A system for quantitative slicing of bone-in meat

Abstract

The application discloses a system for quantitative slicing of bone-in meat, which comprises a data acquisition device for bone-in meat slicing, a data processing device and a slicing execution mechanism, wherein the data acquisition device is used for point cloud data acquisition of the three-dimensional shape of the bone-in meat during quantitative slicing; the data processing device is used for processing and calculation of the collected point cloud data, including curved surface reconstruction processing of the bone-in meat; and the slicing thickness is calculated according to production requirements; and the slicing execution mechanism is used for quantitative slicing of the bone-in meat according to the calculated interval thickness. The system has high intelligence, reduces the loss and waste of the bone-in meat raw material, does not need professional technical personnel to operate and maintain, and reduces production cost. The application promotes the progress and development of the quantitative slicing process of the bone-in meat, and provides guarantee for improving the profit of meat processing enterprises.

Description

Technical Field

[0001] This invention relates to the field of food processing machinery and three-dimensional measurement technology, specifically to a system for quantitative slicing of bone-in meat. This system is mainly used for quantitative slicing of bone-in meat. Background Technology

[0002] With the increasing meat production in my country year by year, bone-in meat, with its unique taste and rich nutritional value, is highly favored by consumers. Slicing, as an important step in the secondary processing of bone-in meat, aims to precisely cut the meat into slices of a certain quality according to production needs. Currently, although most meat processing enterprises have used automated slicing machines to replace manual slicing, their level of automation is low, resulting in significant slicing waste, which directly affects the profits of meat processing enterprises.

[0003] With the increase in meat production and the steady growth of processing companies' revenue, the corresponding processing volume is also constantly increasing. However, if meat processing technology fails to be improved and upgraded accordingly, the waste of raw materials will inevitably intensify. This waste not only leads to the ineffective use of resources but also directly damages the profits of meat processing companies. Therefore, improving and upgrading meat processing technology is particularly necessary; it is not only key to improving economic efficiency but also an inevitable choice for achieving sustainable development.

[0004] Currently, domestic meat processing enterprises generally use manual or mechanical slicing methods for bone-in meat slicing. Manual slicing is problematic due to the cold, noisy, and slippery working environment, the potential for worker injuries from slicing blades, low efficiency, high defect rate, significant raw material loss, and easy contamination of meat surfaces during large-scale slicing operations. Therefore, this method is gradually being phased out. In contrast, mechanical slicing is unaffected by the working environment, and the entire slicing process can be automated, greatly improving slicing efficiency and reducing manpower while minimizing meat surface contamination. However, most bone-in meat slicing equipment in the domestic market still cannot achieve quantitative slicing, only performing simple cutting actions, resulting in low levels of automation and significant waste of bone-in meat.

[0005] To overcome these problems, bone-in meat slicing devices based on machine vision and 3D measurement technology have emerged in recent years. These devices achieve precise control over slice thickness by acquiring 3D morphological data of bone-in meat. However, existing data acquisition devices still have some shortcomings. For example, some devices have low accuracy in data acquisition, making it difficult to meet the requirements of high-precision slicing; others are complex to operate, requiring professional technicians for operation and maintenance, which increases production costs.

[0006] Therefore, developing a data acquisition device capable of accurately and efficiently acquiring three-dimensional point cloud data of bone-in meat and designing corresponding algorithms based on this to achieve quantitative slicing of bone-in meat is of great significance for improving the accuracy and efficiency of quantitative slicing of bone-in meat. Summary of the Invention

[0007] The existing technologies suffer from problems and drawbacks, primarily including high cost, low efficiency, insufficient data accuracy, and operational complexity. These issues limit the effectiveness of existing technologies in quantitative slicing of bone-in meat, necessitating the search for more efficient, accurate, and cost-effective solutions. This invention arose in this context, aiming to improve the efficiency and accuracy of quantitative slicing of bone-in meat by addressing the problems in existing technologies through improvement and innovation.

[0008] Technical solution:

[0009] This invention discloses a system for quantitative slicing of bone-in meat, comprising a data acquisition device, a data processing device, and a slicing execution mechanism, wherein:

[0010] The data acquisition device is used to collect point cloud data on the three-dimensional morphology of bone-in meat during quantitative slicing.

[0011] The data processing device processes and calculates the collected point cloud data, including surface reconstruction of bone-in meat; and calculates the slice thickness according to production requirements.

[0012] The slicing mechanism quantitatively slices the bone-in meat according to the calculated interval thickness.

[0013] Specifically, the data acquisition device includes a data acquisition module and an electronic control module, wherein:

[0014] The data acquisition module includes: a conveyor belt, a scanner stand, a scanner backplate, a line laser scanner, and a host computer. The line laser scanner is mounted on the scanner backplate, which is horizontally fixed to the scanner stand and located above the conveyor belt. The line laser scanner is connected to the host computer via a data cable.

[0015] The electrical control module includes: a PLC all-in-one unit, a servo motor and its driver, a motor synchronous pulse distributor, photoelectric switches, and a servo motor and its reducer. The PLC all-in-one unit, as the core component of the electrical control module, controls the conveyor belt's movement speed and the photoelectric switch's delay time by programming ladder diagrams. The servo motor and its driver are used to control the servo motor's speed, thereby changing the conveyor belt's running speed. The servo motor driver is equipped with differential signal lines and differential signal terminals for differential signal output. The motor synchronous pulse distributor synchronously distributes the differential signals output from the servo motor with its differential signal lines, providing them simultaneously to different line laser scanners. The photoelectric switch detects whether the bone-in tissue is within the scanning range of the line laser scanner, then feeds back a level signal to achieve synchronous acquisition by the line laser scanner. The servo motor reducer uses a gear speed converter to reduce the motor's rotation speed to the desired speed, obtaining a larger torque. Its function is to improve torque output, reduce speed, improve accuracy, and protect the motor.

[0016] Specifically, the data processing device includes a point cloud preprocessing module, a 3D reconstruction module, a volume calculation module, and a slice thickness calculation module, wherein:

[0017] Point cloud preprocessing module: By setting threshold ranges for the X, Y, and Z axes, point cloud data outside the threshold range is removed to achieve region of interest segmentation;

[0018] 3D Reconstruction Module: Based on the processed point cloud data output by the point cloud preprocessing module, it realizes the reconstruction of the surface model with bone and flesh;

[0019] Volume calculation module: Calculates the total volume of meat and bone based on the reconstructed surface model;

[0020] Slice thickness calculation module: Calculates slice thickness based on given slice quality requirements.

[0021] Specifically, the point cloud preprocessing module includes the following steps:

[0022] S1.1. Region of interest segmentation, performed using the following formula:

[0023] Xmin <= X <= Xmax

[0024] Ymin <= Y <= Ymax

[0025] Zmin <= Z <= Zmax

[0026] In the formula, Xmin, Xmax, Ymin, Ymax, Zmin, and Zmax represent the set threshold range of point cloud data coordinates; X, Y, and Z represent the range of X, Y, and Z axis coordinates of the segmented and extracted point cloud data.

[0027] S1.2 Statistical filtering, performed using the following formula:

[0028]

[0029]

[0030]

[0031] In the formula, X n Y n Z n Let X and Y represent the coordinates of the nth point in the point cloud, respectively. m Y m Z m S represents the coordinates of any point in the point cloud; i Represents the nth point (X) in the point cloud. n Y n Z n ) to any point (X) m Y m Z m The distance is μ, where μ represents the average distance, σ represents the standard deviation, and n represents the number of points in the point cloud.

[0032] S1.3, Voxel downsampling, performed using the following formula:

[0033]

[0034]

[0035]

[0036] In the formula, x centroid y centroid , z centroid represents the centroid coordinates of the voxel; m represents the number of points in the voxel; x i y i z i These represent the x, y, and z coordinates of all points in a voxel.

[0037] Specifically, the 3D reconstruction module achieves 3D reconstruction of bone and flesh through the following steps:

[0038] S2.1 Point cloud registration, including:

[0039] (1) Find point q in the point cloud P to be registered and match the point q in the reference point cloud Q. i The nearest neighbor is p i Calculate the squared spatial distance between two points.

[0040] (2) According to the calculation Sort the data in ascending order, keep the first N pairs, and calculate their sum S, where N = kN. p ;k represents the point cloud overlap, N p N represents the number of points in the point cloud p to be registered, and N represents the number of points to be retained.

[0041] The transformation matrix T = (R, t) is calculated using the following formula:

[0042]

[0043] In the formula, R represents the rotation matrix, t represents the translation matrix, and q i p represents any point in the reference point cloud Q. i Indicates the registration point cloud and q i The nearest neighbor;

[0044] (3) Transform the point cloud P to be registered according to the obtained transformation matrix, as shown in the following formula:

[0045]

[0046] in, Represents the coordinates of the transformed point cloud. represents the point cloud coordinates before transformation, and t represents the translation matrix;

[0047] Repeat steps (1)-(3) to calculate the relationship between the point cloud Q to be registered and the reference point cloud P until the set iteration threshold is reached;

[0048] S2.2 Point cloud fusion is performed using the following formula:

[0049] P = P1 + TP2

[0050] In the formula, T represents the acquired transformation matrix, P represents the point cloud model, P1 represents the target point cloud, and P2 represents the point cloud to be registered.

[0051] S2.3 Parameter extraction, performed using the following formula:

[0052] L = x max -x min

[0053] W = y max -y min

[0054] H = z max -z min

[0055] In the formula, L represents the length of the bone-in meat, W represents the width of the bone-in meat, H represents the height of the bone-in meat, and x max ymax , z max This represents the maximum values ​​of the x, y, and z axes of a point in the 3D reconstruction model. min y min , z min This represents the minimum values ​​of the x-axis, y-axis, and z-axis of a point in the 3D reconstruction model.

[0056] Specifically, when the average point spacing in the point cloud is small, the volume calculation module obtains the total volume of the meat with bone through the following steps:

[0057] S3.1 Surface area calculation is performed using the following formula:

[0058]

[0059]

[0060] In the formula, i represents the index of the triangular facet, h i s1 represents half the perimeter of the triangular face. i Let a represent the area of ​​the triangular facet. i b i c i Let S1 represent the length of each side of the triangular facet, n represent the total number of triangular facets, and S1 represent the surface area of ​​the reconstructed model.

[0061] S3.2 Volume calculation, performed using the following formula:

[0062]

[0063] In the formula, H t s represents the distance between the triangular facet in the x0y plane and the original point cloud. t Indicates s1 i The area of ​​each triangular facet projected onto the x0y plane, where V represents the volume of the reconstructed model and n represents the total number of projected triangular facets.

[0064] Specifically, when the average point spacing of the point cloud is large, the volume calculation module obtains the total volume V of the meat with bone through the following steps:

[0065]

[0066] Among them, S i h represents the area of ​​the i-th slice. i Let n represent the interval between adjacent slices, and n represent the number of layers in the slice; the slice area S is obtained by the following formula:

[0067]

[0068] Where m represents the number of vertices in the convex hull, y iLet y and z represent the coordinates of the i-th vertex. i Represents the z-coordinate of the i-th vertex, y i+1 Let y and z represent the coordinates of the (i+1)th vertex. i+1 This represents the z-coordinate of the (i+1)th vertex.

[0069] Specifically, the calculation process for the average point spacing of the point cloud is as follows:

[0070] d p =min(dis(p,q)),q=1,2,…,p≠q

[0071] Where, d p Let represent the minimum distance between point p and other points, and dis(p,q) represent the distance between point p and any point p. Then, the formula for calculating the average point spacing D of the point cloud is:

[0072]

[0073] Where p represents any point in the point cloud, and N represents the number of points in the point cloud.

[0074] Specifically, the slice thickness calculation module obtains the slice thickness through the following steps:

[0075] S4.1 Average density is calculated using the following formula:

[0076]

[0077] Where, ρ sample V represents the average density of the bone-in meat, V represents the total volume of the bone-in meat, and M represents the mass of the bone-in meat obtained by weighing; the density ρ of the slices slice =ρ sample ;

[0078] S4.2, Slice volume V slice calculate:

[0079]

[0080] Among them, M slice Indicates the required slice quality for quantitative sectioning;

[0081] S4.3, Based on the target slice volume V slice A greedy algorithm is used to calculate the slice thickness. Taking the minimum point of the X-axis of the bone-and-flesh 3D model as the starting point, the optimal slice thickness corresponding to this starting point is found, which is to satisfy the required volume V of the quantitative slice as much as possible. slice The slice thickness;

[0082] S4.4 Then, using the end position of this slice as the starting value of the next slice, repeat the process and set the stop criterion.

[0083] Specifically, the slicing mechanism includes a conveyor belt, a conveyor motor, and a cutter. The meat with bones is placed on the conveyor belt, and the conveyor motor receives start / stop signals from the host computer to control the transmission of the conveyor belt; the cutter receives lifting signals from the host computer to perform the slicing action.

[0084] Beneficial effects of the present invention

[0085] This invention employs advanced three-dimensional measurement technology, using laser scanning to precisely capture the surface morphology of bone-in meat. Simultaneously, combined with optimized data processing algorithms and a control system, this invention can quickly and accurately extract key dimensions and shape information of bone-in meat. More importantly, this invention, through volume calculation and slice thickness calculation modules, can calculate slice dimensions according to requirements, achieving fully automated and precise quantitative slicing of bone-in meat.

[0086] The system of this invention has a high degree of intelligence, which reduces the loss and waste of bone-in meat raw materials; it does not require professional technicians for operation and maintenance, thus reducing production costs; this invention promotes the progress and development of quantitative slicing technology for bone-in meat, and provides a guarantee for improving the profits of meat processing enterprises. Attached Figure Description

[0087] Figure 1 This is a hardware system composition diagram of the present invention.

[0088] Figure 2 Model diagram of data acquisition device

[0089] Figure 3 Schematic diagram of the internal layout of the electrical control cabinet

[0090] Figure 4 Human-computer interface program diagram

[0091] Figure 5 Overall design flowchart for software for quantitative slicing of frozen pork chops

[0092] Figure 6 Interface diagram for the "point cloud preprocessing" function

[0093] Figure 7 Schematic diagram of voxel downsampling principle

[0094] Figure 8 Interface diagram for the "3D Reconstruction" function

[0095] Figure 9 Interface diagram for the "Surface Reconstruction" function

[0096] Figure 10Interface diagram for the "Volume Calculation" function

[0097] Figure 11 A flowchart illustrating the method for calculating slice thickness. Detailed Implementation

[0098] The present invention will be further described below with reference to embodiments, but the scope of protection of the present invention is not limited thereto:

[0099] This invention first discloses a data acquisition device for quantitative slicing of bone-in meat. It consists of a hardware system, a host computer storage and processing module, and a data display and processing software module, with the acquisition device and computer serving as the carriers, respectively. The technical content mainly includes two aspects: the hardware system and the software data processing module.

[0100] The following is a breakdown of the specific functions of each major component and device.

[0101] Combination Figure 1 The hardware system consists of two parts: a data acquisition module and an electronic control module. The data acquisition module includes a conveyor belt, a scanner stand and backplate, a line laser scanner, and a host computer. The line laser scanner is mounted on the scanner backplate, which is horizontally fixed to the scanner stand and located above the conveyor belt. The line laser scanner is connected to the host computer via a data cable. Figure 3 The electrical control module includes a programmable logic controller, photoelectric switches, servo motors and their drivers, servo motor reducers, and motor synchronization pulse distributors. Figure 2 In the middle, the bracket and back plate 1 serve as the skeleton to fix the conveyor belt 2 and the line laser scanner. The frozen pork chops 3 are placed on the conveyor belt 2. The photoelectric switch 4 is set on one side of the conveyor belt 2. The servo motor and its reducer 5 are set at one end of the conveyor belt 2. The electrical control cabinet 6 is set on one side of the bracket. The electrical control cabinet 6 is equipped with a touch screen 7 and a start / stop button 8.

[0102] Conveyor belts: These ensured uniform horizontal movement of the frozen pork chops during data acquisition. This study used two conveyor belts with a gap between them to allow for upward-looking point cloud data acquisition. The online laser scanner was installed with its laser emitter positioned at a certain distance above, below, and to the left and right of the gap to acquire point cloud data from four different perspectives of the frozen pork chops and to ensure that the frozen pork chops did not vibrate as they passed through the gap.

[0103] Scanner stands and backplate: Four scanner stands are installed directly above, below, and to the left and right sides of the conveyor belt gap. The backplate on the stands is perforated according to the size of the line laser scanner and fitted tightly to it. During the debugging phase, the line laser scanner needs to be moved to obtain the best imaging position, and then the backplate is used to fix it to the stand.

[0104] Line laser scanner: The core equipment of the point cloud acquisition test platform, used to acquire point cloud data of bone and flesh contours. The selected scanner is the AR7260+ model from UKC Measurement.

[0105] The line laser scanner comes with a 5m power signal cable and a 5m network data cable. The power signal cable supplies power to the scanner, receives differential encoder signals, and handles high-level batch processing signals. The red and black pins receive 24V voltage to power the line laser scanner. The differential encoder input interface uses RS-485 differential input, and the batch processing interface is optocoupler-isolated and active high. The network data cable uses standard Gigabit Ethernet, allowing for point cloud data transmission between the line laser scanner and a PC.

[0106] Computer: The host computer storage and processing module, which is the carrier for using MATLAB software and the controller for realizing data storage, data display and data processing.

[0107] PLC: The GC-070S-32MAA-C model controller manufactured by Shenzhen Xiankong Technology Co., Ltd. is equipped with a corresponding touch screen. It allows for human-machine interaction through a visual interface, integrating screen configuration and ladder diagram programming. As a core component of the electrical control module, it controls the conveyor belt speed and photoelectric switch delay time by writing ladder diagrams.

[0108] Photoelectric switch: E3F-DS30C4 type diffuse reflection NPN normally open photoelectric switch manufactured by Zhejiang Hugong Automation Technology Co., Ltd. When no object is detected, the switch does not perform any action and remains in the normally open state; when an object is detected, the switch performs an action and changes to the closed state.

[0109] Servo motor and driver: The HXGK-060430E servo motor and driver manufactured by Wenzhou Pufeide Electric Co., Ltd. are used to drive the conveyor belt to perform uniform horizontal movement.

[0110] Servo motor reducer: The NMRV40 reducer, manufactured by Hangzhou Shensu Transmission Equipment Co., Ltd., has a speed ratio of 1:25. It is used to improve torque output, reduce speed, improve accuracy, and protect the motor.

[0111] Motor Synchronous Pulse Distributor: The ZP_ServPulse02 motor synchronous pulse distributor manufactured by Zepeng Industrial Control is used to synchronously distribute and output the differential signals output by the servo motor with differential signal lines so that they can be provided to different line laser scanners simultaneously.

[0112] Relay: This is an IDEC relay used to connect a PLC to a photoelectric switch. It is used in automatic control circuits and is an automatic switch that uses a smaller current to control a larger current.

[0113] Power supply: Meanwell power supply, which converts 120V to 24V to power other devices.

[0114] Conveyor belt start / stop button: Used to start and stop the conveyor belt.

[0115] Terminal blocks: used for wiring connections between various devices.

[0116] A fuse is a protective device that breaks the circuit by melting a fusible element with the heat it generates when the current exceeds a specified value for a certain period of time.

[0117] Circuit breaker: Protects power lines and motors, etc., and can automatically disconnect the circuit when a serious overload, short circuit or undervoltage fault occurs.

[0118] Data storage and processing are performed using MATLAB software on a computer. MATLAB is an interactive software program that can significantly improve the software development process and is a commonly used software for data processing. The data processing mainly includes five functions: point cloud preprocessing, 3D reconstruction, surface reconstruction, volume calculation, and slice thickness calculation.

[0119] The implementation method for quantitative slicing of bone-in meat described in this embodiment includes the following steps:

[0120] S1: Mount the line laser scanner on the back panel of the bracket and connect the power signal cable and the network data cable, with the other end of the network data cable connected to the computer.

[0121] S2: Connect the three-prong power plug of the electrical control cabinet to 120V, rotate the switch knob, and the power indicator light will turn green. The line laser scanner will start, and the power indicator light will turn green. The human-machine interface will then operate as follows. Figure 4 As shown, the servo motor output frequency is set to 15000, and the photoelectric delay time is set to 20*100 milliseconds.

[0122] S3: Place the bone-in meat on one side of the conveyor belt and press the conveyor belt start button. The bone-in meat will move at a constant speed along the direction of the conveyor belt. When it blocks the photoelectric switch, the line laser scanner will start scanning. After moving away from the photoelectric switch, the scanning operation will end after a 2-second delay. Then press the conveyor belt stop button to stop the conveyor belt from running.

[0123] S4: The data collected by the line laser scanner will be transmitted to the computer via a network cable, and the file format is txt.

[0124] S5: Data Processing. Open the frozen pork chops quantitative slicing software, log in, and begin processing the collected data. The overall design flowchart of the frozen pork chops quantitative slicing software is shown below. Figure 5 As shown.

[0125] Clicking "Point Cloud Preprocessing" will open the "Point Cloud Preprocessing" window. Enter the threshold ranges for the X, Y, and Z axes to segment the ROI region. Then, enter the number of points to be counted and the distance threshold to complete statistical filtering (the formula is shown below). Finally, enter the voxel grid size for voxel downsampling. The bottom displays the original point cloud and the number of points after preprocessing. Figure 6 As shown.

[0126] Formula for region of interest segmentation:

[0127] Xmin <= X <= Xmax

[0128] Ymin <= Y <= Ymax

[0129] Zmin <= Z <= Zmax

[0130] X,Y,Z - The range of X, Y, and Z axis coordinates of the acquired point cloud data;

[0131] Xmin, Xmax, Ymin, Ymax, Zmin, Zmax - Set the threshold range of point cloud data coordinates.

[0132] The specific principle of the statistical algorithm is based on the assumption that all points in the point cloud follow a Gaussian distribution, the shape of which is determined by the mean distance μ and the standard deviation σ. Assuming the coordinates of the nth point in the point cloud are Pn(Xn, Yn, Zn), then the distance from this point to any point Pm(Xm, Ym, Zm) is S_i. When the value of μ is greater than a threshold, it is judged as noise and removed from the point cloud data.

[0133] Statistical filtering calculation formula:

[0134]

[0135]

[0136]

[0137] A schematic diagram of the voxel downsampling principle is shown below. Figure 7 As shown.

[0138] Clicking "3D Reconstruction" will open the "3D Reconstruction" window. Click "Read Point Cloud File," select point cloud data from four perspectives, and click "3D Reconstruction" again. The 3D reconstructed model will be displayed at the top of the interface. Click "Shape Parameter Extraction" to obtain the bounding box of the 3D model and display it at the bottom of the interface. Its length, width, and height are displayed at the very bottom of the interface. Figure 8 As shown.

[0139] Formula for obtaining the transformation matrix:

[0140] The key to point cloud registration is obtaining the transformation matrix T, which is composed of the rotation matrix R and the translation matrix t, as shown in the following formula:

[0141]

[0142] Suppose a point in the point cloud to be registered is (x, y, z), and the corresponding point in the point cloud to be registered is (x1, y1, z1). Then the relationship between the two can be expressed by the following formula:

[0143]

[0144]

[0145] Point cloud data fusion formula: P = P1 + TP2 where:

[0146] T-The transformation matrix obtained

[0147] P - The final point cloud model obtained

[0148] P1 - Target Point Cloud

[0149] P2 - Point cloud to be registered.

[0150] The shape parameters of frozen pork chops are extracted using a bounding box algorithm. This method first obtains the maximum values ​​x in the X, Y, and Z directions of the point cloud. max y max , z max and minimum value x min y min , z min The length (L), width (W), and height (H) of the frozen pork chop can be obtained using the following formula.

[0151] L = x max -x min

[0152] W = y max -y min

[0153] H = z max -z min

[0154] Clicking "Surface Reconstruction" will open the "Surface Reconstruction" window. Click "Select Point Cloud File" to select the 3D reconstructed model, which will be displayed at the top of the interface. Enter the alpha value, click "Surface Reconstruction," and the resulting surface reconstructed model will be displayed at the bottom of the interface. The model's volume and surface area will be displayed on the right. Figure 9 As shown.

[0155] Formula for obtaining the surface area of ​​a surface reconstruction model:

[0156]

[0157]

[0158] i-index of triangle

[0159] h i - Half the perimeter of the triangular facet

[0160] a i b i c i - Length of each side of the triangular facet

[0161] n - the total number of triangular facets.

[0162] Formula for obtaining the volume of a surface reconstruction model:

[0163]

[0164] H t Distance values ​​between the triangular mesh in the -x0y- plane and the original point cloud

[0165] s t Indicates s1 i The area of ​​each triangular facet projected onto the x0y plane;

[0166] At this point, V represents the volume of the remodeled frozen pork chop, which is the first way to obtain the volume of the frozen pork chop.

[0167] Clicking "Volume Calculation" will open the "Volume Calculation" window. Click "Select Point Cloud File" to select the 3D reconstructed model point cloud, which will be displayed at the top of the interface. Click "Slice Volume Calculation" and enter the mass of the frozen pork chop and the given slice mass value to calculate the average density and slice volume. Figure 10 As shown.

[0168] Formula for calculating the volume of frozen pork chops:

[0169] The three-dimensional model of frozen pork chops was sliced ​​at equal intervals. The convex hull of the point cloud dataset for each slice was constructed using the Graham scan method and then calculated using the discretized Green's theorem.

[0170]

[0171] Where S is the area of ​​the tangent plane, m is the number of vertices of the convex hull, and y i Let z be the y-coordinate of the i-th vertex. i Let z be the z-coordinate of the i-th vertex, and y be the z-coordinate of the ith vertex. i+1 Let z be the y-coordinate of the (i+1)th vertex. i+1 Let z be the z-coordinate of the (i+1)th vertex.

[0172] After obtaining the slice area, the volume of the frozen pork chop is calculated using the platform formula method. Each pair of slices is approximated as a platform, and the volume of the platform is then calculated using the summation formula, as shown in the following formula:

[0173]

[0174] Where V is the volume of the frozen pork chop, and S i S is the area of ​​the i-th slice. i+1 h is the slice area of ​​the (i+1)th layer. i The interval between adjacent slices is n, and the number of slice layers is n; this is the second method for obtaining the volume of frozen pork chops.

[0175] In a preferred embodiment, when the average point spacing of the point cloud is small, the first method is used to obtain the volume of the frozen pork chop; when the average point spacing of the point cloud is large, the second method is used to obtain the volume of the frozen pork chop. The calculation process for the average point spacing of the point cloud is as follows:

[0176] d p =min(dis(p,q)),q=1,2,…,p≠q

[0177] Where, d p Let represent the minimum distance between point p and other points, and dis(p,q) represent the distance between point p and any point p. Then, the formula for calculating the average point spacing D of the point cloud is:

[0178]

[0179] Where p represents any point in the point cloud, and N represents the number of points in the point cloud. In a preferred embodiment, the average point spacing D = 0.3 mm is used as the boundary: D ≤ 0.3 mm is considered to be a small average point spacing, and D > 0.3 mm is considered to be a large average point spacing.

[0180] After determining the total volume of the frozen pork chop, in order to obtain the target slice thickness, the mass of the frozen pork chop is further obtained using an electronic scale, and the given slice mass is set according to actual production needs.

[0181] Formula for calculating average density:

[0182]

[0183] ρ sample - Average density of frozen pork chops

[0184] V-Frozen Pork Chop Volume

[0185] M-Frozen Pork Chops Quality.

[0186] Assume the density of frozen pork chops is uniformly distributed, i.e., ρ sample =ρ slice , ρ sample - Average density of frozen pork chops.

[0187] slice volume V slice calculate:

[0188]

[0189] Among them, M slice This indicates the required slice quality for quantitative sectioning.

[0190] Calculation of slice thickness:

[0191] Based on the target slice volume V slice A greedy algorithm is used to calculate the slice thickness. Taking the minimum point of the X-axis of the bone-and-flesh 3D model as the starting point, the optimal slice thickness corresponding to this starting point is found, which is to satisfy the required volume V of the quantitative slice as much as possible. slice The slice thickness is then determined; the process is repeated, with the end position of this slice serving as the starting point for the next slice, and a stop criterion is set.

[0192] Figure 11 A flowchart of the slice thickness calculation method is given, X min The minimum value of the X-axis of the 3D model, X max X represents the maximum value of the 3D model. begin X is the starting value for the slice. end Here, W is the slice termination value, d and d1 are the slice thicknesses, V is the slice volume, V_target is the target slice volume, error is the absolute error of the slice volume, min-error is the minimum absolute error of the slice volume, and d_best is the optimal slice thickness. The pseudocode for the specific quantitative slice thickness calculation algorithm is as follows:

[0193]

[0194]

[0195] The method for calculating slice thickness based on a 3D reconstruction model can be divided into the following steps:

[0196] (1) Set the starting position X of the slice. begin Set the minimum X-axis value of the 3D model. min The initial slice thickness d was set to 10 mm. Point clouds within this thickness range were extracted, and the volume (V) of the slice point cloud was returned using a volume calculation method based on Graham scanning. q If V q Smaller than the required slice volume (V) for quantitative sectioning S If the slice thickness is increased by 10 mm, the slice volume V is recalculated. q until V q Greater than or equal to V S At that time, the slice thickness d at that time is taken as the candidate value.

[0197] (2) Define the slice thickness range as [d-10mm, d mm], and set the step size to 1mm. Compare the V values ​​under different slice thicknesses. q With V S The absolute difference between the values ​​is used to determine the optimal slice thickness d_best, with the slice thickness d being the smallest absolute difference. X is then set... begin =X begin +d_best.

[0198] (3) Repeat steps (1) and (2) to obtain the d_best of all slices of the sample and store them in the d_best_all list.

[0199] (4) When the initial position of the slice is greater than or equal to X max Or when the slice termination position is greater than or equal to X max When the loop ends, the loop terminates.

[0200] After the above steps, the 3D model of each sample will be divided into several slice point clouds, and the thickness value of the slice point cloud will be obtained. Then, the frozen pork chops will be sliced ​​according to the obtained slice point cloud thickness value.

[0201] The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which this invention pertains may make various modifications or additions to the described specific embodiments or use similar methods to substitute them, without departing from the spirit of the invention or exceeding the scope defined by the appended claims.

Claims

1. A system for quantitative slicing of bone-in meat, characterized by It includes a data acquisition device for slicing bone-in meat, a data processing device, and a slicing execution mechanism, wherein: The data acquisition device is used to collect point cloud data on the three-dimensional morphology of bone-in meat during quantitative slicing. The data processing device processes and calculates the collected point cloud data, including surface reconstruction of bone-in meat; and calculates slice thickness according to production requirements; the data processing device includes a point cloud preprocessing module, a 3D reconstruction module, a volume calculation module, and a slice thickness calculation module, wherein: Point cloud preprocessing module: By setting threshold ranges for the X, Y, and Z axes, point cloud data outside the threshold range is removed to achieve region of interest segmentation; 3D Reconstruction Module: Based on the processed point cloud data output by the point cloud preprocessing module, it realizes the reconstruction of the surface model with bone and flesh; Volume calculation module: Calculates the total volume of meat and bone based on the reconstructed surface model; Slice thickness calculation module: Calculates slice thickness based on given slice quality requirements; When the point cloud average point distance is small, the volume calculation module obtains the total volume of the bone meat by the following steps : S3.1 Surface area calculation is performed using the following formula: In the formula, Indicates the index of the triangular facet. This represents half the perimeter of the triangular face. This represents the area of ​​the triangular facet. These represent the lengths of each side of the triangular facet. This indicates the total number of triangular pieces. Indicates the surface area of ​​the reconstructed model; S3.2 Volume calculation is performed using the following formula: In the formula, express The distance between the triangular facet of the plane and the original point cloud. express Projected to The area of ​​each triangular facet of the plane. Indicates the total number of projected triangular faces; When the point cloud average point distance is large, the volume calculation module obtains the total volume of the bone and meat by the following steps : in, This represents the area of ​​the i-th slice. This represents the area of ​​the slice at layer (i+1). The interval between adjacent slices is represented by , and n represents the number of layers in the slice; where the slice area is... Obtained through the following formula: in, Indicates the number of vertices in the convex hull. This represents the y-coordinate of the i-th vertex. This represents the z-coordinate of the i-th vertex. This represents the y-coordinate of the (i+1)th vertex. This represents the z-coordinate of the (i+1)th vertex; The calculation process for the average point spacing of the point cloud is as follows: in, This represents the minimum distance between point p and other points. Let p represent the distance between point p and any point q, then the average point spacing of the point cloud is... The calculation formula is: wherein, denotes an arbitrary point in the point cloud, denotes the number of points in the point cloud; The slicing mechanism quantitatively slices the bone-in meat according to the calculated interval thickness; The slice thickness calculation module obtains the slice thickness through the following steps: S4.1 Average density is calculated using the following formula: wherein, represents the average density of the bone-in meat, represents the total volume of the bone-in meat, represents the bone-in meat mass obtained by weighing; density of the slice = ; S4.2, Slice volume Calculation: wherein, represents the slice quality required for quantization slicing; S4.3, Based on target slice volume A greedy algorithm is used to calculate the slice thickness, taking the minimum X-axis point of the bone-and-flesh 3D model as the starting point, and finding the optimal slice thickness corresponding to this starting point, that is, to meet the required volume of quantitative slices as much as possible. The slice thickness; S4.4 Then, using the end position of this slice as the starting value of the next slice, repeat the process and set the stop criterion.

2. The system according to claim 1, characterized in that... The data acquisition device includes a data acquisition module and an electronic control module, wherein: The data acquisition module includes: a conveyor belt, a scanner bracket, a scanner backplate, a line laser scanner, and a host computer. The line laser scanner is mounted on the scanner backplate, which is horizontally fixed on the scanner bracket and located above the conveyor belt. The line laser scanner is connected to the host computer via a data cable. When installing the line laser scanner, its laser emitter is positioned at a certain distance above, below, and to the left and right of the gap in the conveyor belt. The electrical control module includes: a PLC all-in-one unit, a servo motor and its driver, a motor synchronous pulse distributor, photoelectric switches, and a servo motor reducer. The PLC all-in-one unit, as the core component of the electrical control module, controls the conveyor belt's movement speed and the photoelectric switch's delay time by programming ladder diagrams. The servo motor and its driver are used to control the servo motor's speed, thereby changing the conveyor belt's running speed. The servo motor driver is equipped with differential signal lines and differential signal terminals for differential signal output. The motor synchronous pulse distributor synchronously distributes the differential signals output from the servo motor with its differential signal lines, providing them simultaneously to different line laser scanners. The photoelectric switch detects whether the bone-in tissue is within the scanning range of the line laser scanner, then feeds back a level signal to achieve synchronous acquisition by the line laser scanner. The servo motor reducer uses a gear speed converter to reduce the motor's rotation speed to the desired speed.

3. The system of claim 1, wherein The point cloud preprocessing module includes the following steps: S1.

1. Region of interest segmentation, performed using the following formula: Xmin <= X <= Xmax Ymin <= Y <= Ymax Zmin <= Z <= Zmax In the formula, Xmin, Xmax, Ymin, Ymax, Zmin, and Zmax represent the set threshold range of point cloud data coordinates; X, Y, and Z represent the X, Y, and Z axis coordinate ranges of the segmented and extracted point cloud data. S1.2 Statistical filtering is performed using the following formula: In the formula, Let these represent the coordinates of the nth point in the point cloud. These represent the coordinates of any point in the point cloud; This represents the nth point in the point cloud. Arrive at any point The distance; Indicates the average distance. Indicates standard deviation, This indicates the number of points in the point cloud; S1.3, Voxel downsampling, performed using the following formula: In the formula, represents the centroid coordinates of the voxel; m represents the number of points in the voxel; , , These represent the points in a voxel. , , Axis coordinate values.

4. The system of claim 1, wherein The 3D reconstruction module achieves 3D reconstruction with bone and flesh through the following steps: S2.1 Point cloud registration, including: (1) Find the nearest neighbor of the point in the reference point cloud Q corresponding to the point in the point cloud to be registered , and calculate the squared spatial distance between the two points ;​​ (2) According to the calculation Sort the data in ascending order, keep the first N pairs, and calculate their sum S, where N = kN. p ;k represents the point cloud overlap, N p N represents the number of points in the point cloud P to be registered, and N represents the number of points to be retained. The transformation matrix T = (R, t) is calculated using the following formula: In the formula, Represents the rotation matrix. Represents the translation matrix. Represents any point in the reference point cloud Q. Indicates the registration point cloud and The nearest neighbor; (3) Transform the point cloud to be registered according to the obtained transformation matrix, as shown in the following formula: wherein, denotes the transformed point cloud coordinates, denotes the point cloud coordinates before transformation, denotes a translation matrix; Repeat steps (1)-(3) to calculate the relationship between the point cloud P to be registered and the reference point cloud Q until the set iteration threshold is reached; S2.2 Point cloud fusion is performed using the following formula: P2 = P1 + TP In the formula, T represents the obtained transformation matrix, T= , Represents the rotation matrix. P1 represents the translation matrix; P2 represents the point cloud model; P1 represents the target point cloud; P represents the point cloud to be registered. S2.3 Parameter extraction, performed using the following formula: In the formula, It indicates the length of meat with bone. Indicates the width of meat with bone. It indicates that the meat with bones is high. Points in a 3D reconstruction model Maximum value Points in a 3D reconstruction model Minimum value.

5. The system of claim 1, wherein The slicing mechanism includes a conveyor belt, a transmission motor, and a cutter. The meat with bones is placed on the conveyor belt, and the transmission motor receives start / stop signals from the host computer to control the transmission of the conveyor belt; the cutter receives lifting signals from the host computer to perform the slicing action.

Images