a second Y-direction guide rail 26, sliding frame, C-axis motion motor 27, C-axis reducer 28, high-pressure gas needle 29, high-pressure gas pipe 30, acoustic sensor 31 and first solenoid valve 51; X-direction is the left-right horizontal direction, Y-direction is front and rear horizontal direction;

On the frame 1, a second support guide rail 56 arranged along the X direction and in the shape of a rectangular frame is respectively provided on the front and rear sides of the horizontal conveyor belt 4, and each second support guide rail 56 is provided with a said second X To the moving screw 22, one end of the second X-direction moving screw 22 is rotatably connected to the second support guide rail 56, and the other end of the second X-direction moving screw 22 is driven with the second X-direction stepping motor 21. Connection, the second X-direction moving screw 22 is threadedly connected with the second X-direction bracket 23 sliding along the second support guide rail 56, and the front and rear ends of the second Y-direction guide rail 26 are fixedly connected to the two second X-direction brackets 23. To the upper end of the bracket 23, the second Y-direction guide rail 26 is a hollow rod-shaped structure, the second Y-direction moving screw 25 is arranged inside the second Y-direction guide rail 26 in the front-rear direction, and one end of the second Y-direction moving screw 25 rotates Connected to one end of the second Y-direction guide rail 26, the other end of the second Y-direction moving screw 25 is connected to the second Y-direction stepping motor 24, the C-axis moving motor 27 and the C-axis reducer 28 are arranged in Inside the sliding frame, the sliding frame is threaded on the second Y-direction moving screw 25 and is slidably arranged along the second Y-direction guide rail 26 , and the power output end of the C-axis motion motor 27 is drive-connected with the power input end of the C-axis reducer 28 , the high-pressure gas needle 29 is installed on the power output shaft of the C-axis reducer 28, the power output shaft of the C-axis reducer 28 is arranged horizontally along the left-right direction, and the high-pressure gas needle 29 is perpendicular to the power output shaft of the C-axis reducer 28, The air inlet of the high pressure air needle 29 is connected with the air outlet of the high pressure air pipe 30, the first solenoid valve 51 is arranged on the high pressure air pipe 30, the air compressor is connected to the air inlet of the high pressure air pipe 30, and the acoustic sensor 31 is arranged on the high pressure air needle 29 at the jet port.

The sorting device 7 includes a sorting support plate 32, a sorting cylinder 33, a sorting rod 34, a sorting rod head 35 and a slideway 36; the slideway 36 is installed on the right front side of the frame 1 by bolting; the sorting support plate 32 is vertically installed on the right rear position of the frame 1 through bolt connection, and the sorting cylinder 33 is fixedly installed on the left side of the sorting support plate 32. The rods 34 are arranged horizontally in the left-right direction.

The chestnut positioning device is a cold air nozzle 37 and a water drip pipe arranged side by side above the lifting conveyor belt 2. The water drip pipe is located on the left side of the cold air nozzle 37. the groove 54; the cold air nozzle 37 is provided with a second solenoid valve 52;

The chestnut unlocking device is a hot air nozzle 38 arranged above the horizontal conveyor belt 4. The nozzle of the hot air nozzle 38 is set downward and faces the groove 54 passing through. The hot air nozzle 38 is provided with a third solenoid valve 53 and a second electromagnetic valve. Both the valve 52 and the third solenoid valve 53 are connected to and controlled by the electronic control module 9 .

As shown in Figure 13, the electronic control module 9 includes an embedded ARM single-chip microcomputer, an input/output sub-module, an industrial Ethernet communication sub-module, a display sub-module, and a light warning sub-module; the embedded ARM single-chip microcomputer communicates with the input/output through the on-chip bus The sub-modules are connected; the embedded ARM single-chip microcomputer is respectively connected with the industrial camera 12 and the laser detection sub-module 20 through the industrial Ethernet communication sub-module; the industrial Ethernet communication sub-module has built-in dual industrial Ethernet network ports, and the input/output sub-modules are respectively Through control signals with the first X-direction stepper motor 13, the first Y-direction stepper motor 16, the second X-direction stepper motor 21, the C-axis motion motor 27, the second Y-direction stepper motor 24, and the sorting cylinder 33 The line is connected, and the corresponding stepper motor is controlled in an orderly manner according to the control process to realize positioning and sorting; the input/output sub-module is also connected with the first solenoid valve 51, the second solenoid valve 52 and the third solenoid valve 53 , according to the action flow, open or close the solenoid valve to realize the functions of detection, freezing and thawing respectively; the input/output sub-module is also connected to and controls the opening and closing of the ring LED light 11 and its brightness; the input/output sub-module is also connected with the acoustic The sensor 31 is connected.

The detection method of the chestnut worm eye detection device based on machine vision, laser and acoustics includes the following steps:

(1) The lifting conveyor belt 2 and the horizontal conveyor belt 4 are driven by the stepping motor to move from left to right, and the chestnuts are placed one by one into the leftmost groove 54 on the top of the lifting conveyor belt 2, and the water in the drip pipe drops directly below. In the groove 54, the chestnuts are placed one by one in the groove 54 with liquid water of the lifting conveyor belt 2 from the left side, and the lifting conveyor belt 2 and the horizontal conveyor belt 4 drive the chestnuts to move to the right;

(2) When the chestnut moves under the spout of the cold air nozzle 37, the electronic control module 9 controls the second solenoid valve 52 to open, and the high-pressure cold air is sprayed to the groove 54 with liquid water, and the liquid water freezes and freezes to fix the chestnut. into the groove 54;

(3) The lifting conveyor belt 2 and the horizontal conveyor belt 4 are driven by a conveying stepper motor (conventional technology in the art), the width of each section of the conveyor belt is 30~50mm, and is initially positioned by the stepper motor, and the chestnut to be tested is sent into the camera darkroom Inside the detection module 3, the position is further calculated by the machine vision algorithm of the electronic control module 9, and the stepping motor is driven for precise positioning, so that the chestnut to be detected is in the area below the industrial camera 12;

The lifting conveyor belt 2 and the horizontal conveyor belt 4 drive the chestnut to move to the right in the camera darkroom detection module 3, as shown in Figures 7 and 8, the chestnut position is calculated by the machine vision method of the electronic control module 9, and the photo darkroom detection module 3 Under the control of the electronic control module 9, the industrial camera 12 takes pictures of the chestnut to be detected, and is sent to the electronic control module 9 through the industrial Ethernet communication sub-module, and processed by the embedded ARM single-chip microcomputer. The process is as follows: by calibrating the industrial camera 12 , the height H from the lens of the industrial camera 12 to the physical space is a known quantity, and the monocular vision method is adopted. The angle a can be calculated by the similar triangle ratio, and L1 is the length of the pixel, which can be obtained by an image processing algorithm (conventional technology in the art). ), thereby calculating the L2 value in the physical space, and obtaining the coordinate position of the chestnut to be detected in the photo darkroom; then through the feature extraction and analysis of the image, the size of the chestnut to be detected and the area information of the suspected bug eye are obtained; the calculated The position and size information of the chestnut to be detected provides positioning information for the subsequent laser detection module 5 and acoustic detection module 6;

(4) The horizontal conveyor belt 4 drives the chestnut to move to the right under the laser detection module 5 , and the first X-direction stepping motor 13 drives the first X-direction moving screw 14 to rotate, thereby making the first X-direction moving screw 14 rotate. The first X-direction bracket 15 moves to the left or right along the first support guide rail 55, and the first Y-direction guide rail 18, the laser detection fixture 19 and the laser detection sub-module 20 also move to the left or right; the first Y-direction steps The motor 16 drives the first Y-direction moving screw 17 to rotate, so that the laser detection fixture 19 and the first Y-direction moving screw 17 move forward or backward along the first Y-direction guide rail 18, and the laser detection sub-module 20 also moves forward. Move forward or backward; move the laser detection fixture 19 to the top of the chestnut to be detected through the X-Y plane positioning, and the laser detection sub-module 20 is aimed at the vertical top of the suspected chestnut worm-eye area; after the laser detection sub-module 20 is detected, the detection After processing the laser reflected signal intensity and position height information, it is handed over to the electronic control module 9 for analysis;

The specific detection process of the laser detection sub-module 20 in step (4) is as follows: as shown in FIG. 9 , the laser detection sub-module 20 scans the suspected chestnut worm eye area in the X and Y directions of the XY plane at an interval of 2 mm. The widths are all 14mm, with a total of 8×8=64 times. The laser detection sub-module 20 obtains the laser measurement height data set H ₆₄ , and the height is the distance from the laser emitting head in the laser detection sub-module to the chestnut surface to be detected, as shown below:

Further, the average height difference is calculated along the X direction and the Y direction for each data in the height data set H ₆₄ , as shown in the following formula:

In the above formula, when the subscript (i or j) of the height dataset is less than or equal to 0 or greater than 9, the average height difference ( ); if the average height difference at the current measurement location ( ) is greater than the threshold of 0.18mm, indicating that the current measurement position is the edge of the suspected chestnut worm's eye; then the detected edge position of the suspected chestnut worm's eye (point position set) is used to close the surface with a B-spline curve, and the area of the suspected chestnut worm's eye area is calculated, If the area is greater than the threshold value of 15 mm ² , the laser detection module 5 judges that the chestnut to be tested currently has worm-eye defects, and on the contrary, judges that the chestnut to be tested does not have worm-eye defects;

(5) The horizontal conveyor belt 4 drives the chestnut to move to the right under the acoustic detection module 6, and the second X-direction stepping motor 21 drives the second X-direction moving screw 22 to rotate, so that the The second X-direction bracket 23 moves to the left or right along the second support guide rail 56 , and the second Y-direction guide rail 26 , the carriage, the C-axis motion motor 27 , the C-axis reducer 28 and the high-pressure gas needle 29 also move to the left or to the right Move to the right; the second Y-direction stepping motor 24 drives the second Y-direction moving screw 25 to rotate, so that the carriage with the second Y-direction moving screw 25 moves forward or backward along the second Y-direction guide rail 26, The C-axis motion motor 27, the C-axis reducer 28 and the high-pressure gas needle 29 also move forward or backward; the second C-axis reducer 28 is moved to the top of the chestnut to be detected through the X-Y plane positioning, and then passes through the second C-axis. The rotary motion of the reducer 28 aligns the high-pressure gas needle 29 at the vertical top of the suspected chestnut worm's eye; after the high-pressure gas needle 29 ejects airflow, the acoustic sensor 31 collects the airflow echo signal at the position of the suspected chestnut worm's eye, and sends it to the electronic control module 9 analysis.

In step (5), the specific process of cooperative detection between the acoustic detection module 6 and the electronic control module 9 is as follows: As shown in FIG. Above; through the movement of the C-axis reducer 28, the acoustic signal data of three positions are collected, the position 1 is the vertical right above the suspected chestnut worm's eye region, and the position 2 is the vertical direction of the surface of the suspected chestnut worm's eye region through the HX ₄₉ and HY ₄₉ data is calculated and obtained, position 3 is the symmetrical position of position 1 about the vertical line of the suspected chestnut worm's eye area; the electronic control module 9 opens the first solenoid valve 51, and the high-pressure gas acts on the suspected chestnut worm's eye area through the high-pressure gas needle 29. After the sensor 31 collects the data for 2 seconds, the data is sent to the electronic control module 9 through the input/output sub-module. After the success, the electronic control module 9 closes the first solenoid valve 51; Acoustic signal data collection is performed; after the acoustic signal data collection of all three positions is completed, the embedded ARM single-chip microcomputer of the electronic control module 9 is processed according to the following process:

A. The acoustic signal data collected at position 1, position 2 and position 3 are processed according to the procedures B~E;

B. Perform pre-emphasis, framing and windowing (12 short-term analysis windows) operations on the 4-second acoustic signal data collected at each location;

C. For each short-term analysis window, obtain the corresponding frequency spectrum through FFT, and then obtain the Mel (Mel) frequency spectrum through the Mel (Mel) filter bank;

D. Perform cepstrum analysis on the Mel (Mel) spectrum (take logarithm, after DCT discrete cosine transform processing, take the 2nd to 13th coefficients after DCT discrete cosine transform as MFCC coefficients), obtain 12 Mel frequency cepstral coefficient MFCC;

E. After the 12 Mel-frequency cepstral coefficients MFCC of the above 12 short-term analysis windows are subjected to 256-level grayscale processing, a Mel-frequency cepstral coefficient (MFCC) feature map is obtained, with a size of 12×12 256-level gray further, the 12×12 256-level grayscale image is sent to the residual network for identification, and the identification result is a two-dimensional vector [P ₀ , P ₁ ], when P ₀ is greater than P ₁ , the acoustic detection module Mark the current position as the chestnut to be tested has worm-eye defects, otherwise, judge that the chestnut to be tested does not have worm-eye defects;

F. Summarize and analyze the preliminary results of position 1, position 2 and position 3. If there is a worm-eye defect mark in any of the positions of position 1, position 2 and position 3, the acoustic detection module judges that there is a worm-eye defect in the chestnut to be tested. There is no bug eye defect in the chestnut to be tested;

The final identification result after the combined detection of camera vision, laser and acoustics is displayed through the display sub-module; chestnuts with insect eyes are also displayed in an alarm through the light warning sub-module to remind the staff; the display sub-module can also detect the detected Statistical analysis of data, displaying statistical data by hour or batch, which is convenient for manual viewing;

(6) When the detected chestnut is transported to the right under the spout of the hot air nozzle 38, the solid ice inside the groove 54 of the horizontal conveyor belt 4 of the metal heat conductor is rapidly heated by the hot air to turn it into liquid. water, which is convenient for classification by the sorting device 7;

(7) When the chestnut with worm eyes is transported to the right to the position corresponding to the front and rear of the sorting device 7, the sorting cylinder 33 drives the sorting rod 34 to move, and pushes the sorting rod head 35 to push the chestnut with the worm eye defect out of the chute. The entrance of 36; the sorting cylinder 33 is connected to the electronic control module 9, and under the control of the electronic control module 9, the sorting rod 34 is pushed to move, so as to realize that the chestnut with bug eye defect falls into the defective product frame through the slideway 36, which is qualified The chestnuts are transported to the right by the horizontal conveyor belt 4.

As shown in Figure 11, the residual network includes two residual blocks, A1 convolution layer, A2 Batch Norm layer, A3 activation layer, B1 convolution layer, B2 Batch Norm layer and B3 activation layer constitute the first residual block, the A1 convolutional layer, A2 BatchNorm layer, A3 activation layer, B1 convolutional layer, B2 Batch Norm layer and B3 activation layer are connected end to end in turn, and further, the data at the entrance of the A1 convolutional layer can be short-circuited and inserted into the B2 Batch Between the Norm layer and the B3 activation layer; the C1 convolutional layer, the C2 BatchNorm layer, the C3 activation layer, the D1 convolutional layer, the D2 Batch Norm layer and the D3 activation layer constitute the second residual block, the C1 convolutional layer , C2 Batch Norm layer, C3 activation layer, D1 convolution layer, D2 Batch Norm layer and D3 activation layer are connected end to end. The first residual block and the second residual block are connected end to end, the MFCC grayscale image is input into the two residual blocks of the residual network, and then enters the E1 fully connected layer, and the data flows into the F1 The soft regression layer finally outputs the recognition result.

Further, the residual network (Res-Net) is trained offline, and a total of 3160 samples of data are collected for training, of which 1510 are data without bug-eye defects, and 1650 are data with bug-eye defects. The model is stored in the embedded ARM microcontroller of the electronic control module 9.

As shown in Figure 12, the area information of suspected bug eyes is obtained from the detection results of machine vision; the position information of suspected bug eyes is detected by machine vision, and the laser detection results are obtained by laser detection; The surface information near the worm's eye is obtained through acoustic detection to obtain the acoustic detection result; the above-mentioned machine vision results (suspected worm's eye position and size information), laser detection results and acoustic detection results are identified through the SVM model, and the final identification results (normal and insect eyes), and then output by the display sub-module and the light warning sub-module.

The SVM model is trained offline and shares the same training samples with the Residual Network (Res-Net), that is, training on data collected from a total of 3160 samples (1510 of which are data without bug-eye defects). , 1650 are data with bug eye defects), and the parameters of the SVM model after training are stored in the embedded ARM single-chip microcomputer of the electronic control module 9.

This embodiment does not limit the shape, material, structure, etc. of the present invention in any form. Any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention belong to the protection of the technical solution of the present invention. scope.

Claims (8)

1. A chestnut worm eye detection device based on machine vision, laser and acoustics, characterized in that: it comprises a frame (1), and the frame (1) is sequentially provided with a lifting conveyor belt (2) and a horizontal conveyor belt ( 4), the lifting conveyor belt (2) is inclined at a low left and a high right, and the frame (1) is provided with a camera darkroom detection module (3) and a laser detection module located above the horizontal conveyor belt (4) in turn from left to right (5), the acoustic detection module (6) and the sorting device (7), the side of the frame (1) is provided with an electronic control module (9), a chestnut positioning device and a chestnut unlocking device, and the working part of the chestnut positioning device is located in the lifting position. Above the conveyor belt (2), the working part of the chestnut unlocking device is located between the acoustic detection module (6) and the sorting device (7) above the horizontal conveyor belt (4). 2. The chestnut worm's eye detection device based on machine vision, laser and acoustics according to claim 1, characterized in that: both the lifting conveyor belt (2) and the horizontal conveyor belt (4) are made of metal thermal conductor materials, and the lifting conveyor belt (2) and the horizontal conveyor belt (4) are made of The upper surfaces of the belt (2) and the horizontal conveyor belt (4) are evenly provided with grooves (54) for storing the chestnuts to be conveyed at intervals of 5 to 8 cm along the left and right directions. The lifting conveyor belt (2) and the horizontal conveyor belt (4) are combined as A complete conveyor belt. 3. The chestnut worm eye detection device based on machine vision, laser and acoustics according to claim 2, characterized in that: the camera darkroom detection module (3) comprises a cavity box (10), a ring LED light (11) and an industrial The camera (12), the cavity box (10) is open at the bottom and faces the horizontal conveyor belt (4) vertically downward, the left and right sides of the cavity box (10) are fixed on the frame (1) by screws, The left and right sides of the bottom edge of the cavity box (10) are provided with rectangular notches for the horizontal transportation of chestnuts; the ring LED light (11) and the industrial camera (12) are installed in the top center of the cavity box (10) Location. 4. The chestnut worm eye detection device based on machine vision, laser and acoustics according to claim 3, characterized in that: the laser detection module (5) comprises a first X-direction stepping motor (13), a first X-direction moving wire Rod (14), first X-direction bracket (15), first Y-direction stepping motor (16), first Y-direction moving screw (17), first Y-direction guide rail (18), laser detection fixture (19) ) and the laser detection sub-module (20); The X-direction is the left-right horizontal direction, the Y-direction is the front-rear horizontal direction, and a first support rail (55) arranged along the X-direction and in the shape of a rectangular frame is respectively provided on the front and rear sides of the horizontal conveyor belt (4) on the frame (1). ), each first support rail (55) is provided with a first X-direction moving screw (14), and one end of the first X-direction moving screw (14) is rotatably connected to the first support rail ( 55), the other end of the first X-direction moving screw (14) is drivingly connected with the first X-direction stepping motor (13), and the first X-direction moving screw (14) is threadedly connected along the first A support rail (55) slides the first X-direction bracket (15), the front and rear ends of the first Y-direction rail (18) are fixedly connected to the upper ends of the two first X-direction brackets (15), the first The Y-direction guide rail (18) is a hollow rod-shaped structure, the first Y-direction moving screw (17) is arranged inside the first Y-direction guide rail (18) along the front-rear direction, and one end of the first Y-direction moving screw (17) rotates Connected to one end inside the first Y-direction guide rail (18), the other end of the first Y-direction moving screw (17) is drivingly connected to the first Y-direction stepping motor (16), and the laser detection fixture (19) is threaded Connected to the first Y-direction moving screw (17) and slidably arranged along the first Y-direction guide rail (18), the laser detection sub-module (20) is mounted on the laser detection fixture (19) and irradiated toward the groove (54) . 5. The chestnut worm eye detection device based on machine vision, laser and acoustics according to claim 4, characterized in that: the acoustic detection module (6) comprises a second X-direction stepping motor (21), a second X-direction moving wire Rod (22), second X-direction bracket (23), second Y-direction stepper motor (24), second Y-direction moving screw (25), second Y-direction guide rail (26), sliding frame, C axis Motion motor (27), C-axis reducer (28), high-pressure gas needle (29), high-pressure gas pipe (30), acoustic sensor (31) and first solenoid valve (51); X direction is the left and right horizontal direction, Y direction for the front and rear horizontal direction, A second support guide rail (56) arranged along the X direction and in the shape of a rectangular frame is respectively provided on the frame (1) on the front and rear sides of the horizontal conveyor belt (4), and each second support guide rail (56) is provided within each of the second support guide rails (56). There is a second X-direction movement screw (22), one end of the second X-direction movement screw (22) is rotatably connected to the second support guide rail (56), and the second X-direction movement screw (22) The other end is drivingly connected with the second X-direction stepping motor (21), and the second X-direction moving screw (22) is threadedly connected with the second X-direction sliding along the second support rail (56). The front and rear ends of the second Y-direction rails (26) are fixedly connected to the upper ends of the two second X-direction brackets (23), and the second Y-direction rails (26) are hollow rod-shaped structures. The Y-direction moving screw (25) is arranged inside the second Y-direction guide rail (26) in the front-rear direction, and one end of the second Y-direction moving screw (25) is rotatably connected to one end of the second Y-direction guide rail (26). The other end of the two Y-direction moving screw (25) is connected to the second Y-direction stepping motor (24) in a transmission connection, and the C-axis moving motor (27) and the C-axis reducer (28) are arranged inside the sliding frame, The sliding frame is threaded on the second Y-direction motion screw (25) and is slidably arranged along the second Y-direction guide rail (26). The power output end of the C-axis motion motor (27) is connected to the power of the C-axis reducer (28). The input end is connected to the transmission. The high-pressure gas needle (29) is installed on the power take-off shaft of the C-axis reducer (28). Vertical to the power output shaft of the C-axis reducer (28), the air inlet of the high-pressure gas needle (29) is connected to the air outlet of the high-pressure gas pipe (30), and the first solenoid valve (51) is arranged on the high-pressure gas pipe (30) , an air compressor is connected to the air inlet of the high-pressure gas pipe (30), and the acoustic sensor (31) is arranged at the air-jet port of the high-pressure gas needle (29). 6. The chestnut worm eye detection device based on machine vision, laser and acoustics according to claim 5, characterized in that: the sorting device (7) comprises a sorting support plate (32), a sorting cylinder (33), a sorting The rod (34), the sorting rod head (35) and the slideway (36); the slideway (36) is installed on the right front side of the frame (1) by bolting; the sorting support plate (32) is vertically installed by bolting At the rear right position of the frame (1), the sorting cylinder (33) is fixedly installed on the left side of the sorting support plate (32), and the telescopic rod of the sorting cylinder (33) is coaxial with the sorting rod (34). Connection, the sorting rod (34) is arranged horizontally along the left and right directions. 7. The chestnut worm's eye detection device based on machine vision, laser and acoustics according to claim 6, characterized in that: the chestnut positioning device is a cooling air nozzle (37) and a water drip pipe arranged side by side above the lifting conveyor belt (2). , the water drip pipe is located on the left side of the cold air nozzle (37), the water drip hole of the water drip pipe and the nozzle of the cold air nozzle (37) are both set downward and toward the passing groove (54); the cold air nozzle (37) is provided with a second solenoid valve (52); The chestnut unlocking device is a hot air nozzle (38) arranged above the horizontal conveyor belt (4). There is a third solenoid valve (53). 8. The chestnut worm's eye detection device based on machine vision, laser and acoustics according to claim 7, characterized in that: the electronic control module (9) comprises an embedded ARM single-chip microcomputer, an input/output sub-module, and an industrial Ethernet communication sub-module , display sub-module, light warning sub-module; embedded ARM single-chip microcomputer is connected with the input/output sub-module through the on-chip bus; The embedded ARM single-chip microcomputer is respectively connected with the industrial camera (12) and the laser detection sub-module (20) through the industrial Ethernet communication sub-module; the industrial Ethernet communication sub-module has built-in dual industrial Ethernet network ports, and the input/output sub-modules are respectively connected with The first X-direction stepper motor (13), the first Y-direction stepper motor (16), the second X-direction stepper motor (21), the C-axis motion motor (27), the second Y-direction stepper motor (24) ), the sorting cylinder (33) is connected through the control signal line, and the corresponding stepper motor is controlled to act in an orderly manner according to the control process to realize positioning and sorting; the input/output sub-module is also connected with the first solenoid valve (51), The second solenoid valve (52) is connected to the third solenoid valve (53), and according to the action flow, the solenoid valve is opened or closed to realize the functions of detection, freezing and thawing respectively; the input/output sub-module is also connected to and controls the annular The LED light (11) is turned on and off and its brightness; the input/output sub-module is also connected with the acoustic sensor (31).