CN110766702B - Agaricus bisporus grading judgment method - Google Patents

Abstract

The invention discloses a method for grading and judging Agaricus bisporus, which belongs to the field of Agaricus bisporus production. The device includes the following steps: S1, acquiring mushroom image information; S2, extracting the region of interest S3, image segmentation, which includes: S31, converting the image to be extracted into a grayscale image; S32, using the OSTU threshold segmentation algorithm for processing; S33, Perform morphological transformation; S34, perform expansion operation on the morphological transformation map; S35, perform distance transformation; S36, perform fixed threshold binarization with distance as the threshold to determine the foreground image; S37, subtract the background image from the foreground image, extract Image outline; S38, according to the uncertainty region in the markers, the boundary of the original image is finally obtained; S4, according to the acquired foreground image, calculate the number of pixels m of the foreground image; S5, according to the preset rules, according to the mushroom cap area judgment level. The invention can improve the accuracy of grading judgment of Agaricus bisporus, and reduce errors caused by manual grading.

Description

A kind of grading judgment method of Agaricus bisporus

technical field

The invention relates to the field of Agaricus bisporus production, in particular to a method for judging the classification of Agaricus bisporus.

Background technique

Agaricus bisporus is also known as white mushroom, mushroom, and sea mushroom, and producers and operators in European and American countries often call it common cultivated mushroom or button mushroom. Agaricus bisporus is a mushroom that is cultivated and consumed worldwide. It is known as the "World Mushroom" and can be sold fresh, canned, or salted. The mycelium of Agaricus bisporus is also used as a raw material for medicine. China's Agaricus bisporus is most cultivated in Fujian, Shandong, Henan, Zhejiang and other provinces. Cultivation methods include mushroom house cultivation, greenhouse frame cultivation and greenhouse border cultivation. Different regions, different climatic conditions and different seasons can adopt their own cultivation methods. It is widely distributed and is widely cultivated in China. However, with the continuous development of Agaricus bisporus cultivation technology, factory production of Agaricus bisporus has been realized, and uninterrupted production can be realized throughout the year through the environmental control of the mushroom house. The factory production of Agaricus bisporus can precisely control the temperature, humidity, _CO2 concentration and ventilation of the mushroom house, thus providing a very suitable growth environment for Agaricus bisporus. At present, the daily output of large-scale Agaricus bisporus factories can reach hundreds of tons.

When Agaricus bisporus is produced and sold, it is generally necessary to grade Agaricus bisporus according to parameters such as cap size, incompleteness, and browning degree, so as to obtain a higher market price according to market demand. However, at present, manual grading is still used in the factory production of Agaricus bisporus, and manual grading requires high proficiency of workers, and the accuracy of grading by personnel of different working levels is uneven. ,low efficiency.

Contents of the invention

The invention provides a method for grading Agaricus bisporus, which can solve the problems of large errors and low efficiency in grading Agaricus bisporus in the prior art.

A method for judging the classification of Agaricus bisporus, comprising the steps of:

S1, acquiring mushroom image information;

S2, extract the region of interest: according to the shooting environment of the mushroom image, set the parameters and intercept the mushroom image with a single background color to obtain the image to be extracted; calculate the coordinate value of the upper left point of the minimum circumscribing rectangle of the mushroom outer contour, and the minimum circumscribing The width W and height H of the rectangle;

S3, image segmentation, which includes:

S31, converting the image to be extracted into a grayscale image;

S32, adopting the OSTU threshold segmentation algorithm to process to obtain a binarized image;

S33, perform morphological transformation: use a 3*3 matrix as a template to perform a closed operation, expand first, place each pixel x of the image in the center of the template, traverse all other elements covered by the template, and modify the value of the pixel x The maximum value of all pixels, corroding the expanded image, traversing each pixel of the image to modify the pixel to the minimum value in the template, and obtaining the morphological transformation map;

S34, performing an expansion operation on the morphological transformation map to obtain a background image;

S35, perform distance transformation: set the mask size to 3*3, set the RBG value of the foreground picture to (255,255,255), i.e. white; set the RBG value of the background picture to (0,0,0), i.e. black ; Use non-zero pixels as the foreground target, and zero pixels as the background; calculate all pixel distances between the foreground image and the background image, and use the least squares method to replace the distance with pixels to generate a distance transformation map;

S36, using the distance as a threshold to perform fixed threshold binarization to determine the foreground image;

S37, subtracting the background image and the foreground image, determining the uncertain area where the foreground image and the background image overlap, extracting the contour of the image, and obtaining markers;

S38, according to the uncertain region, the boundary of the original image is finally obtained through the watershed change in the markers;

S4, according to the obtained foreground image, calculate the number of pixels m of the foreground image, then the area of the mushroom cap is: m*25.4/d square millimeter, wherein, d is the resolution of the camera;

S5, according to the preset rules, the grade is determined according to the area of the mushroom cap.

More preferably, in S2, the library functions findContours and boundingRect in OpenCV are used to calculate the coordinate value of the upper left point of the minimum circumscribed rectangle of the mushroom, and the width W and height H of the rectangle.

More preferably, in S2, further comprising extending the minimum circumscribed rectangle outward by 13 mm to obtain an extended rectangle.

More preferably, S4 also includes calculating the areas of the front and back sides of the mushroom respectively, and taking the maximum value of the two as the area of the mushroom.

More preferably, it also includes S41, which includes:

S411, extracting feature points: constructing a Hessian matrix, the Hessian matrix H(X, σ) of any pixel point X=(x, y) in the image to be extracted is as follows:

Among them, σ is the scale, Lxx(X,σ), Lxy(X,σ), and Lyy(X,σ) are the second derivatives of the image to be extracted in each direction after Gaussian filtering;

The convolution of the integral image and the box filter is approximately expressed as Dxx, Dxy, Dyy, and the approximate calculation of the Hessian determinant is obtained as:

det(Hessian)＝D _xx D _yy -(λD _xy ) ² (2)

Among them, λ is the weight coefficient, which is used to balance the error caused by using the box filter approximation;

Compare all the pixels processed by the Hessian matrix with the points in the scale space for non-maximum values to find out the points of interest in the image;

Perform linear interpolation in scale space and image space to obtain the final stable feature points;

S412, converting the original image into a grayscale image;

According to the extracted feature points, draw a circle with the feature points as the center;

Take the front threshold of the mushroom as 5mm, and the reverse threshold as 13mm;

If the size of the circle exceeds the threshold, it is considered incomplete, otherwise, it is not incomplete;

S5. According to the preset rules, the grade is determined according to the area and incompleteness of the mushroom cap.

The invention provides a method for grading and judging Agaricus bisporus, which can improve the accuracy of grading and judging Agaricus bisporus, reduce errors caused by manual grading, effectively reduce labor workload, improve efficiency and reduce costs.

Description of drawings

Fig. 1 is the structural representation of Agaricus bisporus classification system;

Fig. 2 is the top view of Fig. 1;

Fig. 3 is the front view of Fig. 1;

Fig. 4 is the structural representation of turning over device in Fig. 1;

Figure 5 is a top view of Figure 4;

Fig. 6 is the front view of Fig. 4;

Fig. 7 is the structural representation of classifying device in Fig. 1;

Figure 8 is the original image acquired by the camera;

Fig. 9 is the original image after interception;

Figure 10 is the image to be extracted;

Figure 11 is an example diagram of the process of image segmentation;

Figure 12 is a flowchart of image segmentation;

Fig. 13 is a flow chart of cap area calculation;

Fig. 14 is a process example diagram of S41;

Fig. 15 is a flowchart of S41;

Figure 16 is an example diagram of the process of browning screening.

Detailed ways

A specific embodiment of the present invention will be described in detail below in conjunction with the accompanying drawings, but it should be understood that the protection scope of the present invention is not limited by the specific embodiment.

As shown in Figures 1 to 3, a kind of Agaricus bisporus grading system provided during the implementation of the present invention includes:

Rack 10;

The centrifugal feeding tray 20 is arranged on the frame 10, wherein the centrifugal feeding tray 20 is a feeding tray in the prior art, and it utilizes centrifugal force to send out the materials sequentially;

The first conveyer belt 11 is fixedly installed on the frame 10, and one end thereof is connected with the centrifugal feeding tray 20 for conveying the mushrooms sent out from the centrifugal feeding tray 20;

The second conveyor belt 12 is fixedly installed on the frame 10 and is located below the first conveyor belt 11;

The transition connecting cylinder 13 is connected to the end of the first conveyor belt 11 and the front end of the second conveyor belt 12, and is used to transfer the mushrooms conveyed on the first conveyor belt 11 to the second conveyor belt 12. The transition connecting cylinder 13 is C-shaped The arc connecting cylinder of the structure has a through cavity, and the cavity is also in a C-shaped structure. When the mushrooms transported by the first conveyor belt 11 move to the end of the first conveyor belt 11, they fall into the transition connecting cylinder 13, and then Under the action of gravity, the mushrooms slide down to the second conveyor belt 12 through the transition connecting cylinder 13;

Image acquisition system, for acquiring the image of the mushroom on the first conveyor belt 11 and determining the grade of the mushroom;

A grading device, used for grading and removing the mushrooms on the second conveyor belt 12;

The control device is used to acquire mushroom grade information and control the action of the grader device.

Specifically, the grading device includes a photoelectric sensor and a sorting grab 40, the photoelectric sensor is arranged on the frame 10, and is used to detect whether there are mushrooms on the second conveyor belt 12, and the photoelectric sensor signal is connected to the control device; the sorting grab 40 includes a sorting Sorting motor 41 and some driving rods 42, several driving rods 42 are arranged evenly along the circumference of the output shaft of sorting motor 41, one end of several driving rods 42 is fixedly connected to the output end of sorting motor 41, and the output shaft of sorting motor 41 There is an included angle A between the axis line and the vertical direction, and the driving lever 42 is used to remove the mushrooms on the second conveyor belt 12 when it rotates. Wherein, the included angle A is 60°, the driving lever 42 and the axis of the output shaft of the sorting motor 41 have an included angle B, and the included angle B is 120°.

Mushrooms enter the first conveyor belt 11 under the conveyance of the centrifugal feeding tray 20, and during the conveying process on the first conveyor belt 11, when passing through the first image acquisition device 30, the first image acquisition device 30 acquires the image information of the mushrooms And send it to the processor, the processor judges the grade of the mushroom according to the preset algorithm and sends it to the control device, the control device receives the signal and controls the grading device. Since the mushrooms are transported in sequence, the mushrooms acquired and judged by the first image acquisition device 30 will be sequentially transported to the second conveyor belt 12, and the mushrooms can be removed according to the grade by the grading device according to the order. Specifically, when the mushroom whose image information is collected by the first image acquisition device 30 and whose grade is judged is transported to the second conveyor belt 12, the photoelectric sensor detects the mushroom and sends a signal to the control device, and the control device, according to the received grade information, Control the action of the grading device. When it is necessary to reject, the control device controls the rotation of the sorting motor 41. The sorting motor 41 drives the driving lever 42 to rotate. The mushrooms are transferred to the recovery device on one side, wherein, in order to better remove the mushrooms, a paddle 43 is also provided at the end of the driving rod 42 away from the sorting motor 41, and the paddle 43 and the driving rod 42 are vertically arranged. The paddle 43 can increase the working area of the paddle 42, making it easier to remove mushrooms. Different from the traditional air pump push away, the grading device in this embodiment has no so-called return stroke, and it will not affect the sorting of mushrooms when the mushrooms are relatively concentrated, so as to avoid missed detection.

As shown in Figures 4 to 6, the system also includes a turning device and a fixed mount 50, the turning device includes an infrared sensor 51, two cylinders 52 oppositely arranged and a baffle plate 53 arranged at the output end of the cylinder 52, and the fixed mount 50 is arranged on the frame 10, two baffles 53 are arranged in a figure-eight shape, and one end with a larger opening of the baffle 53 is fixedly connected to the fixed frame 50, and when the cylinder 52 is in action, it is used to control the movement of the free ends of the two baffles 53 Opening angle, the infrared sensor 51 signal is connected to the control device, and is used to detect whether the mushroom has turned over; wherein, the image acquisition system includes a first image acquisition device 30 and a second image acquisition device 31, and the first image acquisition device 30 is used to collect mushrooms The image information on one side, the second image acquisition device 31 is used to collect the image information on the other side of the mushroom; The device 31 is along the front side in the direction of movement of the first conveyor belt 11 .

Specifically, the image acquisition device includes a dark box arranged on the frame 10, a camera and a light source, the camera and the light source are arranged in the dark box, and the signal of the camera is connected to the control device.

When the mushroom is transported on the first conveyor belt 11, it first passes through the first image acquisition device 30, and when it moves to the first image acquisition device 30, the first image acquisition device 30 takes pictures of the information on one side of the mushroom and judges the grade of the mushroom, and then The mushroom continues to move under the conveyance of the first conveyor belt 11. When moving to the turning device, because the size of the mushroom is basically in an interval, the distance between the smaller opening ends of the two baffle plates 53 is set to be smaller than the outer edge of the mushroom. The diameter is small, and the distance between the larger opening end is larger than the outer diameter of the mushroom, so that the mushroom can smoothly enter between the two baffles 53, and then when it moves to a position close to the smaller opening end, the mushroom part contacts the baffle 53 and is subjected to resistance , due to the special shape of the mushroom, its upper part contacts the baffle 53, but the bottom part does not touch the baffle 53, and at the same time, the bottom also receives the friction force of the first conveyor belt 11, so an oblique upward resultant force will be formed on the mushroom , prompting the mushroom to turn over, because the height of the mushroom is generally smaller than the outer diameter, so when the mushroom is turned over, as shown in Figure 6, the infrared sensor 51 will not detect the mushroom under normal transportation, and when the mushroom is turned over, in the vertical direction When it becomes higher, it will enter the detection range of the infrared sensor 51. The infrared sensor 51 detects that the mushroom has turned over and sends a signal to the control device. The control device controls the retraction of the cylinder 52 and drives the baffle 53 to open. In order to prevent that the baffle plate 53 has been opened before the mushroom is fully turned over, causing the mushroom to turn back again, a delay operation can be set for the retraction stroke of the cylinder 52, such as after the infrared sensor 51 detects that the mushroom is turned over, it sends a signal to the control device, The control device sets the delay operation, and after 1-2S, the cylinder 52 is controlled to retract. The overturned mushrooms enter the second image acquisition device 31 under the transport of the first conveyor belt 11 to acquire the image information of the other side and further judge the grade.

Embodiment one:

When making a specific judgment, the system uses the size of Agaricus bisporus as a characteristic parameter, and divides it into four grades: A grade, B grade, C grade and D grade.

Grading judgments include the following steps:

S1, acquiring mushroom image information through a camera;

S2, extract the region of interest: according to the shooting environment of the mushroom image, set the parameters and intercept the mushroom image with a single background color to obtain the image to be extracted; calculate the coordinate value of the upper left point of the minimum circumscribing rectangle of the mushroom outer contour, and the minimum circumscribing The width W and height H of the rectangle.

Specifically, in this embodiment, the camera resolution is 96 as an example. The mushroom image acquired by the system includes the conveyor belt and its edge. In order to accurately obtain the characteristic parameters of Agaricus bisporus, it is necessary to extract the region of interest from the mushroom image. First, the In Figure 8 captured by the camera, 158 pixels are intercepted on the left and 150 pixels are intercepted on the right, and the aluminum material on the side of the conveyor belt is removed to obtain the image shown in Figure 9. The specific intercepted pixel values depend on the width of the conveyor belt and the viewfinder size of the camera To specifically set, the final goal is to obtain an image of a single background color; then use the library function findContours and boundingRect in OpenCV to set specific parameters for the picture in Figure 9 to calculate the upper left point position of the smallest circumscribed rectangle of the mushroom ( x, y) and the width w and height h of the rectangle, and then extend the rectangle outward by 50 pixels to obtain the position of the rectangle: length: x-50~x+w+50, width: y-50~y+ h+50, the image shown in Figure 10 is obtained according to the position of the rectangle.

The region of interest extraction separates the image of Agaricus bisporus from the whole image, removes the interference, and obtains the image to be extracted, which is used for subsequent image processing.

As shown in Figure 11, S3, image segmentation, which includes:

S31, converting the image to be extracted into a grayscale image;

S32, adopting the OSTU threshold segmentation algorithm to process to obtain a binarized image;

S33, perform morphological transformation: use a 3*3 matrix as a template to perform a closed operation, expand first, place each pixel x of the image in the center of the template, traverse all other elements covered by the template, and modify the value of the pixel x The maximum value of all pixels, corroding the expanded image, traversing each pixel of the image to modify the pixel to the minimum value in the template, and obtaining the morphological transformation map;

S34, performing an expansion operation on the morphological transformation map to obtain a background image;

S35, perform distance transformation: set the mask size to 3*3, set the RBG value of the foreground picture to (255,255,255), i.e. white; set the RBG value of the background picture to (0,0,0), i.e. black ; Use non-zero pixels as the foreground target, and zero pixels as the background; calculate all pixel distances between the foreground image and the background image, and use the least squares method to replace the distance with pixels to generate a distance transformation map;

S36, using the distance as a threshold to perform fixed threshold binarization to determine the foreground image;

S37, subtracting the background image and the foreground image, determining the uncertain area where the foreground image and the background image overlap, extracting the contour of the image, and obtaining markers;

S38, according to the uncertain region, the boundary of the original image is finally obtained through the watershed change in the markers;

S4, according to the obtained foreground image, calculate the number of pixels m of the foreground image, then the area of the mushroom cap is: m*25.4/d square millimeter, wherein, d is the resolution of the camera;

Specifically, as shown in Figure 12, first convert the image to be extracted, that is, (a) in Figure 11 into the grayscale image (b) in Figure 11, and use the OSTU threshold segmentation algorithm to obtain the binarization in Figure 11 Image (c). Then perform morphological transformation, use the 3*3 matrix as the template to perform the closing operation, expand first, place each pixel x of the image in the center of the template, traverse all other elements covered by the template, and modify the value of the pixel x to all The largest value in the pixel, and then corrode the expanded image, and traverse each pixel of the image to modify the pixel to the minimum value in the template. As a result, small black holes and cracks are eliminated, and noise is removed to obtain the image in Figure 11 (d ), the expansion operation is performed on the morphological transformation map obtained by the morphological transformation, and the background image (e) in Figure 11 is obtained. Then perform distance transformation, set the mask size to 3*3, set the RGB value of the foreground image to (255,255,255), which is white, and the background RGB value to (0,0,0), which is black, so non-zero pixels are the foreground The target, the zero pixel point is the background, the farther the pixel in the foreground target is from the background, the greater the distance, using the least squares method, replace the pixel with this distance, generate the image (f) in Figure 11, and use these distances as the threshold Perform fixed threshold binarization to determine the foreground image (g) in Figure 11, and subtract the background image (e) in Figure 11 from the foreground image (g) in Figure 11 to determine the uncertain area where the foreground and background overlap, Extract image contours to get markers. The boundary (h) of the image to be extracted in Figure 11 is finally obtained through the watershed change in the markers according to the uncertain region. The algorithm flow chart is shown in Figure 12.

S5, according to the preset rules, the grade is determined according to the area of the mushroom cap.

Specifically, as shown in Figure 13, according to the Agaricus bisporus industry standard NY/T1790-2009, the system selects the cap area of Agaricus bisporus as the characteristic parameter of size classification, and divides it into three grades of "large, medium and small" (diameter> 45mm is the first grade, 25≤diameter≤45mm is the second grade, and the diameter <25mm is the third grade). On this basis, this study defines the diameter <10mm as the fourth grade. 1 inch=2.54 cm, pixel height/resolution=image height size, pixel width/resolution=image width size The resolution of the camera selected by the device is 96, and the pixels of the foreground image are calculated according to the foreground image obtained in the preceding steps A few m, then the area of the mushroom is: m*25.4/96 square millimeters. Calculate the area of the front and back of the mushroom separately. Since the handle of the mushroom may cause the mushroom to appear sideways, resulting in a decrease in the front area, the maximum value of the two is taken as the area of the mushroom.

Embodiment two:

The SURF (Speeded-up Robust Features) algorithm is an improved algorithm based on the SIFT (Scale-invariant feature transform) algorithm, which was proposed by Herbert Bay at the European International Conference on Computer Vision in 2006. The algorithm does not depend on pixel values and is occluded, angled etc. The shooting effect is less affected, and it has the characteristics of fast calculation speed and high stability. The SURF algorithm mainly includes two parts: feature point extraction and feature point description.

On the basis of Embodiment 1, as shown in Figure 14 and Figure 15, this embodiment also includes S41, which includes:

S411, extracting feature points: constructing a Hessian matrix, the Hessian matrix H(X, σ) of any pixel point X=(x, y) in the image to be extracted is as follows:

Among them, σ is the scale, Lxx(X,σ), Lxy(X,σ), and Lyy(X,σ) are the second derivatives of the image to be extracted in each direction after Gaussian filtering;

The convolution of the integral image and the box filter is approximately expressed as Dxx, Dxy, Dyy, and the approximate calculation of the Hessian determinant is obtained as:

det(Hessian)＝D _xx D _yy -(λD _xy ) ² (2)

Among them, λ is the weight coefficient, which is used to balance the error caused by using the box filter approximation;

Compare all the pixels processed by the Hessian matrix with the points in the scale space for non-maximum values to find out the points of interest in the image;

Perform linear interpolation in scale space and image space to obtain the final stable feature points;

S412, converting the original image into a grayscale image;

According to the extracted feature points, draw a circle with the feature points as the center;

The front threshold of the mushroom is 5mm, and the back threshold is 13mm; in this embodiment, the front threshold is 22 pixels, and the back threshold is 50 pixels. Due to the presence of the reverse stalk of the mushroom, the negative threshold is greater than the positive threshold.

If the size of the circle exceeds the threshold, it is considered incomplete, otherwise, it is not incomplete;

Among them, FIG. 14 shows several image examples during specific implementation.

S5. According to the preset rules, the grade is determined according to the area and incompleteness of the mushroom cap.

Embodiment three:

On the basis of embodiment one or two, the step of screening according to browning is also included. As shown in Figure 16, the L value in the Lab format image can better reflect the brightness of the mushroom, thereby reflecting the degree of mushroom browning, so the RGB format image acquired by the camera is converted into a Lab format image. L=0 means black, L=100 means white, large L value means white, small L value means black; L value 86 and above is good quality mushroom, L value between 80-85 is better Mushrooms, L value between 70-79 are poor mushrooms, and mushrooms with L value lower than 69 have no edible value, corresponding to four grades of 1, 2, 3, and 4 respectively. Traverse each pixel of the cap, count the number of pixels corresponding to each level, and calculate the number of pixels corresponding to the four levels of 1, 2, 3, and 4 according to the number of pixels of the cap obtained by the watershed algorithm The ratio of the number to the total number of cap pixels is R1, R2, R3, R4. After a large number of experiments, it is determined that R1≥0.65 is grade 1, 0.58≤R2<0.65 is grade 2, 0.53≤R3<0.58 is grade 3, and R4<0.53 for 4 grades.

Currently, prototypes are being tested in the laboratory. 100 mushrooms were picked from the edible fungus base of Shenyang Agricultural University, and transported directly to the laboratory after harvesting. The prototype was used for testing and compared with the manual grading results, and the manual grading results were used as the standard. Manual identification uses a vernier caliper to measure the cap diameter of Agaricus bisporus as a size parameter, and the browning and incomplete characteristic parameters are identified by relevant industry experts by visually observing the appearance of the mushroom. The test results are shown in Table 1.

Table 1 Classification test results

Table 2 Result of Agaricus bisporus grading

It can be seen from Table 1 that the average correct rate of using the Agaricus bisporus automatic grading system is about 96.45%. Among them, the detection errors are mainly due to the incompleteness and browning appearing on the stalk or side of the mushroom, which cannot be captured by the camera. The test results showed that the grading method was effective for the size, browning and incomplete detection of Agaricus bisporus.

The above disclosures are only a few specific embodiments of the present invention, however, the embodiments of the present invention are not limited thereto, and any changes conceivable by those skilled in the art shall fall within the protection scope of the present invention.

Claims (5)

1. a Agaricus bisporus grading system, is characterized in that, described system comprises: frame; Centrifugal feeding tray, set on the frame, is used to send out the materials sequentially by centrifugal force; The first conveyor belt is fixedly installed on the frame, and one end thereof is connected with the centrifugal feeding tray, and is used for conveying the mushrooms sent from the centrifugal feeding tray; The second conveyor belt is fixedly installed on the frame and is located below the first conveyor belt; The transition connecting cylinder is connected to the end of the first conveyor belt and the front end of the second conveyor belt, and is used to transfer the mushrooms conveyed on the first conveyor belt to the second conveyor belt. The transition connecting cylinder is a circular arc connecting cylinder with a C-shaped structure , which has a through cavity, and the cavity also has a C-shaped structure. When the mushrooms transported by the first conveyor belt move to the end of the first conveyor belt, they fall into the transition tube. Under the action of gravity, the mushrooms slide through the transition tube. onto the second conveyor belt; The image acquisition system is used to acquire the image of the mushrooms on the first conveyor belt and determine the grade of the mushrooms by using the Agaricus bisporus grading judgment method; A grading device is used for grading and removing the mushrooms on the second conveyor belt; The control device is used to obtain mushroom grade information and control the action of the grader device; Specifically, the grading device includes a photoelectric sensor and a sorting grab. The photoelectric sensor is arranged on the frame to detect whether there are mushrooms on the second conveyor belt. The signal of the photoelectric sensor is connected to the control device; the sorting grab includes a sorting motor and several Driving rods, several driving rods are evenly arranged along the circumference of the output shaft of the sorting motor, one end of several driving rods is fixedly connected to the output end of the sorting motor, and there is a clamp between the axis line of the output shaft of the sorting motor and the vertical direction Angle, used to remove the mushrooms on the second conveyor belt when the driving lever is rotated; a paddle is also provided at the end of the driving rod away from the sorting motor, and the paddle and the driving rod are vertically arranged; The system also includes a turning device and a fixing frame. The turning device includes an infrared sensor, two opposite cylinders and a baffle set at the output end of the cylinder. The fixing frame is set on the frame, and the two baffles are arranged in a figure-eight shape. , the end with the larger opening of the baffle is fixedly connected to the fixed frame. When the cylinder is in motion, it is used to control the opening angle of the free ends of the two baffles. The signal of the infrared sensor is connected to the control device to detect whether the mushroom has turned over; among them, The image acquisition system includes a first image acquisition device and a second image acquisition device, the first image acquisition device is used to collect image information on one side of the mushroom, the second image acquisition device is used to collect image information on the other side of the mushroom, and the first image acquisition device The image information flipping device for obtaining mushrooms is located on the rear side of the first image acquisition device along the movement direction of the first conveyor belt, and is located on the front side of the second image acquisition device along the movement direction of the first conveyor belt; The method for judging the classification of Agaricus bisporus includes the following steps: S1, acquiring mushroom image information; S2, extract the region of interest: according to the shooting environment of the mushroom image, set the parameters and intercept the mushroom image with a single background color to obtain the image to be extracted; calculate the coordinate value of the upper left point of the minimum circumscribing rectangle of the mushroom outer contour, and the minimum circumscribing The width W and height H of the rectangle; S3, image segmentation, which includes: S31, converting the image to be extracted into a grayscale image; S32, adopting the OSTU threshold segmentation algorithm to process, and obtain the binarized image; S33, perform morphological transformation: use a 3*3 matrix as a template to perform a closed operation, expand first, place each pixel x of the image in the center of the template, traverse all other pixels covered by the template, and modify the value of the pixel x The maximum value of all pixels, corroding the expanded image, traversing each pixel of the image to modify the pixel to the minimum value in the template, and obtaining the morphological transformation map; S34, performing an expansion operation on the morphological transformation map to obtain a background image; S35, perform distance transformation: set the mask size to 3*3, set the RBG value of the foreground picture to (255,255,255), i.e. white; set the RBG value of the background picture to (0,0,0), i.e. black ; Use non-zero pixels as the foreground target, and zero pixels as the background; calculate all pixel distances between the foreground image and the background image, and use the least squares method to replace the distance with pixels to generate a distance transformation map; S36, using the distance as a threshold to perform fixed threshold binarization to determine the foreground image; S37, subtracting the background image and the foreground image, determining the uncertain area where the foreground image and the background image overlap, extracting the contour of the image, and obtaining markers; S38, according to the uncertain region, the boundary of the original image is finally obtained through the watershed change in the markers; S4, according to the obtained foreground image, calculate the number of pixels m of the foreground image, then the area of the mushroom cap is: m*25.4/d square millimeter, wherein, d is the resolution of the camera; S5, according to the preset rules, the grade is determined according to the area of the mushroom cap. 2. a kind of Agaricus bisporus grading system as claimed in claim 1, is characterized in that, in S2, adopts library function findContours and boundingRect among the OpenCV to calculate the coordinate value of the upper left point of the minimum circumscribed rectangle of mushroom, and rectangle The width W and height H. 3. A kind of Agaricus bisporus grading system as claimed in claim 1, is characterized in that, in S2, also comprises outwardly extending 13mm to described minimum circumscribed rectangle, obtains expanded rectangle. 4. A kind of Agaricus bisporus grading system as claimed in claim 1, is characterized in that, also includes calculating the area of the front and back side of mushroom respectively in S4, gets both maximum value as the area of mushroom. 5. A kind of Agaricus bisporus grading system as claimed in claim 1, is characterized in that, also comprises S41, and it comprises: S411, extracting feature points: constructing a Hessian matrix, the Hessian matrix H(X, σ) of any pixel point X=(x, y) in the image to be extracted is as follows: Among them, σ is the scale, Lxx(X,σ), Lxy(X,σ), and Lyy(X,σ) are the second derivatives of the image to be extracted in each direction after Gaussian filtering; The convolution of the integral image and the box filter is approximately expressed as Dxx, Dxy, Dyy, and the approximate calculation of the Hessian determinant is obtained as: det(Hessian)＝D _xx D _yy -(λD _xy ) ² (2) Among them, λ is the weight coefficient, which is used to balance the error caused by using the box filter approximation; Compare all the pixels processed by the Hessian matrix with the points in the scale space for non-maximum values to find out the points of interest in the image; Perform linear interpolation in scale space and image space to obtain the final stable feature points; S412, converting the original image into a grayscale image; According to the extracted feature points, draw a circle with the feature points as the center; Take the front threshold of the mushroom as 5mm, and the reverse threshold as 13mm; If the size of the circle exceeds the threshold, it is considered incomplete, otherwise, it is not incomplete; S5. According to the preset rules, the grade is determined according to the area and incompleteness of the mushroom cap.