Grain particle material full-automatic crushing and detecting method and system

Abstract

The invention discloses a full-automatic crushing and detecting method and system for grain particle materials, and relates to the technical field of crushing and detecting. The method comprises the following steps that 1, whether first unprocessed grains are unhulled rice or not is judged, if the first unprocessed grains are not the unhulled rice, the next step is executed, otherwise, if the first unprocessed grains are the unhulled rice, the step 3 is executed, 2, volume fixing treatment is conducted on the first unprocessed grains, the step 4 is executed, 3, volume fixing treatment is conducted on the unhulled rice, the unhulled rice is hulled through a rice huller module, and brown rice is obtained; step 4, judging whether the moisture content of unhulled brown rice exceeds 15% or not, if the moisture content does not exceed 15%, skipping to step 7, if the moisture content exceeds 15%, executing the next step, and if the moisture content does not exceed 15%, executing step 4, and if the moisture content does not exceed 15%, executing step 5; according to the invention, automation and unmanned operation of the whole process of crushing, quantitative weighing, liquid adding, oscillation extraction, separation, pipetting, dilution, incubation and physicochemical index detection of a sample can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of crushing detection, and in particular to a method and system for fully automatic crushing and detection of grain particles. Background Art

[0002] The content of mycotoxins and heavy metals in grains needs to be tested during the grain collection and storage process. The detection of mycotoxins and heavy metals has national and industry standards, both of which require that the raw grains must first be crushed and mixed, then weighed to a specified weight, extracted with extraction solution, and diluted with diluent; this series of steps, including card counting, incubation, and testing, is currently mainly done manually.

[0003] The existing crushing method process cannot achieve full-process autonomous adjustment and detection, and requires a large amount of manpower to intervene in the crushing and detection process, especially the identification and screening of grain varieties, which all rely on manual selection. On the one hand, the detection process is inefficient, and on the other hand, manual detection is easily affected by subjective factors, thereby affecting the test results. Therefore, using machine vision to improve the fully automatic grain crushing detection process is a technical problem that technical personnel in this field need to solve. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides a fully automatic crushing and detection method and system for grain particles. In view of the current situation where the detection of physical and chemical indicators of grain procurement is mainly done manually, the present invention uses a third-party fungal toxin and heavy metal detection module and a robotic arm module for transition, successfully realizing fully automatic operation from receiving raw grain to outputting results, thus solving the problems raised in the above-mentioned background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a method for fully automatic crushing and detection of grain particles, comprising the following steps:

[0006] Step 1: The grain type, sample number, test items, and moisture content of the first raw grain are issued, and it is determined whether the first raw grain is rice. If the first raw grain is not rice, the process proceeds to step 2. Conversely, if the first raw grain is rice, the process jumps to step 3.

[0007] Step 2: Fix the volume of the first raw grain in a fixed-volume container, discard the excess first raw grain, and jump to step 4;

[0008] Step 3: The rice is subjected to a constant volume treatment, and the rice hulling machine module hulls the rice to obtain brown rice. The rice hulling machine module uses an image recognition algorithm to determine the hulling efficiency. The rice hulling machine module automatically adjusts the gap to adapt to different types of rice hulling;

[0009] Step 4: Determine whether the moisture content of the shelled brown rice exceeds 15%. If the moisture content does not exceed 15%, jump to step 7. If the moisture content exceeds 15%, proceed to step 5.

[0010] Step 5: Microwave heating and drying the first raw grain or brown rice to obtain the second raw grain. Specifically, use the mechanical hand module to transfer the first raw grain or brown rice to the microwave heating link for heating and dehydration. The microwave heating and moisture measurement module is used to detect the moisture content of the first raw grain or brown rice. If the moisture content exceeds the set moisture threshold, transfer it to the microwave heating unit for microwave heating and drying through the mechanical hand module, so that the moisture content of the second raw grain after heating and drying is lower than the set moisture threshold. Then transfer it to the crushing module through the mechanical hand module.

[0011] Step 6: Determine whether the moisture content of the second raw grain exceeds 15% again. If the moisture content does not exceed 15%, proceed to step 7. If the moisture content exceeds 15%, re-label the second raw grain as the first raw grain and jump to step 5.

[0012] Step 7: The mechanical hand module pours the brown rice or the second raw grain into the crushing module, which is specifically a crusher. The second raw grain or brown rice is crushed to obtain a powder sample. The brown rice or the second raw grain is classified and crushed, and the powder sample is sieved by airflow. After the powder sample is transferred from the crushing module to the weighing and feeding module by the mechanical hand module, the mechanical hand module uses the mounted spray head to clean the upper cover and the cavity. The crushing module is used to crush the brown rice or the second raw grain. After crushing is completed, the crushed powder sample is transferred to the weighing and feeding module by the mechanical hand module.

[0013] Step 8: The mechanical hand module transfers about 15 grams to the weighing and feeding module for weighing and feeding. The amount of powder sample entering the subsequent operation is accurately controlled. The weighing and feeding module is used to weigh the crushed powder sample, and then enters the feeding station. The mechanical hand module takes the large centrifuge tube with the opened cap to the weighing and feeding area. The weighing and feeding module accurately weighs the corresponding grams into the large centrifuge tube. After weighing is completed, clean the residual pipeline.

[0014] Step 9: The mechanical hand module grabs the large centrifuge tube consumable to the bottle cap opening station. The mechanical hand module performs the bottle cap opening operation. The mechanical hand module takes the large centrifuge tube with the opened cap to the weighing and feeding module. The weighing and feeding module weighs the large centrifuge tube. The mechanical hand module grabs the weighed large centrifuge tube to the bottle cap opening station. The bottle cap opening and closing module performs the bottle cap closing operation. Then the mechanical hand module grabs the large centrifuge tube with the closed cap to the detection module.

[0015] Step 10: Accurately weigh 5 grams into the large centrifuge tube. Weigh and feed. Re-calibrate to ensure weighing accuracy. Clean the residual pipeline after weighing is completed.

[0016] Step 11: The mechanical hand module grabs the weighed large centrifugal tube to the bottle cap opening station, and performs the bottle cap closing operation after weighing;

[0017] Step 12: The mechanical hand module grabs the large centrifugal tube with closed bottle cap and transfers it to the detection module to start detection;

[0018] Step 13: The detection items are configured, including setting the powdered sample, detection number and detection item content, querying and reporting the results, and the detection module is used for corresponding health index detection of the detected sample in the large centrifugal tube.

[0019] Further, the hulling machine module adopts an image recognition algorithm to judge the hulling efficiency, and the specific steps are as follows:

[0020] Step 31: The hulling machine module completes the initial hulling of the rice to obtain brown rice, and the brown rice after hulling enters the screening tray through the discharge port of the hulling machine module;

[0021] Step 32: The processing unit sends instructions to the vibration module and the airflow module respectively, the airflow module works to generate airflow to blow to the screening tray, and the rice that has not completed hulling will carry the rice husk into the screening tray. Because the weight of the rice husk is much smaller than the weight of the rice itself, when the airflow module blows to the surface of the vibrating screening tray, the rice husk will deviate from the vibration direction under the influence of the airflow, which is convenient for the visual module to find;

[0022] Step 33: A plurality of image sensors are arranged above the screening tray, each image sensor corresponds to a different recognition angle and focal length, and the intersection points of the plurality of image sensors are located on the inner surface of the screening tray. By setting a plurality of image sensors with different recognition angles and focal lengths to form a sensor array, high-definition image acquisition of the screening tray can be realized by using image fusion technology, which provides image data support for subsequent machine vision algorithms and helps to improve the accuracy of image recognition algorithms. The processing unit sets the acquisition frequency;

[0023] Step 34: The image sensor acquires image information according to the acquisition frequency and transmits it to the processing unit, and the processing unit executes the image fusion program to obtain a frame image;

[0024] Step 35: The processing unit uses the optical flow method to compare the front and rear frame images in time sequence, the gray value of the same pixel point in the adjacent frame images before and after is unchanged, and the formula is obtained by Taylor expansion and ignoring high-order terms , x is the initial x direction in the frame image, y is the initial y direction in the frame image, and t is the initial time in the frame image. The initial x direction, the initial y direction and the initial time are located in the previous frame of the compared frame images, wherein , , u and v are the velocity components of the same pixel point in the x direction and the y direction respectively, Δx is the displacement distance of the pixel point in the x direction of the front and back frame images, Δy is the displacement distance of the pixel point in the y direction of the front and back frame images, and Δt is the interval time between the front and back frame images s, Ix is the x direction value in the frame image spatial gradient, Iy is the y direction value in the frame image spatial gradient, and It is the value of the frame image time gradient, the processing unit pre-processes the frame image, performs Gaussian blur noise reduction on the frame image, is used to reduce the interference of high frequency noise on gradient calculation, normalizes the gray value of all pixel points in the frame image, and ensures consistency under different illumination conditions;

[0025] Step 36: The processing unit calculates the spatial gradient (Ix, Iy) and the time gradient It of the frame image respectively;

[0026] Step 37: For each pixel point (x, y) in the frame image, all points in the adjacent region window of the pixel point are extracted, and an equation group is constructed:

[0027] , wherein , The matrix A is a motion constraint set of all pixels in the adjacent region, each row represents the contribution of a pixel to the optical flow equation, and each row in the vector b is the time gradient of each pixel point in the adjacent region. The velocity vector is solved by the least square method ATA is the covariance matrix of the spatial gradient of the adjacent region, reflecting the distribution of the gradient, and ATb is the correlation between the time gradient and the time change, which determines the direction of the optical flow. The processing unit analyzes the displacement characteristics of each pixel point in the frame image, calculates the velocity amplitude Vmag of the pixel point according to the formula , calculates the motion direction θ of the pixel point according to the formula , and calculates the trajectory length L of the pixel point i in the frame image according to the formula N is the number of frame images. When the optical flow method is used to calculate the displacement of a single frame image, it is easy to be disturbed by noise, illumination change and other factors, resulting in large fluctuation of single frame displacement data. By accumulating the displacement of consecutive N frames, the influence of short-term noise can be smoothed, and more stable motion characteristics can be extracted.

[0028] Step 38: The processing unit sets the velocity threshold Vth, the direction threshold θth and the length threshold Lth, and sets the trigger condition one, the trigger condition two and the trigger condition three. The trigger condition one is that the velocity amplitude of the pixel point in the frame image is greater than the velocity threshold Vth, the trigger condition two is that the motion direction θ of the pixel point in the frame image is greater than the direction threshold θth, and the trigger condition three is that the trajectory length L of the pixel point in the frame image is less than the length threshold Lth. When any two of the trigger condition one, the trigger condition two and the trigger condition three are met, the processing unit marks the corresponding pixel point as a rice hull.

[0029] Step 39: The processing unit presets an adjustment threshold, and the processing unit calculates in real time the proportion of rice husks in rice in the frame image. When the calculated proportion is greater than the adjustment threshold, the processing unit controls the rice huller module to increase the extrusion friction force by a fixed proportion. When the extrusion friction force of the rice huller module is increased to the maximum, it no longer increases. The processing unit presets an adjustment cycle. If the proportion is never greater than the adjustment threshold during the adjustment cycle, the processing unit controls the rice huller module to reduce the extrusion friction force by half of the fixed proportion. After husking is completed, brown rice is obtained, which is used to improve the husking efficiency of the rice huller and avoid excessive rice husks affecting subsequent processing steps.

[0030] Furthermore, the image fusion program specifically includes the following steps:

[0031] Step 341: The processing unit fuses the image information of multiple recognition angles at the same focal length, identifies feature points in each image information based on the SIFT recognition algorithm, and fuses the image information based on the auxiliary graphic information of the recognition angle. After the image information at the same focal length is fused, a focal length map is obtained.

[0032] Step 342: The processing unit repeats step 341 until all image information is fused into focal length maps of different focal lengths, arranges the focal length maps in ascending order of focal length to obtain focal length queues, and counts the number of focal length queues p;

[0033] Step 343: The processing unit divides each focal length map into Specifically, the length and width of the rectangular focal length image are divided into p equal parts, the image block is converted into a grayscale image, the grayscale value difference between adjacent pixels in the image block is calculated and accumulated to obtain the difference value, and the image blocks are arranged in descending order of the difference value to obtain the difference queue of the focal length image;

[0034] Step 344: The processing unit extracts the frontmost value in the difference queue of each focal length map. The processing unit places the extracted image blocks according to their positions in the focal length map to obtain a spliced ​​image;

[0035] Step 345: The processing unit scans the mosaic image. The extracted tiles may not be able to completely piece together the frame image. Tiles at the same position in the focal image may overlap or be missing. If tiles at the same position in the mosaic image overlap, the processing unit compares the difference between the overlapping tiles and retains the tile with the largest difference. If tiles are missing in the mosaic image, the processing unit extracts the tile with the largest difference from the corresponding position in all focal images and adds it to the mosaic image.

[0036] Step 346: The processing unit repeatedly executes step 345 until there are no overlapping blocks or missing blocks in the mosaic image, and then the processing unit marks the complete mosaic image as a frame image.

[0037] Further, the spatial gradient (Ix, Iy) calculation of the frame image specifically includes the following steps:

[0038] The spatial gradient (Ix, Iy) is calculated by a Sobel operator or a Scharr operator, both of which are discrete differential operators that calculate the spatial gradient of an image by convolution. The Sobel operator is a 3x3 kernel that is robust to noise in combination with Gaussian smoothing, and the Scharr operator is also a 3x3 kernel with larger coefficients that is more sensitive to low-contrast edges.

[0039] The Sobel operator calculation steps are as follows:

[0040] Step 361:

[0041] The x-direction kernel of the horizontal gradient, with higher center column weights to highlight horizontal edges.

[0042] The y-direction kernel of the vertical gradient, with higher center row weights to highlight vertical edges.

[0043] Step 362: Pad the frame image with blank or mirrored pixels to increase the frame image size and avoid size reduction after convolution.

[0044] Step 363: For each pixel (g, k) in the frame image, take the sum of the multiplication of the 3x3 adjacent region, with the specific formula as follows:

[0045] ,

[0046] where Igray is the gray value of the corresponding pixel point when the frame image is converted to a grayscale image.

[0047] Step 364: Normalize Ix and Iy to [0, 255] with the formula , The convolution result may be negative, so normalization is needed.

[0048] Further, the Scharr operator calculation steps are as follows:

[0049] The x-direction kernel of the horizontal gradient: The y-direction kernel of the vertical gradient: with larger center coefficients for more sensitive weak edges.

[0050] The remaining Scharr operator calculation steps are consistent with the Sobel operator, except for the kernel.

[0051] Further, the time gradient It calculation of the frame image specifically includes the following steps:

[0052] Step 365: Calculate the time gradient It using the difference method, the formula is as follows:

[0053] For each pixel point (e, o) in the frame image, subtract the gray value of the corresponding pixel point in the next frame image from the gray value of the corresponding pixel point in the current frame image;

[0054] Step 366: Smooth the obtained time gradient It using Gaussian filtering. Due to image noise and discontinuity of object motion, the directly calculated time gradient may contain noise and outliers. In order to make the calculation result more stable and reliable, the calculated time gradient can be smoothed.

[0055] Further, the image recognition algorithm is used to judge the crushing effect in the hierarchical crushing, and the specific steps are as follows:

[0056] Step 71: The crushing is divided into coarse crushing stage, medium crushing stage and fine crushing stage, the crushing range of the coarse crushing stage of the brown rice or the second raw grain is set to 10-20 mm, the crushing range of the medium crushing stage is 3-5 mm, and the crushing threshold size of the fine crushing stage is 20 mesh;

[0057] Step 72: The processing unit controls the crushing module to crush the brown rice or the second raw grain, and the processing unit obtains image information from the vision module to identify whether the diameter of 95% of the brown rice or the second raw grain after crushing meets the range set in the corresponding stage. If it meets the range, it enters the next stage, otherwise, if it does not meet the set range, it repeatedly executes the current stage, and the hierarchical crushing is completed when 98% of the brown rice or the second raw grain is smaller than 20 mesh in the fine crushing stage;

[0058] Step 73: After the brown rice or the second raw grain completes the fine crushing stage, a powder sample is obtained, air flow screening is used, a 20 mesh screen is set, the screen is perpendicular to the blowing direction of the air flow module, the processing unit sends instructions to the air flow module, the air flow module blows air flow to the crushed powder sample, and the processing unit adjusts the air volume and air speed of the air flow module, the air volume is controlled at 500-1000 cubic meters / hour, and the air speed is controlled at 10-15 meters / second, so that the air flow can lift the powder sample and perform screening.

[0059] Further, the vision module judges the crushing efficiency of the crushing module, which includes the following steps:

[0060] Step 76: The processing unit obtains the focal length f, pixel size s and erection height H of each image sensor in the vision module, the processing unit obtains the image information of each image sensor, randomly selects one of the powder samples and measures the imaging size , the imaging size Specifically, the number of pixels occupied by the particles of the powder sample;

[0061] Step 77: The processing unit calculates the size of the selected powdered sample particles according to the formula The size of the selected powdered sample particles is calculated and recorded;

[0062] Step 78: Repeat step 77 to continuously calculate the proportion of the number of powdered sample particles within the specified range, and recalculate the proportion every time step 77 is repeated.

[0063] The full-automatic crushing and detection system for grain particle materials includes a constant volume module, a huller module, a vision module, an air flow module, a vibration module, a microwave heating and moisture measurement module, a crushing module, a large centrifuge tube management module, a weighing and feeding module, an open and close bottle cap module, a detection module, and a mechanical hand module. The output ends of the vision module, the weighing and feeding module, the open and close bottle cap module, and the detection module are connected with the input end of the processing unit. The output end of the processing unit is connected with the input ends of the constant volume module, the microwave heating and moisture detection module, the air flow module, the crushing module, the large centrifuge tube management module, the mechanical hand module, the vibration module, and the huller module.

[0064] The constant volume module is used to transfer and perform constant volume processing on the first raw grain or paddy sampled from the upper level. The huller module is used to perform hulling processing on the paddy. The microwave heating and moisture measurement module is used to perform drying processing on the first raw grain or brown rice with excessive moisture. The crushing module is used to crush the second raw grain to pass through a 20-mesh sieve. The large centrifuge tube management module is used to manage large centrifuge tube consumables. The weighing and feeding module is used to accurately weigh the powdered sample and transfer it to the large centrifuge tube to be detected. The open and close bottle cap module is used to open and close the bottle cap before weighing. The detection module is used to detect the specified hygiene indicators. The mechanical hand module is used to transfer the powdered sample between different modules.

[0065] Further, the processing unit is internally integrated with a GPU for image processing. The vibration module is located directly below the screening tray of the huller module. The bottom of the screening tray is a slide rail, and the direction of the slide rail is consistent with the vibration direction of the vibration module, which can realize reciprocating vibration of the screening tray. The vibration module is a vibration motor, which is used to drive the entire screening tray to vibrate and drive the brown rice to turn over. The air flow module is two groups of adjustable speed air blowers. The air outlet of one group of air blowers is perpendicular to the vibration direction of the vibration module. The air blower is located on one side of the huller discharge port and is used for screening rice hulls. The other group of air blowers is located on one side of the crushing module discharge port and is used for screening different sizes of powdered samples after crushing.

[0066] The visual module is composed of image sensors with different focal lengths and different recognition angles, and is arranged above the huller module and the crushing module respectively, and is used for detecting the shelling efficiency of the huller module and the crushing efficiency of the crushing module.

[0067] The present application has the following advantages:

[0068] 1. The sample can realize full-process automation and unmanned from sample crushing, quantitative weighing, liquid adding, shock extraction, separation, pipetting, dilution, incubation to physicochemical index detection.

[0069] 2. The visual module can quickly identify the rice hull through the cooperation of the vibration module and the airflow module, the airflow module can also screen the powdered sample after crushing, realize the full-automatic crushing and detection process without manual intervention, and improve the crushing and detection efficiency.

[0070] 3. The image recognition algorithm of the visual module can dynamically identify the size of the rice hull and the powdered sample during shelling and crushing, and compared with manual screening, the screening accuracy of machine vision is higher and the error rate is lower, which is beneficial to the data reliability of subsequent detection.

[0071] Of course, any product implementing the present application does not necessarily need to achieve all the advantages described above. BRIEF DESCRIPTION OF DRAWINGS

[0072] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0073] Figure 1 A flow chart of the full-automatic crushing and detection method of the grain particle material of the present application;

[0074] Figure 2 A block diagram of the full-automatic crushing and detection system of the grain particle material of the present application. DETAILED DESCRIPTION

[0075] The technical solutions in the embodiments of the present application will be described clearly and completely in the following with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0076] Please refer to Figures 1-2 The present application provides a technical solution: a full-automatic breaking and detecting method for grain particle materials, as shown in the figure, comprising the following steps: Figure 1

[0077] Step 1: issue the first raw grain, sample number, detection item and moisture content, judge whether the first raw grain is rice, if the first raw grain is not rice, proceed to the next step, otherwise, if the first raw grain is rice, jump to step 3;

[0078] Step 2: perform constant volume treatment on the first raw grain, fix the volume of the container, discard the excess first raw grain, and jump to step 4;

[0079] Step 3: constant volume treatment is performed on the rice, and the rice huller module is used to hull the rice to obtain brown rice, and the rice huller module automatically adjusts the gap to adapt to different types of rice hulling;

[0080] Step 4: judge whether the moisture content of the hulled brown rice exceeds 15%, if the moisture content does not exceed 15%, jump to step 7, if the moisture content exceeds 15%, proceed to the next step;

[0081] Step 5: microwave heating and drying the first raw grain or brown rice to obtain the second raw grain;

[0082] Step 6: judge again whether the moisture content of the second raw grain exceeds 15%, if the moisture content does not exceed 15%, proceed to the next step 7, if the moisture content exceeds 15%, re-label the second raw grain as the first raw grain and jump to step 5;

[0083] Step 7: break the second raw grain or brown rice to obtain a powder sample, and use the brown rice or second raw grain to grade break and air screen the powder sample;

[0084] Step 8: weigh the sample, accurately control the amount of powder sample entering the subsequent operation;

[0085] Step 9: weigh the centrifuge tube;

[0086] Step 10: weigh the sample, recalibrate to ensure the accuracy of the weighing;

[0087] Step 11: close the bottle cap after weighing;

[0088] Step 12: transfer to the detection module to start detection;

[0089] Step 13: query and report the results.

[0090] The rice huller module uses image recognition algorithm to judge the hulling efficiency, the specific steps are as follows:

[0091] ​Step 31: the huller module completes the initial hulling of the paddy to obtain brown rice, and the brown rice after hulling is discharged into the screening tray through the discharge port of the huller module;

[0092] Step 32: the processing unit sends instructions to the vibration module and the airflow module respectively, the airflow module works to generate airflow blowing to the screening tray, and the paddy that has not completed hulling will carry the rice husk into the screening tray. Because the weight of the rice husk is much smaller than the weight of the paddy itself, when the airflow module blows air to the surface of the vibrating screening tray, the rice husk will deviate from the vibration direction under the influence of the airflow, which is convenient for the visual module to find;

[0093] Step 33: a plurality of image sensors are arranged above the screening tray, each image sensor corresponds to a different recognition angle and focal length, and the intersection points of the plurality of image sensors are located on the inner surface of the screening tray. For example, the first image sensor has a recognition angle of 15 degrees east and a focal length of 1 meter, and the second image sensor has a recognition angle of 10 degrees south and a focal length of 0.9 meters. By setting multiple image sensors with different recognition angles and focal lengths to form a sensor array, high-definition image acquisition of the screening tray can be achieved using image fusion technology, providing image data support for subsequent machine vision algorithms and helping to improve the accuracy of image recognition algorithms. The processing unit sets the acquisition frequency to 24 Hz, that is, the image sensor acquires 24 times of image information per second;

[0094] Step 34: the image sensor acquires image information according to the acquisition frequency and transmits it to the processing unit, and the processing unit executes the image fusion program to obtain frame images;

[0095] Step 35: the processing unit uses the optical flow method to compare the front and rear frame images in time sequence, the gray value of the same pixel point in the adjacent frame images is unchanged, and the formula is obtained by Taylor expansion and ignoring high-order terms , where x is the initial x direction in the frame image, y is the initial y direction in the frame image, and t is the initial time in the frame image. The initial x direction, the initial y direction and the initial time are located in the previous frame of the compared frame images, where , , u and v are the velocity components of the same pixel point in the x direction and the y direction, Δx is the displacement distance of the pixel point in the x direction of the front and rear frame images, Δy is the displacement distance of the pixel point in the y direction of the front and rear frame images, Δt is the interval time between the front and rear frame images s, Ix is the x direction value in the frame image spatial gradient, Iy is the y direction value in the frame image spatial gradient, and It is the value of the frame image time gradient. The processing unit pre-processes the frame image, performs Gaussian blur noise reduction on the frame image to reduce the interference of high-frequency noise on gradient calculation, and normalizes the gray value of all pixel points in the frame image to ensure consistency under different lighting conditions.

[0096] Step 36: The processing unit calculates the spatial gradient (Ix, Iy) and the temporal gradient It for each frame image respectively;

[0097] Step 37: For each pixel point (x, y) in the frame image, all points in the adjacent region window of the pixel point are extracted, for example, 8 adjacent pixel points in a 3x3 window, and an equation group is constructed:

[0098] wherein , The matrix A is a motion constraint set of all pixels in the adjacent region, each row represents the contribution of a pixel to the optical flow equation, and each row in the vector b is the negation of the temporal gradient of each pixel point in the adjacent region. The velocity vector is solved by the least square method ATA is the covariance matrix of the spatial gradient of the adjacent region, reflecting the distribution of the gradient, and ATb is the correlation between the temporal gradient and the temporal change, which determines the direction of the optical flow. The processing unit analyzes the displacement characteristics of each pixel point in the frame image, calculates the velocity amplitude Vmag of the pixel point according to the formula , calculates the motion direction θ of the pixel point according to the formula , and calculates the trajectory length L of the pixel point i in the frame image according to the formula N is the number of frame images. When the optical flow method is used to calculate the displacement of a single frame image, it is easily disturbed by noise, light changes and other factors, resulting in large fluctuations in single-frame displacement data. By accumulating the displacement of consecutive N frames, short-term noise effects can be smoothed out, and more stable motion features can be extracted;

[0099] Step 38: The processing unit sets a velocity threshold Vth, a direction threshold θth and a length threshold Lth. The velocity threshold Vth is 2.8 pixels / frame, the direction threshold θth is 5 degrees, and the length threshold Lth is 0.3 meters. The processing unit sets trigger condition one, trigger condition two and trigger condition three. Trigger condition one is that the velocity amplitude of the pixel point in the frame image is greater than the velocity threshold Vth, trigger condition two is that the motion direction θ of the pixel point in the frame image is greater than the direction threshold θth, and trigger condition three is that the trajectory length L of the pixel point in the frame image is less than the length threshold Lth. When any two of trigger condition one, trigger condition two and trigger condition three are met, the processing unit marks the corresponding pixel point as a rice hull;

[0100] Step 39: The processing unit presets the adjustment threshold value as 1.5%, and the processing unit calculates the proportion of rice hulls in rice in the frame image in real time. The proportion is specifically obtained by calculating the proportion of the number of pixel points marked as rice hulls in the total pixels in the frame image. Because there are pixel points corresponding to the screening tray and the environment in the frame image in addition to the rice, the adjustment threshold value needs to be set lower, and sufficient data redundancy is reserved. When the calculated proportion is greater than the adjustment threshold value, the processing unit controls the huller module to increase the extrusion friction force by a fixed proportion of 1%. When the extrusion friction force of the huller module is increased to the maximum, it will not be increased any more. The processing unit presets the adjustment period as 5 minutes. If the proportion is always less than the adjustment threshold value within the adjustment period, the processing unit controls the huller module to reduce the extrusion friction force by a fixed proportion of 0.5%, which is half of the previous fixed proportion. This is used to improve the hulling efficiency of the huller and avoid excessive rice hulls affecting subsequent processing steps.

[0101] It should be noted that the adjustment of the extrusion friction force of the huller module is realized by adjusting the speed difference between the two extrusion rollers inside the huller. The huller on the market with a speed-adjustable motor can meet the requirements, and the specific model is not limited here.

[0102] The image fusion program specifically includes the following steps:

[0103] Step 341: The processing unit fuses the image information of multiple recognition angles at the same focal length. Based on the SIFT recognition algorithm, the feature points in each image information are recognized, and the image information is fused according to the recognition angle. For example, the recognition angle of the first image sensor is 15 degrees east and vertically downward. Taking the discharge direction of the huller as east, the image information corresponding to the first image sensor is arranged mainly in the east direction before the recognition algorithm is executed. Then, the SIFT recognition algorithm is executed. After the image information at the same focal length is fused, the focal length map is obtained.

[0104] Step 342: The processing unit repeats step 341 until all image information is fused into focal length maps of different focal lengths. The focal length maps are arranged in order of increasing focal length to obtain a focal length queue. The processing unit counts the number p of the focal length queue.

[0105] Step 343: The processing unit equally divides each focal length map into parts to obtain tiles. Specifically, the length and width of the rectangular focal length map are equally divided into p parts. The tiles are converted into grayscale maps, and the grayscale value range of the grayscale maps is [0, 255]. The difference between the grayscale values of adjacent pixels in the tiles is calculated and accumulated to obtain a difference value. The tiles are arranged in order of decreasing difference value to obtain a difference value queue of the focal length map.

[0106] Step 344: The processing unit extracts the front The tiles with the top difference represent the clarity of the tiles in the focus map, which is in the front of all tiles, facilitating subsequent identification of the powdery sample and the rice hull. The processing unit places the extracted tiles according to the positions in the focus map to obtain a spliced image;

[0107] Step 345: The processing unit scans the spliced image, and the extracted tiles cannot completely piece together the frame image. The tiles at the same position in the focus map may overlap or be missing. If the tiles overlap at the same position in the spliced image, the processing unit compares the difference values of the overlapping tiles and retains the tile with the largest difference value. In theory, the larger the difference value, the clearer the pixels covered by the tile. If there is a missing tile in the spliced image, the processing unit extracts the tile with the largest difference value from the corresponding position in all focus maps and supplements it into the spliced image.

[0108] Step 346: The processing unit repeatedly executes step 345 until there is no overlapping tile or missing tile in the spliced image. The processing unit marks the complete spliced image as a frame image.

[0109] The spatial gradient (Ix, Iy) calculation of the frame image specifically includes the following steps:

[0110] The spatial gradient (Ix, Iy) is calculated by a Sobel operator or a Scharr operator. Both the Sobel operator and the Scharr operator are discrete differential operators that calculate the spatial gradient of an image through convolution. The Sobel operator has a 3x3 kernel and is combined with Gaussian smoothing, which is robust to noise. The Scharr operator also has a 3x3 kernel, but with larger coefficients, making it more sensitive to low-contrast edges.

[0111] The Sobel operator calculation steps are as follows:

[0112] Step 361:

[0113] The x-direction kernel of the horizontal gradient, with higher center column weights, highlighting horizontal edges;

[0114] The y-direction kernel of the vertical gradient, with higher center row weights, highlighting vertical edges;

[0115] Step 362: Fill in the blank or mirror pixels for the edges of the frame image to increase the size of the frame image and avoid size reduction after convolution.

[0116] Step 363: For each pixel (g, k) in the frame image, take the sum of the multiplication of the 3x3 adjacent region, and the specific formula is as follows:

[0117] ,

[0118] where Igray is the gray value of the corresponding pixel point when the frame image is converted to a grayscale image.

[0119] Step 364: Normalize Ix and Iy to [0, 255], formula is , The convolution result may be negative, so normalization processing is needed.

[0120] Wherein, the Scharr operator calculation steps are as follows:

[0121] Horizontal gradient x direction kernel: , Vertical gradient y direction kernel: , the center coefficient is larger, and the weak edge is more sensitive;

[0122] The Scharr operator has the same calculation steps as the Sobel operator, only the kernel is different.

[0123] Wherein, the frame image time gradient It calculation specifically includes the following steps:

[0124] Step 365: Calculate the time gradient It using the difference method, the formula is as follows:

[0125] For each pixel point (e, o) in the frame image, subtract the gray value of the corresponding pixel point in the next frame image from the gray value of the corresponding pixel point in the current frame image, and the gray value range is [0, 255];

[0126] Step 366: Use Gaussian filter to smooth the obtained time gradient It. Due to image noise and discontinuity of object motion, the directly calculated time gradient may contain noise and outliers. In order to make the calculation result more stable and reliable, the calculated time gradient can be smoothed.

[0127] Wherein, the grading crushing adopts image recognition algorithm to judge the crushing effect, and the specific steps are as follows:

[0128] Step 71: The crushing is divided into coarse crushing stage, medium crushing stage and fine crushing stage, the crushing range of the coarse crushing stage of the brown rice or the second raw grain is set to 10-20mm, the crushing range of the medium crushing stage is 3-5mm, and the crushing threshold size of the fine crushing stage is 20 mesh;

[0129] Step 72: The processing unit controls the crushing module to crush the brown rice or the second raw grain, and the processing unit obtains image information from the vision module, identifies whether the diameter of 95% of the brown rice or the second raw grain after crushing meets the range set by the stage, if it meets the range, it enters the next stage, otherwise, if it does not meet the set range, it repeats the current stage, and the grading crushing is completed when 98% of the brown rice or the second raw grain is smaller than 20 mesh in the fine crushing stage;

[0130] Step 73: After the fine crushing stage of the brown rice or the second raw grain is completed, a powdery sample is obtained, an air flow screen is used, a 20-mesh screen is set, the screen is perpendicular to the blowing direction of the air flow module, the processing unit sends a command to the air flow module, the air flow module blows air flow to the crushed powdery sample, the processing unit adjusts the air volume and air speed of the air flow module, the air volume is controlled at 500-1000 cubic meters / hour, the air speed is controlled at 10-15 meters / second, so that the air flow can lift the powdery sample and screen it.

[0131] The visual module determines the crushing efficiency of the crushing module, which includes the following steps:

[0132] Step 76: The processing unit obtains the focal length f, pixel size s, and erection height H of each image sensor in the visual module, obtains the image information of each image sensor, randomly selects one of the powdery samples, and measures the imaging size The imaging size is the number of pixels occupied by the particles of the powdery sample;

[0133] Step 77: The processing unit calculates the size of the selected powdery sample particles according to the formula and records it;

[0134] Step 78: Repeat step 77 to continuously count the proportion of the number of powdery samples within the range, and recalculate the proportion every time step 77 is repeated.

[0135] A full-automatic crushing and detecting system for grain particle materials, as shown in Figure 2 , includes a constant volume module, a huller module, a visual module, an air flow module, a vibration module, a microwave heating and moisture measurement module, a crushing module, a large centrifuge tube management module, a weighing and feeding module, an opening and closing cap module, a detection module, and a mechanical hand module. The output ends of the visual module, the weighing and feeding module, the opening and closing cap module, and the detection module are connected to the input end of the processing unit. The output end of the processing unit is connected to the input ends of the constant volume module, the microwave heating and moisture detection module, the air flow module, the crushing module, the large centrifuge tube management module, the mechanical hand module, the vibration module, and the huller module.

[0136] The constant volume module is used for transferring and constant volume processing of the first raw grain or paddy of the upper sampling, the huller module is used for hulling the paddy, the microwave heating and moisture measuring module is used for drying the first raw grain or brown rice with excessive moisture, the crushing module is used for crushing the second raw grain to more than 99% passing through a 20-mesh screen, the large centrifuge tube management module is used for managing consumables of the large centrifuge tube, the weighing and feeding module is used for accurately weighing the powder sample and transferring it to the large centrifuge tube to be detected, the bottle cap opening and closing module is used for the operation of opening and closing the bottle cap before weighing, the detection module is used for detecting the specified health indicators, and the manipulator module is used for transferring the powder sample between different modules.

[0137] The processing unit is internally integrated with a GPU for image processing, i.e., a graphics acceleration computing unit, which can share the computing pressure of the processing unit on image information processing and improve the overall response speed. The vibration module is located directly below the screening tray of the huller module, the bottom of the screening tray is a slide rail, the direction of the slide rail is consistent with the vibration direction of the vibration module, and the reciprocating vibration of the screening tray can be realized. The vibration module is a vibration motor, which is used to drive the vibration of the entire screening tray to drive the brown rice to turn over. The vibration frequency of the vibration module ranges from 100 to 120 times per minute. The air flow module is two groups of adjustable speed blowers, one group of blowers has an air outlet perpendicular to the vibration direction of the vibration module. The blower is located on one side of the huller discharge port and is used for screening rice hulls. The other group of blowers is located on one side of the discharge port of the crushing module and is used for screening different sizes of powder samples after crushing.

[0138] The vision module is composed of a plurality of image sensors with different focal lengths and different recognition angles. The vision module is located above the huller module and the crushing module, respectively, and is used to detect the hulling efficiency of the huller module and the crushing efficiency of the crushing module.

[0139] The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any skilled person in the art can make equivalent replacements or changes to the technical solution and the inventive concept of the present application within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A method for fully automatic crushing and detection of grain particles, characterized by: The following steps are involved: Step 1: Determine whether the first raw grain is rice. If the first raw grain is not rice, proceed to the next step. Otherwise, if the first raw grain is rice, jump to step 3. Step 2: Perform volume control on the first raw grain and skip to step 4; Step 3: The rice is subjected to a constant volume treatment, and the rice hulling machine module hulls the rice to obtain brown rice. The rice hulling machine module uses an image recognition algorithm to determine the hulling efficiency. The rice hulling machine module automatically adjusts the gap to adapt to different types of rice hulling; Step 4: Determine whether the moisture content of the shelled brown rice exceeds 15%. If the moisture content does not exceed 15%, jump to step 7. If the moisture content exceeds 15%, proceed to the next step. Step 5: Microwave-heating and drying the first raw grain or brown rice to obtain a second raw grain; Step 6: Determine again whether the moisture content of the second raw grain exceeds 15%. If the moisture content does not exceed 15%, proceed to step 7. If the moisture content exceeds 15%, re-mark the second raw grain as the first raw grain and jump to step 5. Step 7: crushing the second raw grain or brown rice to obtain a powdered sample, and performing graded crushing and airflow screening; Step 8: Weigh and load the material to accurately control the amount of powdered sample entering the subsequent operation; Step 9: Weigh the centrifuge tube; Step 10: Weigh and load, and recalibrate; Step 11: Close the bottle cap after weighing; Step 12: Transfer to the detection module to start detection; Step 13: Query and report the results.

2. The method for fully automatic crushing and detecting grain particles according to claim 1, characterized in that: The rice hulling module uses an image recognition algorithm to determine the hulling efficiency. The specific steps are as follows: Step 31: The rice huller module completes the preliminary hulling of the rice to obtain brown rice, and the hulled brown rice passes through the discharge port of the rice huller module and enters the screening tray; Step 32: The processing unit sends instructions to the vibration module and the airflow module respectively, and the airflow module works to generate airflow blowing toward the screening tray; Step 33: A plurality of image sensors are mounted above the screening tray, each corresponding to a different recognition angle and focal length. The plurality of image sensors are positioned on the inner surface of the screening tray toward the intersection, and the acquisition frequency is set; Step 34: The image sensor acquires image information according to the acquisition frequency and transmits it to the processing unit, which executes the image fusion program to obtain a frame image; Step 35: Use the optical flow method to compare the previous and next frame images in chronological order. The grayscale value of the same pixel in the previous and next adjacent frame images remains unchanged. According to the formula Taylor expansion and ignoring higher-order terms yields , x is the initial x direction in the frame image, y is the initial y direction in the frame image, and t is the initial time in the frame image, where , , u and v are the velocity components of the same pixel in the x and y directions respectively, Δx is the displacement distance of the pixel in the previous and next frame images in the x direction, Δy is the displacement distance of the pixel in the previous and next frame images in the y direction, and Δt is the interval time between the previous and next frame images s, Ix is the x-direction value of the spatial gradient of the frame image, Iy is the y-direction value of the spatial gradient of the frame image, It is the value of the temporal gradient of the frame image, the frame image is preprocessed, Gaussian blurring is performed on the frame image to reduce noise, and the grayscale values ​​of all pixels in the frame image are normalized; Step 36: Calculate the spatial gradient (Ix, Iy) and the temporal gradient It for the frame image respectively; Step 37: For each pixel (x, y) in the frame image, extract all points in the adjacent area window of the pixel point and construct the equation system: ,in , , solve the velocity vector by the least squares method , ATA is the covariance matrix of spatial gradients in adjacent regions, ATb is the correlation between temporal gradients and temporal changes, and the displacement characteristics of each pixel in the frame image are analyzed. According to the formula Calculate the velocity amplitude Vmag of the pixel point according to the formula Calculate the movement direction of the pixel point θ, according to the formula Calculate the trajectory length L of pixel point i in the frame image, where N is the number of frame images; Step 38: Set a speed threshold Vth, a direction threshold θth, and a length threshold Lth, and set trigger conditions one, two, and three. Trigger condition one is that the speed amplitude of the pixel in the frame image is greater than the speed threshold Vth, trigger condition two is that the motion direction θ of the pixel in the frame image is greater than the direction threshold θth, and trigger condition three is that the trajectory length L of the pixel in the frame image is less than the length threshold Lth. When any two of trigger conditions one, two, and three are met, the corresponding pixel is marked as rice husk. Step 39: Preset an adjustment threshold, and calculate in real time the proportion of rice husks in rice in the frame image. When the calculated proportion is greater than the adjustment threshold, the processing unit controls the rice hulling machine module to increase the extrusion friction force by a fixed proportion. When the extrusion friction force of the rice hulling machine module reaches the maximum, it will no longer increase. Preset an adjustment cycle. If the proportion is never greater than the adjustment threshold during the adjustment cycle, the processing unit controls the rice hulling machine module to reduce the extrusion friction force by half of the fixed proportion, and brown rice is obtained after completing the hulling.

3. The method for fully automatic crushing and detecting grain particles according to claim 2, characterized in that: The image fusion procedure specifically includes the following steps: Step 341: fusing image information at multiple recognition angles at the same focal length, identifying feature points in each image information based on a recognition algorithm, and fusing the image information based on the recognition angle auxiliary graphic information. After fusing the image information at the same focal length, a focal length map is obtained. Step 342: Repeat step 341 until all image information is fused into focal length maps of different focal lengths, arrange the focal length maps in ascending order of focal length to obtain focal length queues, and count the number of focal length queues p; Step 343: Divide each focal length map into equal parts The image blocks are obtained in batches, and the image blocks are converted into grayscale images. The grayscale value differences between adjacent pixels in the image blocks are calculated and accumulated to obtain the differences. The image blocks are arranged in descending order of the differences to obtain a difference queue. Step 344: Extract the top of the difference queue Blocks, place the extracted blocks according to their positions in the focal length map to obtain a mosaic image; Step 345: Scan the mosaic image. If there are overlapping tiles at the same position in the mosaic image, compare the difference between the overlapping tiles and retain the tile with the largest difference. If there are missing tiles in the mosaic image, extract the tile with the largest difference from the corresponding position in all focal length images and add it to the mosaic image. Step 346: Repeat step 345 until there are no overlapping blocks or missing blocks in the mosaic image, and then mark the complete mosaic image as a frame image.

4. The method for fully automatic crushing and detecting grain particles according to claim 2, characterized in that: The calculation of spatial gradient (Ix, Iy) of a frame image specifically includes the following steps: The spatial gradient (Ix, Iy) is calculated by the Sobel operator or the Scharr operator. Both the Sobel operator and the Scharr operator are discrete differential operators. The spatial gradient of the image is calculated by convolution. The calculation steps of the Sobel operator are as follows: Step 361: The x-direction kernel of the transverse gradient, ; The y-direction kernel of the longitudinal gradient, ; Step 362: Fill the edge of the frame image with blanks or mirror pixels; Step 363: For each pixel (g, k) in the frame image, take the 3x3 adjacent regions and multiply them together to obtain the sum. The specific formula is as follows: ， Among them, Igray is the grayscale value of the corresponding pixel when the frame image is converted into a grayscale image; Step 364: Normalize Ix and Iy to [0, 255].

5. The method for fully automatic crushing and detecting grain particles according to claim 4, characterized in that: The calculation steps of the Scharr operator are as follows: Transverse gradient x-direction kernel: , longitudinal gradient y direction kernel: ; The rest of the calculation steps of the Scharr operator are the same as those of the Sobel operator, with only the kernel being different.

6. The method for fully automatic crushing and detecting grain particles according to claim 2, characterized in that: The calculation of the time gradient It of the frame image specifically includes the following steps: Step 365: Calculate the time gradient It using the difference method. The formula is as follows: , for each pixel (e, o) in the frame image, subtract the grayscale value of the pixel corresponding to the current frame from the grayscale value of the pixel corresponding to the next frame; Step 366: Use Gaussian filtering to smooth the obtained time gradient It.

7. The method for fully automatic crushing and detecting grain particles according to claim 1, characterized in that: Grading crushing uses image recognition algorithm to judge the crushing effect. The specific steps are as follows: Step 71: Crushing is divided into a coarse crushing stage, a medium crushing stage, and a fine crushing stage. The crushing range of the coarse crushing stage of the brown rice or the second raw grain is set to 10-20 mm, the crushing range of the medium crushing stage is set to 3-5 mm, and the crushing threshold size of the fine crushing stage is set to 20 mesh; Step 72: The processing unit controls the crushing module to crush the brown rice or the second raw grain, obtains image information from the vision module, and identifies whether the diameter of 95% of the brown rice or the second raw grain after crushing meets the range set for the stage. If it meets the range, it proceeds to the next stage. Otherwise, if it does not meet the set range, the current stage is repeated. In the fine crushing stage, when the size of 98% of the brown rice or the second raw grain is less than 20 mesh, the graded crushing is completed. Step 73: After the brown rice or the second raw grain completes the fine crushing stage, a powdered sample is obtained, and airflow screening is used. A 20-mesh screen is set, and the screen is perpendicular to the blowing direction of the airflow module. The processing unit sends an instruction to the airflow module, and the airflow module blows air to the crushed powdered sample, adjusts the air volume and wind speed of the airflow module, and performs screening.

8. The method for fully automatic crushing and detecting grain particles according to claim 1, characterized in that: The visual module determines the crushing efficiency of the crushing module by the following steps: Step 76: Obtain the focal length f, pixel size s, and mounting height H of each image sensor in the vision module, obtain the image information of each image sensor, randomly select one of the powder samples and measure its image size ; Step 77: According to the formula Calculate the size of the selected powdered sample and record it; Step 78: Repeat step 77 to continuously count the percentage of powdered samples that meet the range, and recalculate the percentage each time step 77 is repeated.

9. A fully automatic crushing and detection system for grain particles, characterized by A fully automatic crushing and detection method for grain particles according to claim 1, comprising a volume control module, a rice hulling machine module, a vision module, an airflow module, a vibration module, a microwave heating and moisture measurement module, a crushing module, a large centrifuge tube management module, a weighing and loading module, a bottle cap opening and closing module, a detection module, and a manipulator module, wherein the output ends of the vision module, the weighing and loading module, the bottle cap opening and closing module, and the detection module are all connected to the input end of the processing unit, and the output end of the processing unit is respectively connected to the input ends of the volume control module, the microwave heating and moisture detection module, the airflow module, the crushing module, the large centrifuge tube management module, the manipulator module, the vibration module, and the rice hulling machine module; The constant volume module is used to undertake the transfer of the first raw grain or rice sampled by the upper level and perform constant volume treatment. The rice huller module is used to hull the rice. The microwave heating and moisture measurement module is used to dry the first raw grain or brown rice with excessive moisture. The crushing module is used to crush the second raw grain to pass through a 20-mesh sieve. The large centrifuge tube management module is used to manage large centrifuge tube consumables. The weighing and loading module is used to accurately weigh powdered samples and transfer them to large centrifuge tubes to be tested. The bottle cap opening and closing module is used for opening and closing the bottle cap before weighing. The detection module is used to detect specified hygiene indicators. The manipulator module is used to transfer powdered samples between different modules.

10. The fully automatic crushing and detection system for grain particles according to claim 9, characterized in that: The processing unit integrates a GPU for image processing. The vibration module is located directly below the rice huller module. The vibration module is used to drive the turning of the brown rice. The airflow module is used to screen the rice husks and screen the powder samples of different sizes after crushing. The vision module consists of several image sensors with different focal lengths and different recognition angles. The vision modules are located above the rice huller module and above the crushing module respectively, and are used to detect the shelling efficiency of the rice huller module and the crushing efficiency of the crushing module.

Images