A machine vision-based flexible intelligent air separation system for tobacco shreds

Abstract

This invention discloses a machine vision-based flexible intelligent air separation system for tobacco shreds, comprising: a feeding conveyor belt, an air separation box, a secondary separation box, a negative pressure fan, a frequency converter for the negative pressure fan, and a feeding conveyor belt motor. Normal tobacco shreds are conveyed under negative pressure by the negative pressure fan to the air separation box for image recognition, thus allowing qualified tobacco shreds to fall onto the conveyor belt at the bottom outlet of the air separation box for transport to the next process. Abnormal tobacco shreds are conveyed under negative pressure by the negative pressure fan to the secondary separation box for image recognition and secondary sorting, while the operating status of the feeding conveyor belt motor is monitored. This invention employs machine vision intelligent algorithms to identify the air separation conditions inside the air separator and feeds the identification results back to the air separator's execution system to achieve intelligent closed-loop control of the air separator, thereby improving air separation quality and efficiency, improving production processes, and ensuring the stability of tobacco shred products.

Description

Technical Field

[0001] This invention belongs to the field of machine vision technology, specifically a flexible intelligent air-separation system for tobacco shreds based on machine vision. Background Technology

[0002] The air separator is one of the equipment used in the cigarette manufacturing production line to separate materials. It is used for different processing techniques of tobacco materials such as tobacco sheets, tobacco shreds, tobacco stems, stem shreds, and expanded tobacco.

[0003] Currently, the control method for air separators basically relies on sensors to obtain the difference between the negative pressure value inside the air separator and the preset negative pressure value when the tobacco leaves reach the air separator inlet, which is used to manually adjust the opening degree of the air separator's damper. The patent "An Image Monitoring Device for a Tobacco Air Separator and a Tobacco Air Separator" (Authorization No.: CN 216500744 U) addresses the issue that "Currently, the negative pressure airflow control of air separators in cigarette manufacturing production lines in my country is mostly done manually. During manual adjustment, it is necessary to frequently observe the discharge from the waste outlet and the material outlet of the tobacco air separator to further determine the composition of the discharge. However, the manual movement between the waste outlet and the material outlet takes time, and the observation effect is not timely enough, therefore the judgment of airflow adjustment is not accurate enough." The patent proposes a method to use an image monitoring device to simultaneously monitor both the material outlet and the waste outlet of the tobacco air separator, and observe them simultaneously on the same display. This method does indeed improve the difficulty of adjusting the damper of the original air classifier, and it is more direct than directly controlling the negative pressure value of the air duct, because the final effect of the air classifier is to see the quality of the material at the discharge port. However, this method still requires human judgment of the material and waste at the outlet, and it lacks real-time and continuous operation, and is not intelligent.

[0004] The main drawbacks of current sorting machines are as follows:

[0005] Existing air separators are all open-loop controlled, and once the air outlet and air pressure are adjusted, they cannot be adjusted in a timely manner according to changes in actual working conditions.

[0006] Even if some devices employ vision, this vision only makes it more convenient for the observer to observe, and does not have intelligent recognition capabilities. Summary of the Invention

[0007] The present invention addresses the shortcomings of the prior art by proposing a flexible intelligent air separation system for tobacco shreds based on machine vision. This system aims to identify the sorting status within the air separator in real time and continuously, and feed the identification results back to the air separator's actuator. This enables the final closed-loop intelligent control of the air separator, thereby improving production efficiency, refining the production process, and ensuring the stability of the tobacco shreds.

[0008] To achieve the above-mentioned objectives, the present invention adopts the following technical solution:

[0009] The present invention provides a flexible intelligent air separation system for tobacco shreds based on machine vision, which includes: a feeding conveyor belt, an air separation box, a secondary separation box, a negative pressure fan, a negative pressure fan frequency converter, and a feeding drive belt motor.

[0010] The feeding conveyor belt is driven by a motor, transporting the tobacco shreds on it to the inlet of the air-separation box. A secondary separation box is located at the bottom of the air-separation box and at the inlet. A negative pressure fan provides different intensities of negative pressure air to the air-separation box and the secondary separation box, and the negative pressure value of the negative pressure fan is controlled by a frequency converter. Normal tobacco shreds on the feeding conveyor belt are transported to the air-separation box under the negative pressure of the fan, while abnormal tobacco shreds are transported to the secondary separation box under the negative pressure of the fan. A first CCD linear array sensor is installed in the air-separation box; a second CCD linear array sensor is installed in the secondary separation box. The first CCD linear array sensor is used to capture an image of the current normal tobacco shreds under negative pressure. F The system is used to determine whether the negative pressure value inside the air separation box is within the set negative pressure range. If it is, the current negative pressure value of the air separation box remains unchanged; if it is less than the negative pressure range, the current negative pressure value of the air separation box is increased; if it exceeds the negative pressure range, the current negative pressure value of the air separation box is decreased. This allows qualified tobacco shreds within the negative pressure range to fall onto the conveyor belt at the bottom outlet of the air separation box for transport to the next process. A tobacco shredding mechanism is installed inside the secondary separation box to shred abnormal tobacco shreds. The shredded tobacco shreds are blown into the air separation box under negative pressure for air separation, and the remaining portion falls to the bottom of the secondary separation box. The second CCD linear array sensor is used to capture images of the current abnormal tobacco shreds at the bottom of the secondary separation box. E It is used to determine whether the conveying speed of the feeding conveyor belt exceeds the speed threshold. If it does, an audible and visual alarm is triggered and the speed of the feeding conveyor belt motor is reduced. Otherwise, it indicates that the secondary separation box is working normally.

[0011] The machine vision-based flexible intelligent air-sorting system for tobacco shreds described in this invention is also characterized in that the current normal tobacco shred image I... F The judgment is made according to the following process:

[0012] Step 1: Calculate the current normal tobacco image I using the Lucas-Kanade algorithm. F Any i-th pixel p in the optical flow field i velocity V(x) i ,y i )={v i,x ,v i,y}, where (x i ,yi Image I represents the current normal tobacco shreds. F medium pixel p i Position coordinates, v i,x and v i,y p, representing the i-th pixel. i Velocity values ​​in the horizontal and vertical directions;

[0013] Step 2: Determine whether equation (1) is true. If it is true, set the i-th pixel p... i velocity V(x) i ,y i If the i-th pixel is not added to the velocity set VC, then discard the i-th pixel p. i velocity V(x) i ,y i );

[0014]

[0015] In equation (1), δ represents the angle threshold, and θ i p represents the i-th pixel. i The angle is obtained from equation (2):

[0016]

[0017] Step 3: Process each pixel in the optical flow field according to the process in Step 2 to obtain the final velocity set VC;

[0018] Step 4: Calculate the average value Vmean of the velocity set VC using equation (3):

[0019]

[0020] In equation (3), K is the number of velocity values ​​of pixels in the velocity set VC;

[0021] Step 5: If Vmean < ε1, it means that the current negative pressure value of the air separation box is too low, and the current negative pressure value of the air separation box is increased. If Vmean > ε2, it means that the current negative pressure value of the air separation box is too high, and the current negative pressure value of the air separation box is decreased. Otherwise, it means that the current negative pressure value of the air separation box is reasonable, and the current negative pressure value of the air separation box is kept unchanged. Here, ε1 and ε2 represent the minimum and maximum speed thresholds, respectively.

[0022] The current abnormal tobacco I E The judgment is made according to the following process:

[0023] Step a: Use equation (4) to process the current abnormal image I EThe RGB components of any j-th pixel are converted to the YCbCr color space, and the YCbCr components of the j-th pixel are obtained, thus obtaining the current abnormal image I. E YCbCr spatial image Y E ;

[0024]

[0025] In equation (4), Y j This represents the Y component of the YCbCr color space at the j-th pixel, where Cb is the color space component. j This represents the Cb component of the YCbCr color space at the j-th pixel, where Cr is the Cb component. j R represents the Cr component of the YCbCr color space at the j-th pixel. j G represents the R component of the RGB color space of the j-th pixel. j Let G and B represent the RGB color space components of the j-th pixel. j This represents the B component of the RGB color space of the j-th pixel, and the operator ">>" indicates a right shift.

[0026] Step b: Calculate the YCbCr spatial image Y E The local binary pattern value L of any j-th pixel in the dataset j (x j,c ,y j,c ), where (x j,c ,y j,c Let be the image coordinates of the j-th pixel;

[0027] Step c, Y E Divide the code into several sub-blocks of size N×N, and set the flag of any nth sub-block as flag. n ;

[0028] Step d: Calculate the histogram of N×N local binary pattern values ​​in the nth sub-block, and sort the N×N histograms in the nth sub-block in ascending order to obtain the sorted histogram of the nth sub-block, sortHist. n ;

[0029] Step e: Use equation (5) to sort the histogram of the nth sub-block (sortHist). n By fitting the data, we obtain the combined exponential function h of the nth sub-block. n (m):

[0030]

[0031] In equation (5), λ 1,n and λ 2,n They are exponential functions h n(m) has two parameters; m represents the independent variable;

[0032] Step f: If equation (6) holds, then the nth sub-block is an abnormal sub-block, and the flag is set. n Set it to 1; otherwise, it indicates that the nth sub-block is a normal sub-block, and the flag is set. n Set to 0;

[0033]

[0034] In equation (6), λ 1,L and λ 1,U These are parameters λ 1,n The lower and upper limits of λ; 2,L and λ 2,U These are parameters λ 2,n The lower and upper limits;

[0035] Step g: Based on the labels of all sub-blocks, calculate Y. E The number of abnormal sub-blocks, num, indicates that the conveyor belt speed is too fast and the speed of the conveyor belt motor should be reduced. Otherwise, it indicates that the conveyor belt speed is normal. M is the maximum number of abnormal sub-blocks. Otherwise, it indicates that the secondary separation box is working normally and the speed of the conveyor belt motor should be kept constant.

[0036] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0037] 1. This invention uses the Lucas-Kanade algorithm to calculate the optical flow field, which is fast and can meet the requirements of online real-time processing;

[0038] 2. In this invention, only velocity points that meet certain conditions are included in the velocity set VC, thereby eliminating the influence of optical flow field noise points;

[0039] 3. This invention uses local binary mode values ​​to extract local image descriptive information, resulting in good stability.

[0040] 4. This invention utilizes an exponential function to fit the sorted local binary pattern histogram, resulting in good descriptive accuracy;

[0041] 5. This invention improves equipment reliability and reduces production accidents caused by equipment failure;

[0042] 6. This invention provides an intelligent upgrade to the current tobacco air separator, reducing labor intensity. Attached Figure Description

[0043] Figure 1 The present invention divides the image into several sub-blocks of size N×N; Detailed Implementation

[0044] In this embodiment, a machine vision-based flexible intelligent air separation system for tobacco shreds is described. A feeding conveyor belt is driven by a motor, transporting the tobacco shreds to the inlet of an air separation box. A secondary separation box is located at the bottom of the air separation box and near the inlet. A negative pressure fan provides negative pressure air of varying intensities to both the air separation box and the secondary separation box, with the negative pressure value controlled by a frequency converter. Normal tobacco shreds on the feeding conveyor belt are transported to the air separation box under the negative pressure of the fan, while abnormal tobacco shreds are transported to the secondary separation box under the same negative pressure. A first CCD linear array sensor is installed in the air separation box. Abnormal tobacco shreds, generally due to their heavy clumps, cannot pass through the negative pressure fan to enter the air separation box and instead fall into the secondary separation box for further processing.

[0045] A second CCD linear array sensor is used inside the secondary separation chamber; the first CCD linear array sensor is used to capture images of normal tobacco shreds under negative pressure. F This is used to determine whether the negative pressure value inside the air separation box is within the set negative pressure range. If it is, the current negative pressure value of the air separation box remains unchanged; if it is less than the negative pressure range, the current negative pressure value of the air separation box is increased; if it exceeds the negative pressure range, the current negative pressure value of the air separation box is decreased. This allows qualified tobacco shreds within the negative pressure range to fall onto the conveyor belt at the bottom outlet of the air separation box for transport to the next process. A tobacco shredding mechanism is installed inside the secondary separation box to shred abnormal tobacco shreds. The shredded tobacco shreds are then blown into the air separation box under negative pressure for air separation, with the remaining portion falling to the bottom of the secondary separation box. A second CCD linear array sensor is used to capture images of the current abnormal tobacco shreds at the bottom of the secondary separation box. E It is used to determine whether the conveying speed of the feeding conveyor belt exceeds the speed threshold. If it does, an audible and visual alarm is triggered and the speed of the feeding conveyor belt motor is reduced. Otherwise, it indicates that the secondary separation box is working normally.

[0046] In this embodiment, image I of normal tobacco shreds inside the air separator box F The judgment is made according to the following process:

[0047] Step 1: Calculate the current normal tobacco image I using the Lucas-Kanade algorithm. F Any i-th pixel p in the optical flow field i velocity V(x) i ,y i )={v i,x ,v i,y}, where (x i ,y i Image I represents the current normal tobacco shreds. F medium pixel p i Position coordinates, vi,x and v i,y p, representing the i-th pixel. i The speed values ​​in both the horizontal and vertical directions are processed using the Lucas-Kanade algorithm, which offers advantages such as high processing speed, meeting the real-time processing requirements of production lines. Another advantage is the algorithm's maturity, stability, and ease of implementation.

[0048] Step 2: Determine whether equation (1) is true. If it is true, set the i-th pixel p... i velocity V(x) i ,y i If the i-th pixel is not added to the velocity set VC, then discard the i-th pixel p. i velocity V(x) i ,y i This step requires the speed to meet a certain angle range, which can filter out some of the noise.

[0049]

[0050] In equation (1), δ represents the angle threshold, and θ i p represents the i-th pixel. i The angle is obtained from equation (2):

[0051]

[0052] Step 3: Process each pixel in the optical flow field according to the process in Step 2 to obtain the final velocity set VC;

[0053] Step 4: Calculate the average value Vmean of the velocity set VC using equation (3):

[0054]

[0055] In equation (3), K is the number of velocity values ​​of pixels in the velocity set VC;

[0056] Step 5: If Vmean < ε1, it indicates that the tobacco movement in the image is too slow, meaning the current negative pressure value of the air separation box is too low. The system will increase the current negative pressure value of the air separation box. If Vmean > ε2, it indicates that the tobacco movement in the image is too fast, meaning the current negative pressure value of the air separation box is too high. The system will decrease the current negative pressure value of the air separation box. Otherwise, it indicates that the current negative pressure value of the air separation box is reasonable, and the current negative pressure value of the air separation box will remain unchanged. Here, ε1 and ε2 represent the minimum and maximum speed thresholds, respectively.

[0057] If there is an excessive accumulation of abnormal tobacco shreds at the bottom of the secondary separation chamber, it indicates that the conveyor belt is moving too fast and needs to be adjusted. Specifically, the abnormal tobacco shreds in the secondary separation chamber are identified as follows: EThe judgment is made according to the following process:

[0058] Step a: Use equation (4) to process the current abnormal image I E The RGB components of any j-th pixel are converted to the YCbCr color space, and the YCbCr components of the j-th pixel are obtained, thus obtaining the current abnormal image I. E YCbCr spatial image Y E Since the speed of computer shift operations is much faster than that of multiplication operations, the advantage of using equation (4) for color space conversion is that it is fast and suitable for real-time processing on the production line.

[0059]

[0060] In equation (4), Y j This represents the Y component of the YCbCr color space at the j-th pixel, where Cb is the color space component. j This represents the Cb component of the YCbCr color space at the j-th pixel, where Cr is the Cb component. j R represents the Cr component of the YCbCr color space at the j-th pixel. j G represents the R component of the RGB color space of the j-th pixel. j Let G and B represent the RGB color space components of the j-th pixel. j This represents the B component of the RGB color space of the j-th pixel, and the operator ">>" indicates a right shift.

[0061] Step b: Calculate the YCbCr spatial image Y E The local binary pattern value L of any j-th pixel in the dataset j (x j,c ,y j,c ), where (x j,c ,y j,c ) represents the image coordinates of the j-th pixel; Local Binary Mode has the feature of effectively extracting local textures of an image and is suitable for processing tobacco features.

[0062] Step c, Y E Divide into several sub-blocks of size N×N, such as Figure 1 As shown, the flag of any nth sub-block is set as flag. n ;

[0063] Step d: Calculate the histogram of N×N local binary pattern values ​​in the nth sub-block, and sort the N×N histograms in the nth sub-block in ascending order to obtain the sorted histogram of the nth sub-block, sortHist. n ;

[0064] Step e: Use equation (5) to sort the histogram of the nth sub-block (sortHist). n By fitting the data, we obtain the combined exponential function h of the nth sub-block. n (m):

[0065]

[0066] In equation (5), λ 1,n and λ 2,n They are exponential functions h n (m) has two parameters; m represents the function h n The independent variable is (m). The curve characteristics of the exponential function are suitable for the sorted histogram in this invention. By adjusting the two parameters, the sorted histogram can be accurately fitted, and the normality of the current sub-block can be identified.

[0067] Step f: If equation (6) holds, then the nth sub-block is an abnormal sub-block, and the flag is set. n Set it to 1; otherwise, it indicates that the nth sub-block is a normal sub-block, and the flag is set. n Set to 0;

[0068]

[0069] In equation (6), λ 1,L and λ 1,U These are parameters λ 1,n The lower and upper limits of λ; 2,L and λ 2,U These are parameters λ 2,n The lower and upper limits;

[0070] Step g: Based on the labels of all sub-blocks, calculate Y. E The number of abnormal sub-blocks, num, indicates that the conveyor belt speed is too fast and the speed of the conveyor belt motor should be reduced. Otherwise, it indicates that the conveyor belt speed is normal. M is the maximum number of abnormal sub-blocks. Otherwise, it indicates that the secondary separation box is working normally and the speed of the conveyor belt motor should be kept constant.

Claims

1. A flexible intelligent air-separation system for tobacco shreds based on machine vision, characterized in that, include: Feeding conveyor belt, air separation box, secondary separation box, negative pressure fan, negative pressure fan frequency converter, feeding conveyor belt motor; The feeding conveyor belt is driven by a motor, which transports the tobacco shreds on the conveyor belt to the inlet of the air-separation box. A secondary separation box is located at the bottom of the air-separation box and at the inlet. A negative pressure fan provides negative pressure air of different intensities to the air-separation box and the secondary separation box, and the negative pressure value of the negative pressure fan is controlled by a frequency converter. Normal tobacco shreds on the feeding conveyor belt are transported to the air-separation box under the negative pressure of the fan, while abnormal tobacco shreds are transported to the secondary separation box under the negative pressure of the fan. A first CCD linear array sensor is installed in the air-separation box; a second CCD linear array sensor is installed in the secondary separation box. The first CCD linear array sensor is used to capture an image of the current normal tobacco shreds under negative pressure. F The system is used to determine whether the negative pressure value inside the air separation box is within the set negative pressure range. If it is, the current negative pressure value of the air separation box remains unchanged; if it is less than the negative pressure range, the current negative pressure value of the air separation box is increased; if it exceeds the negative pressure range, the current negative pressure value of the air separation box is decreased. This allows qualified tobacco shreds within the negative pressure range to fall onto the conveyor belt at the bottom outlet of the air separation box for transport to the next process. A tobacco shredding mechanism is installed inside the secondary separation box to shred abnormal tobacco shreds. The shredded tobacco shreds are blown into the air separation box under negative pressure for air separation, and the remaining portion falls to the bottom of the secondary separation box. The second CCD linear array sensor is used to capture an image of the current abnormal tobacco shreds at the bottom of the secondary separation box. E This is used to determine whether the conveyor speed of the feeding conveyor exceeds a speed threshold. If it does, an audible and visual alarm is triggered, and the speed of the feeding conveyor motor is reduced. Otherwise, it indicates that the secondary separation box is working normally, and the speed of the feeding conveyor motor remains unchanged. The current normal tobacco image I F The judgment is made according to the following process: Step 1: Calculate the current normal tobacco image I using the Lucas-Kanade algorithm. F Any i-th pixel p in the optical flow field i velocity V(x) i , y i )={v i,x , v i,y }, where (x i , y i Image I represents the current normal tobacco shreds. F medium pixel p i Position coordinates, v i,x and v i,y p, which is the i-th pixel. i Velocity values ​​in the horizontal and vertical directions; Step 2: Determine whether equation (1) is true. If it is true, set the i-th pixel p... i velocity V(x) i , y i If the i-th pixel is not added to the velocity set VC, then the i-th pixel p is discarded. i velocity V(x) i , y i ); (1) In equation (1), Indicates the angle threshold. p represents the i-th pixel. i The angle is obtained from equation (2): (2) Step 3: Process each pixel in the optical flow field according to the process in Step 2 to obtain the final velocity set VC; Step 4: Calculate the average value Vmean of the velocity set VC using equation (3): (3) In equation (3), K is the number of velocity values ​​of pixels in the velocity set VC; Step 5: If Vmean < ε1, it means that the current negative pressure value of the air separation box is too low, and the current negative pressure value of the air separation box is increased. If Vmean > ε2, it means that the current negative pressure value of the air separation box is too high, and the current negative pressure value of the air separation box is decreased. Otherwise, it means that the current negative pressure value of the air separation box is reasonable, and the current negative pressure value of the air separation box is kept unchanged. Here, ε1 and ε2 represent the minimum and maximum speed thresholds, respectively.

2. The machine vision-based flexible intelligent air-separation system for tobacco shreds according to claim 1, characterized in that, The current abnormal tobacco image I E The judgment is made according to the following process: Step a: Use equation (4) to generate the current abnormal tobacco image I. E The RGB components of any j-th pixel are converted to the YCbCr color space, and the YCbCr components of the j-th pixel are obtained, thus obtaining the current abnormal tobacco image I. E YCbCr spatial image Y E ; (4) In equation (4), This represents the Y component of the YCbCr color space at the j-th pixel. This represents the Cb component of the YCbCr color space at the j-th pixel. This represents the Cr component of the YCbCr color space at the j-th pixel. This represents the R component of the RGB color space for the j-th pixel. This represents the G component of the RGB color space for the j-th pixel. The operator "" represents the B component of the RGB color space of the j-th pixel. " indicates a right shift; Step b: Calculate the YCbCr spatial image Y E Local binary pattern value of any j-th pixel , where (x j,c , y j,c Let be the image coordinates of the j-th pixel; Step c, Y E Divide the code into several sub-blocks of size N×N, and set the flag of any nth sub-block as flag. n ; Step d: Calculate the histogram of N×N local binary pattern values ​​in the nth sub-block, and sort the N×N histograms in the nth sub-block in ascending order to obtain the sorted histogram of the nth sub-block, sortHist. n ; Step e: Use equation (5) to sort the histogram of the nth sub-block (sortHist). n By fitting the data, we obtain the combined exponential function of the nth sub-block. : (5) In equation (5), λ 1,n and λ 2,n Exponential functions The two parameters; m represents the independent variable; Step f: If equation (6) holds, then the nth sub-block is an abnormal sub-block, and the flag is set. n Set it to 1; otherwise, it indicates that the nth sub-block is a normal sub-block, and the flag is set. n Set to 0; (6) In equation (6), and These are parameters λ 1,n The lower and upper limits; and These are parameters λ 2,n The lower and upper limits; Step g: Based on the labels of all sub-blocks, calculate Y. E The number of abnormal sub-blocks is num. If num is greater than M, the conveyor belt speed is too fast, and the conveyor belt motor speed is reduced. Otherwise, it means that the conveyor belt speed is normal, the secondary separation box is working normally, and the conveyor belt motor speed is kept constant. M is the maximum number of abnormal sub-blocks.

Images