Method and system for measuring three-dimensional flow velocity of flowing tobacco shreds in tobacco shred drying cavity

Abstract

The invention relates to the technical field of cigarette technology and detection, in particular to a method and system for measuring the three-dimensional flow velocity of flowing tobacco shreds in a tobacco shred drying cavity, and the method comprises the steps: obtaining a multi-angle image of the flowing tobacco shreds; preprocessing the multi-angle image, and taking the processed image as an original image; detecting the original image by adopting a feature point detection algorithm, detecting all tobacco shreds and a plurality of corresponding feature points in the original image, and obtaining a tobacco shred-feature point data set; and calculating the three-dimensional flow velocity of the plurality of feature points corresponding to each tobacco shred in the tobacco shred-feature point data set, and obtaining the three-dimensional flow velocity of each tobacco shred according to the three-dimensional flow velocity of the plurality of feature points corresponding to each tobacco shred. According to the technical method, the flow velocity of the flowing tobacco shreds in the tobacco shred drying cavity is accurately measured by adopting the optical flow algorithm and the three-dimensional velocity vector method. Compared with the prior art, the measurement precision and reliability are remarkably improved, and the device has the advantages of non-contact, real-time monitoring, data storage and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of cigarette manufacturing processes and detection technologies, and particularly to a method and system for measuring the three-dimensional flow velocity of flowing tobacco shreds in a drying chamber. Background Art

[0002] In modern tobacco processing, the drying process of tobacco shreds is a crucial step, which directly affects the quality and taste of the final tobacco shreds. During the drying process, the flow state and velocity of tobacco shreds in the drying chamber have a significant impact on the moisture evaporation, release of aroma substances, and maintenance of structure of the tobacco shreds. Therefore, accurately measuring the flow velocity of flowing tobacco shreds in the drying chamber is of great significance for optimizing the drying process parameters and improving the quality of tobacco shreds.

[0003] Traditional flow velocity measurement methods, such as using an anemometer or a hot-wire anemometer, are often difficult to accurately measure in a high-temperature, high-humidity, and sealed drying chamber due to the limitations of contact measurement. In addition, these methods may interfere with the flow of tobacco shreds and affect the accuracy of measurement.

[0004] To overcome these challenges, non-contact optical measurement methods have gradually attracted attention. The optical flow method is a non-contact flow velocity measurement method that calculates the motion velocity of an object by analyzing the brightness changes of pixels in consecutive image frames. The optical flow method has the advantages of high precision, fast response, and no interference, and is suitable for measuring the flow velocity of tobacco shreds in a complex environment in the drying chamber. However, the optical flow measurement from a single angle can only obtain the velocity information on a two-dimensional plane and is difficult to comprehensively reflect the flow state of tobacco shreds in three-dimensional space. Therefore, how to effectively perform multi-angle and three-dimensional flow velocity measurement is a challenge faced by the current technical field. Summary of the Invention

[0005] To solve the above technical problems, the object of the present invention is to provide a method and system for measuring the three-dimensional flow velocity of flowing tobacco shreds in a drying chamber. The method and system capture image sequences of flowing tobacco shreds from different angles by arranging high-speed imaging devices at multiple angles in the drying chamber, and use the optical flow algorithm to calculate the velocity vectors at each viewing angle, so as to obtain the motion information of tobacco shreds at different viewing angles. Further, through multi-angle data fusion, a three-dimensional flow velocity distribution of tobacco shreds is constructed. This multi-angle optical flow measurement method overcomes the limitations of traditional measurement methods and can measure the flow velocity of flowing tobacco shreds in the drying chamber in real time and accurately without disturbing the flow of tobacco shreds, comprehensively reflecting the flow state of tobacco shreds in three-dimensional space and providing accurate data support for process parameter optimization.

[0006] To achieve the above object, an embodiment of the present invention provides a method for measuring the three-dimensional flow velocity of flowing tobacco shreds in a drying chamber, including:

[0007] Obtaining multi-angle images of flowing tobacco shreds;

[0008] Preprocess the multi - angle image and use the processed image as the original image;

[0009] Use a feature - point detection algorithm to detect the original image, detect all the cut tobacco and the corresponding multiple feature points in the original image, and obtain a cut - tobacco - feature - point data set;

[0010] Calculate the three - dimensional flow velocity of the multiple feature points corresponding to each cut tobacco in the cut - tobacco - feature - point data set, and obtain the three - dimensional flow velocity of each cut tobacco according to the three - dimensional flow velocity of the multiple feature points corresponding to each cut tobacco.

[0011] Optionally, obtaining multi - angle images of flowing cut tobacco includes:

[0012] Preset high - speed camera devices at multiple shooting angles to shoot the flowing cut tobacco from multiple angles;

[0013] Obtain the captured images of each high - speed camera device with the same frame rate and synchronized shooting time to obtain the multi - angle image sequence.

[0014] Optionally, preprocessing the multi - angle image includes:

[0015] Perform grayscale operation, denoising operation and enhancement operation on the multi - angle image.

[0016] Optionally, using a feature - point detection algorithm to detect the original image, detect all the cut tobacco and the corresponding multiple feature points in the original image, and obtain a cut - tobacco - feature - point data set, including:

[0017] Convolve the original image with a Gaussian kernel according to formula (1) to obtain Gaussian blurred images of each layer,

[0018] L(x,y,σ) = G(x,y,σ)*I(x,y), (1)

[0019] where L(x,y,σ) is the Gaussian blurred image at scale σ, G(x,y,σ) is the Gaussian kernel function, I(x,y) is the original image, σ is the scale parameter used to control the degree of Gaussian blur, and the Gaussian kernel function is the following formula (2),

[0020]

[0021] Obtain a difference image according to formula (3),

[0022] D(x,y,σ) = L(x,y,kσ)-L(x,y,σ), (3)

[0023] Among them, D(x, y, σ) is the difference image at scale σ, L(x, y, kσ) is the Gaussian blurred image at scale kσ, and k is a constant used to control the ratio between adjacent scales;

[0024] Compare the value of each pixel point in the difference image with the values of 26 pixel points adjacent to it in space and scale. If the value of this point is the maximum or minimum among the values of its 26 adjacent pixel points, then this point is a feature point;

[0025] Solve the exact position of the feature point according to formula (4),

[0026]

[0027] Among them, is the displacement vector of the feature point, and H is the Hessian matrix used to describe the second-order derivative information around the feature point, is the gradient vector of the difference image;

[0028] Calculate the main direction of the feature point according to formula (5),

[0029]

[0030] Among them, L x is the image gradient in the x direction of the Gaussian blurred image, and L y is the image gradient in the y direction of the Gaussian blurred image;

[0031] Divide the pixel area around the feature point according to the main direction of the feature point to generate a feature descriptor with direction invariance.

[0032] Optionally, calculate the three-dimensional flow velocity of multiple feature points corresponding to each cut tobacco in the cut tobacco-feature point dataset, and obtain the three-dimensional flow velocity of each cut tobacco according to the three-dimensional flow velocities of the multiple feature points corresponding to each cut tobacco, including:

[0033] Obtain an undetected cut tobacco in the cut tobacco-feature point dataset as the cut tobacco to be measured;

[0034] Judge whether the feature points of the cut tobacco to be measured flow in one direction;

[0035] In the case of judging that the feature points of the cut tobacco to be measured flow in one direction, use the optical flow algorithm to obtain the velocities of multiple feature points of the cut tobacco to be measured in the two-dimensional direction, and calculate the average velocity of each direction of the multiple feature points of the cut tobacco to be measured as the three-dimensional flow velocity of the cut tobacco to be measured;

[0036] When it is determined that the characteristic points of the to-be-tested cut tobacco do not flow in one direction, the three-dimensional velocity vectors are used to obtain the velocities of multiple characteristic points of the to-be-tested cut tobacco in three-dimensional directions, and the average velocity of each direction of the multiple characteristic points of the to-be-tested cut tobacco is calculated as the three-dimensional flow velocity of the to-be-tested cut tobacco;

[0037] Determine whether there is still untested cut tobacco in the cut tobacco - characteristic point dataset;

[0038] When it is determined that there is still untested cut tobacco in the cut tobacco - characteristic point dataset, return to execute the step of obtaining an untested cut tobacco in the cut tobacco - characteristic point dataset as the to-be-tested cut tobacco;

[0039] When it is determined that there is no untested cut tobacco in the cut tobacco - characteristic point dataset, the calculated three-dimensional flow velocity is used as the three-dimensional flow velocity of each corresponding cut tobacco.

[0040] Optionally, determining whether the characteristic points of the to-be-tested cut tobacco flow in one direction includes:

[0041] Determine whether the proportion of the velocity component in the dominant direction of the characteristic points of the to-be-tested cut tobacco in the total velocity is higher than n%;

[0042] When it is determined that the proportion of the velocity component in the dominant direction of the characteristic points of the to-be-tested cut tobacco in the total velocity is higher than n%, then the characteristic points of the to-be-tested cut tobacco flow in one direction;

[0043] When it is determined that the proportion of the velocity component in the dominant direction of the characteristic points of the to-be-tested cut tobacco in the total velocity is not higher than n%, then the characteristic points of the to-be-tested cut tobacco do not flow in one direction.

[0044] Optionally, using the optical flow algorithm to obtain the velocities of multiple characteristic points of the to-be-tested cut tobacco in two-dimensional directions includes:

[0045] Construct the basic optical flow equation according to formula (6),

[0046] I(x,y,t) = I(x + Δx,y + Δy,t + Δt), (6)

[0047] where I(x,y,t) is the image brightness function;

[0048] Determine the optical flow constraint equation according to formula (7),

[0049]

[0050] where u is the component of the optical flow velocity vector in the x direction, v is the component of the optical flow velocity vector in the y direction, is the change rate of the image brightness over time, is the gradient of the image brightness in the x direction, is the gradient of the image brightness in the y direction;

[0051] Calculate the gradient of the original image at the feature point in the x direction and the gradient in the y direction Calculate the rate of change of brightness over time based on the difference between two frames

[0052] Construct an optical flow constraint equation system for multiple pixel points in the local window of the original image;

[0053] Minimize the optical flow constraint equation system using the least squares method, and determine the velocity vector of the feature point of the to-be-detected cut tobacco according to formulas (8) to (10),

[0054]

[0055] where A is the brightness gradient matrix, and b is the negative vector of the time gradient, is the gradient of pixel point I 1 in the x direction, is the gradient of pixel point I 1 in the y direction, is the gradient of pixel point I 2 in the x direction, is the gradient of pixel point I 2 in the y direction, is the gradient of pixel point I n in the x direction, is the gradient of pixel point I n in the y direction, is the gradient of pixel point I 1 rate of change over time, is the rate of change over time of pixel point I 2 rate of change over time, is the rate of change over time of pixel point I n rate of change over time.

[0056] Optionally, use a three-dimensional velocity vector to obtain the velocities of multiple feature points of the to-be-detected cut tobacco in three dimensions, including:

[0057] Determine the three-dimensional position of the feature point P of the to-be-detected cut tobacco;

[0058] Calculate the velocity vector of the feature point of the to-be-detected cut tobacco according to formula (11),

[0059]

[0060] where ΔP is the three-dimensional position change of the feature point P of the to-be-detected cut tobacco in consecutive frames, and Δt is the time difference between consecutive frames;

[0061] Fuse the multi-view velocity vectors according to formula (12) to obtain the three-dimensional flow velocity of the cut tobacco.

[0062]

[0063] where V k,x is the x-direction component of the velocity vector of the k-th view, V k,y is the y-direction component of the velocity vector of the k-th view, V k,z is the z-direction component of the velocity vector of the k-th view, ω k is the weight of the k-th view.

[0064] Optionally, determining the three-dimensional position of the feature point P of the cut tobacco to be measured includes:

[0065] Determine the three-dimensional position of the feature point of the cut tobacco to be measured according to formula (13).

[0066]

[0067] where P is the feature point of the cut tobacco to be measured, and its three-dimensional point position is (x, y, z), and argmin P represents selecting the value of P with the minimum sum of squared errors among all possible P, p i is the position of the two-dimensional image point observed in the i-th view, P i is the projection matrix of the i-th view, P i,3 is the third row of the projection matrix P i , p j is the position of the two-dimensional image point observed in the j-th view, P j is the projection matrix of the j-th view, P j,3 is the third row of the projection matrix P j for normalization processing.

[0068] On the other hand, the present invention also provides a measurement system for the three-dimensional flow velocity of the cut tobacco flowing in the cut tobacco drying cavity. The measurement system includes a processor, and the processor is used to execute the measurement method as described in any one of the above.

[0069] Advantages of the present invention:

[0070] In the embodiment of the present invention, high-speed imaging devices are arranged at multiple angles inside the cut tobacco drying cavity to capture image sequences of the flowing cut tobacco from different angles, and the optical flow algorithm is used to calculate the velocity vectors at each viewing angle, so that the motion information of the cut tobacco at different viewing angles can be obtained. Further, through multi-angle data fusion, the three-dimensional flow velocity distribution of the cut tobacco can be accurately constructed. Compared with the prior art, this multi-angle optical flow measurement method greatly improves the accuracy and spatial resolution of the flow velocity measurement, can provide more detailed and reliable flow information than the traditional method, comprehensively reflects the flow state of the cut tobacco in the three-dimensional space, and provides accurate data support for the optimization of process parameters.

[0071] The embodiment of the present invention adopts the optical flow method as the core measurement technology, and captures images externally through a high-speed imaging device, realizing a completely non-contact measurement. Compared with the prior art, this non-contact measurement method effectively avoids interfering with the flow of the cut tobacco, ensuring the accuracy of the measurement results and the durability of the equipment.

[0072] Other features and advantages of the present invention will be described in detail in the subsequent specific embodiment part. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] The drawings are used to provide a further understanding of the embodiments of the present invention, and constitute a part of the specification, and are used to explain the embodiments of the present invention together with the following specific embodiments, but do not constitute a limitation to the embodiments of the present invention. In the drawings:

[0074] Figure 1 is a flowchart of a method for measuring the three-dimensional flow velocity of flowing cut tobacco in a cut tobacco drying cavity according to an embodiment of the present invention;

[0075] Figure 2 is a flowchart of a method for detecting cut tobacco feature points according to an embodiment of the present invention;

[0076] Figure 3 is a flowchart of a method for calculating the three-dimensional flow velocity of cut tobacco according to an embodiment of the present invention;

[0077] Figure 4 is a flowchart of a method for judging the flow direction of cut tobacco feature points according to an embodiment of the present invention;

[0078] Figure 5 is a flowchart of a method for obtaining the three-dimensional flow velocity of cut tobacco by using the optical flow algorithm according to an embodiment of the present invention;

[0079] Figure 6 is a flowchart of a method for obtaining the three-dimensional flow velocity of cut tobacco by using three-dimensional velocity vectors according to an embodiment of the present invention;

[0080] Figure 7Schematic diagram of a test device according to an embodiment of the present invention;

[0081] Figure 8 Schematic diagram of a test process according to an embodiment of the present invention;

[0082] Figure 9 Velocity measurement diagram of a single cut tobacco flowing in a cavity according to an embodiment of the present invention;

[0083] Figure 10 Velocity measurement diagram of multiple cut tobaccos flowing in a cavity according to an embodiment of the present invention. Detailed implementation manners

[0084] The following will describe in detail the specific implementation manners of the embodiments of the present invention with reference to the accompanying drawings. It should be understood that the specific implementation manners described herein are only for explaining and illustrating the embodiments of the present invention, and are not used to limit the embodiments of the present invention.

[0085] It should be noted that the acquisition, transmission, storage, use, processing, etc. of data in the technical solution of this application all comply with the relevant regulations of national laws and regulations. In the embodiments of this application, some industry-existing solutions such as certain software, components, models, etc. may be mentioned. They should be regarded as exemplary. The purpose is only to illustrate the feasibility in the implementation of the technical solution of this application, but it does not mean that the applicant has already or necessarily used this solution.

[0086] As Figure 1 shown is a flowchart of a method for measuring the three-dimensional flow velocity of cut tobacco flowing in a drying chamber according to an embodiment of the present invention. In this Figure 1 the measurement method may include the following steps:

[0087] In step S10, multi-angle images of the flowing cut tobacco are acquired;

[0088] In step S11, the multi-angle images are preprocessed, and the processed images are used as original images;

[0089] In step S12, a feature point detection algorithm is used to detect the original images, all cut tobaccos and the corresponding multiple feature points in the original images are detected, and a cut tobacco-feature point data set is obtained;

[0090] In step S13, the three-dimensional flow velocities of the multiple feature points corresponding to each cut tobacco in the cut tobacco-feature point data set are calculated, and the three-dimensional flow velocity of each cut tobacco is obtained according to the three-dimensional flow velocities of the multiple feature points corresponding to each cut tobacco.

[0091] In this as Figure 1In the method for measuring the three-dimensional flow velocity of flowing cut tobacco in the cut tobacco drying cavity shown, since the movement trajectory of the cut tobacco is three-dimensional, it is necessary to obtain the velocity vectors of the cut tobacco at various perspectives to comprehensively capture the movement characteristics of the cut tobacco in the three-dimensional space. Therefore, it is necessary to obtain multi-angle images of the flowing cut tobacco through step S10. For the specific method of obtaining multi-angle images, although it can be various forms known to those skilled in the art, including but not limited to methods such as rotating shooting, multi-camera array, and drone shooting. In an example of the present invention, considering non-contact measurement without disturbing the natural flow of the cut tobacco, high-speed imaging devices are installed at multiple positions in the cut tobacco drying cavity. These devices are used to capture the flow images of the cut tobacco from different angles, and at the same time, the captured images of each high-speed imaging device are obtained with the same frame rate and synchronized shooting time. Specifically, in this example, the imaging device has good imaging capabilities under high resolution and low light conditions to clearly record the flow condition of the cut tobacco in the cut tobacco drying cavity. The frame rate of the high-speed imaging device can be set according to the shooting requirements. In this example, the frame rate can be not less than 300 frames per second to ensure that the rapid movement and subtle changes of the cut tobacco can be captured. To ensure the temporal consistency of the multi-angle images collected, a synchronous acquisition system is used to control the image acquisition. In this example, the synchronous acquisition system can be a synchronous triggering system, which enables all high-speed imaging devices to start shooting simultaneously and capture image sequences at the same frame rate by uniformly triggering the shooting operation.

[0092] After obtaining the multi-angle images, considering that the noise interference in the collected images is relatively strong, in order to improve the accuracy and stability of subsequent image analysis and processing, it is necessary to preprocess the images through step S11. For the specific method of image preprocessing, although it can be various forms known to those skilled in the art, including but not limited to methods such as filtering and denoising, grayscale conversion, and edge detection. In an example of the present invention, grayscale conversion operations, denoising operations, and enhancement processing operations can be adopted to improve the image quality to enhance the detection accuracy of subsequent feature points and the stability of cut tobacco flow velocity calculation.

[0093] Step S12 is used to detect the feature points of the cut tobacco in the original image, so as to form a cut tobacco-feature point data set. Since the flow of the cut tobacco is random, the shape of the cut tobacco may also change during the flow process. Therefore, in order to accurately track the position change of the cut tobacco and calculate the flow rate of the cut tobacco, it is necessary to identify and track the points with significant characteristics in the original image, that is, the feature points, so as to analyze the motion state of the cut tobacco, that is, step S12: using a feature point detection algorithm to detect the original image, detecting all the cut tobacco and the corresponding multiple feature points in the original image, and obtaining the cut tobacco-feature point data set. In this embodiment, for the specific method of detecting the feature points of the cut tobacco in the original image in step S12, it can be various forms known to those skilled in the art. In an example of the present invention, this step S12 may include, for example, Figure 2 as shown in Figure 2 . In this

[0094] , this step S12 may include:

[0095] L(x, y, σ) = G(x, y, σ) * I(x, y), (1)

[0096] where L(x, y, σ) is the Gaussian blurred image at scale σ, G(x, y, σ) is the Gaussian kernel function, I(x, y) is the original image, σ is the scale parameter used to control the degree of Gaussian blur, and the Gaussian kernel function is the following formula (2),

[0097]

[0098] In step S21, the difference image is obtained according to formula (3),

[0099] D(x, y, σ) = L(x, y, kσ) - L(x, y, σ), (3)

[0100] where D(x, y, σ) is the difference image at scale σ, L(x, y, kσ) is the Gaussian blurred image at scale kσ, and k is a constant used to control the ratio between adjacent scales;

[0101] In step S22, the value of each pixel point in the difference image is compared with the values of the 26 pixel points adjacent to it in space and scale. If the value of this point is the maximum or minimum among the values of its adjacent 26 pixel points, then this point is a feature point;

[0102] In step S23, the exact position of the feature point is solved according to formula (4),

[0103]

[0104] Among them, is the displacement vector of the feature point, and H is the Hessian matrix, which is used to describe the second-order derivative information around the feature point. is the gradient vector of the difference image;

[0105] In step S24, the main direction of the feature point is calculated according to formula (5).

[0106]

[0107] where L x is the image gradient in the x direction of the Gaussian blurred image, and L y is the image gradient in the y direction of the Gaussian blurred image;

[0108] In step S25, according to the main direction of the feature point, the pixel area around the feature point is divided to generate a feature descriptor with direction invariance.

[0109] In the method shown as follows Figure 2 Step S20 is used to construct a Gaussian pyramid. The Gaussian pyramid gradually blurs and shrinks the image on the basis of the original image to form a series of images with different scales. Since detecting feature points at different scales requires smoothing the original image at different scales, in this step S20, by convolving the original image with a Gaussian kernel, Gaussian blurred images at different scales are generated. This helps to detect feature points in the multi-scale space and ensures that the algorithm can detect features of different sizes.

[0110] Step S21 is used to obtain the difference image. Since potential feature points exist in the changes of the detected image at different scales. Therefore, it is necessary to generate a difference image by calculating the difference between adjacent-scale Gaussian blurred images. The difference image can highlight the significant change regions in the image, and these regions are usually the candidate positions of feature points. In this step S21, by subtracting the Gaussian blurred images of adjacent scales at each scale, a difference Gaussian pyramid is constructed, and these difference images strengthen the edges and feature points in the image.

[0111] Step S22 is used to identify the feature points. In the difference Gaussian pyramid, for each pixel point, compare its value with the values of 26 adjacent points in space and scale (3×3×3 window). If the value of this point is the largest or smallest among these points, then this point is considered a potential feature point. By step S22, local extreme points are found, and these extreme points are used as potential feature points.

[0112] Step S23 is used to solve the exact position of the feature points. Since after finding the potential feature points, the positioning of the feature points needs to be precise for further analysis of the motion characteristics of the feature points. Therefore, it is necessary to solve the exact position of the feature points through Step S23. As for the way to solve this position, there are various methods known to those skilled in the art. In this embodiment, three-dimensional quadratic interpolation can be used to accurately locate the potential feature points and obtain their exact positions in the scale space. Through Taylor expansion, each feature point in the image is accurately located in the three-dimensional space (x, y, σ) to remove low-contrast feature points and unstable edge responses.

[0113] Step S24 is used to determine the main direction of the feature points in order to generate a feature descriptor with rotational invariance. This step S24 determines the main direction of the feature points by calculating the gradient directions and gradient magnitudes of the pixels around the feature points and statistically analyzing the histogram. This makes the subsequent generated feature descriptor rotationally invariant, that is, no matter how the image rotates, the feature descriptor can remain consistent.

[0114] Step S25 is used to generate the feature descriptor. Since it is necessary to match the feature points in different images, it is necessary to generate a unique and invariant feature descriptor for each feature point. Step S25 divides the pixel region around the feature point according to the main direction of the feature point to generate a feature descriptor with rotational invariance. There are various ways to divide the pixel region, which are known to those skilled in the art. In this embodiment, a 16×16 window can be used and divided into 4×4 small regions. An 8-dimensional orientation histogram is generated for each region, forming a 128-dimensional feature vector in total. These feature descriptors are used for subsequent feature point matching and can effectively identify the corresponding relationships of the same feature points under different perspectives, different scales, and different rotations.

[0115] Step S13 is used to calculate the three-dimensional flow velocity of the characteristic points corresponding to each cut tobacco, and then determine the three-dimensional flow velocity of each cut tobacco. Since the calculation of the three-dimensional flow velocity of the cut tobacco is obtained through the three-dimensional flow velocity of the characteristic points corresponding to each cut tobacco, it is necessary to calculate the three-dimensional flow velocity of multiple characteristic points corresponding to each cut tobacco, and then average the velocities of the multiple characteristic points corresponding to each cut tobacco in each direction to obtain the three-dimensional flow velocity of each cut tobacco, that is, step S13. In this embodiment, for the specific method of calculating the three-dimensional flow velocity of the cut tobacco in step S13, it can be various forms known to those skilled in the art. In an example of the present invention, since the cut tobacco mostly flows in a slender pipe, its flow characteristics are relatively simple and mainly flow in one direction, and only the velocity of the cut tobacco in the two-dimensional direction needs to be calculated. However, there are some cases where the cut tobacco moves in a wider cavity, and at this time, the flow characteristics of the cut tobacco are relatively complex and the flow velocity of the cut tobacco in the three-dimensional direction needs to be calculated. Therefore, it is necessary to adopt different calculation methods to calculate the velocity of the cut tobacco according to the flow direction of the cut tobacco, so as to accurately describe the flow state of the cut tobacco. This step S13 may include, for example, Figure 3 as shown in Figure 3 . In this

[0116] In step S30, an undetected cut tobacco in the cut tobacco-characteristic point dataset is obtained as the cut tobacco to be measured;

[0117] In step S31, it is judged whether the characteristic points of the cut tobacco to be measured flow in one direction;

[0118] In step S32, when it is judged that the characteristic points of the cut tobacco to be measured flow in one direction, the optical flow algorithm is used to obtain the velocities of multiple characteristic points of the cut tobacco to be measured in the two-dimensional direction, and the average velocity of each direction of the multiple characteristic points of the cut tobacco to be measured is calculated as the three-dimensional flow velocity of the cut tobacco to be measured;

[0119] When it is judged that the characteristic points of the cut tobacco to be measured do not flow in one direction, the three-dimensional velocity vector is used to obtain the velocities of multiple characteristic points of the cut tobacco to be measured in the three-dimensional direction, and the average velocity of each direction of the multiple characteristic points of the cut tobacco to be measured is calculated as the three-dimensional flow velocity of the cut tobacco to be measured;

[0120] In step S33, it is judged whether there is still undetected cut tobacco in the cut tobacco-characteristic point dataset;

[0121] In step S34, when it is judged that there is still undetected cut tobacco in the cut tobacco-characteristic point dataset, return to execute the step of obtaining an undetected cut tobacco in the cut tobacco-characteristic point dataset as the cut tobacco to be measured;

[0122] When it is judged that there is no undetected cut tobacco in the cut tobacco-characteristic point dataset, the calculated three-dimensional flow velocity is used as the three-dimensional flow velocity of the corresponding each cut tobacco.

[0123] In the method shown as follows Figure 3 In this method, step S30 is used to obtain an untested cut tobacco from the cut tobacco-feature point dataset as the cut tobacco to be tested. By step S31, it is judged whether the feature points of the cut tobacco to be tested flow in one direction, and then the corresponding calculation method is used to calculate the three-dimensional flow velocity of the cut tobacco. In this embodiment, for the judgment method of judging whether the cut tobacco flows in one direction, there can be various forms known to those skilled in the art. In one example of the present invention, step S31 may include as follows Figure 4 shown in Figure 4 In this

[0124] In step S40, it is judged whether the proportion of the velocity component in the dominant direction of the feature points of the cut tobacco to be tested in the total velocity is higher than n%;

[0125] In step S41, when it is judged that the proportion of the velocity component in the dominant direction of the feature points of the cut tobacco to be tested in the total velocity is higher than n%, then the feature points of the cut tobacco to be tested flow in one direction;

[0126] When it is judged that the proportion of the velocity component in the dominant direction of the feature points of the cut tobacco to be tested in the total velocity is not higher than n%, then the feature points of the cut tobacco to be tested do not flow in one direction.

[0127] In this Figure 4 shown method, step S40 judges whether the cut tobacco to be tested flows in one direction by calculating the proportion of the velocity component in the dominant direction of the feature points of the cut tobacco to be tested in the total velocity and comparing this proportion with n%. The value of n% can be set according to the measurement requirements. In this example, the value of n% can be 70%-90%.

[0128] Step S32 is used to calculate the three-dimensional flow velocity of the cut tobacco by using different three-dimensional flow velocity calculation methods according to different situations of the cut tobacco flow direction. Considering that when the feature points of the cut tobacco to be tested flow in one direction, the velocity in other directions is almost 0, the optical flow algorithm is used to calculate the velocity of the cut tobacco in the two-dimensional direction, and the motion state of the cut tobacco can be more accurately described; when the feature points of the cut tobacco to be tested do not flow in one direction, the flow characteristics of the cut tobacco are more complex and the flow law is not obvious, such as in eddy currents and elbows, and the three-dimensional velocity vector needs to be used to calculate the flow velocities of the cut tobacco in the three-dimensional directions respectively, so as to accurately describe the flow state of the cut tobacco. In this embodiment, the specific steps for calculating the velocity of the cut tobacco in the two-dimensional direction can be various forms known to those skilled in the art. In one example of the present invention, the steps for calculating the two-dimensional velocity of the cut tobacco by the optical flow algorithm may include as follows Figure 5 shown in Figure 5 In this

[0129] In step S50, the basic optical flow equation is constructed according to formula (6).

[0130] I(x,y,t) = I(x + Δx,y + Δy,t + Δt), (6)

[0131] where I(x,y,t) is the image luminance function;

[0132] In step S51, the optical flow constraint equation is determined according to formula (7).

[0133]

[0134] where u is the component of the optical flow velocity vector in the x direction, and v is the component of the optical flow velocity vector in the y direction, is the rate of change of image luminance with time, is the gradient of the image luminance in the x direction, is the gradient of the image luminance in the y direction;

[0135] In step S52, the gradient of the original image at the feature point in the x direction and the gradient in the Y direction are calculated, and the rate of change of luminance with time is calculated based on the difference between two frames.

[0136] In step S53, an optical flow constraint equation system is constructed for multiple pixel points in the local window of the original image;

[0137] In step S54, the least squares method is used to minimize the optical flow constraint equation system, and the feature point velocity vector of the to-be-detected cut tobacco is determined according to formulas (8) to (10).

[0138]

[0139] where A is the luminance gradient matrix, b is the negative vector of the time gradient, is the gradient of pixel point I 1 in the x direction, is the gradient of pixel point I 1 in the y direction, is the gradient of pixel point I 2 in the x direction, is the gradient of pixel point I 2 in the y direction, is the gradient of pixel point I n in the x direction, is the gradient of pixel point I n in the y direction, is the gradient of pixel point I 1 rate of change with time, is the pixel point I 2 rate of change over time, is the pixel point I n rate of change over time.

[0140] In the method shown as Figure 5 In the method shown as

[0141] In this embodiment, the specific steps for calculating the velocity of cut tobacco in three-dimensional directions can be in various forms known to those skilled in the art. In one example of the present invention, the steps for calculating the three-dimensional velocity vector of cut tobacco can include those Figure 6 shown as Figure 6 In this

[0142] In step S60, determine the three-dimensional position of the feature point P of the cut tobacco to be measured;

[0143] In step S61, calculate the velocity vector of the feature point of the cut tobacco to be measured according to formula (11),

[0144]

[0145] where ΔP is the three-dimensional position change of the feature point P of the cut tobacco to be measured in consecutive frames, and Δt is the time difference between consecutive frames;

[0146] In step S62, fuse the multi-view velocity vectors according to formula (12) to obtain the three-dimensional flow velocity of the cut tobacco,

[0147]

[0148] where V k,x is the x-direction component of the velocity vector of the k-th view, V k,y is the y-direction component of the velocity vector of the k-th view, V k,z is the z-direction component of the velocity vector of the k-th view, and ω k is the weight of the k-th view.

[0149] In this Figure 6In the method shown, step S60 is used to determine the three-dimensional position of the feature points of the to-be-detected cut tobacco, so as to further determine the change in the three-dimensional position of the three-dimensional points in consecutive frames. In this embodiment, the method for determining the three-dimensional position of the feature points of the to-be-detected cut tobacco can be in various forms known to those skilled in the art. In an example of the present invention, the position P=(x, y, z) of the three-dimensional points is reconstructed through the velocity vectors of multiple perspectives, and the projection matrices P i and P j are used to solve the minimization problem to reconstruct the three-dimensional points. The three-dimensional position of the feature points of the to-be-detected cut tobacco can be determined according to formula (13):

[0150]

[0151] where P is the feature point of the to-be-detected cut tobacco, and its three-dimensional point position is (x, y, z), and argmin P represents selecting the value of P with the minimum sum of squared errors among all possible Ps. p i is the position of the two-dimensional image point observed in the i-th perspective, P i is the projection matrix of the i-th perspective, P i,3 is the third row of the projection matrix P i , p j is the position of the two-dimensional image point observed in the j-th perspective, P j is the projection matrix of the j-th perspective, P j,3 is the third row of the projection matrix P j for normalization processing.

[0152] In step S62, the weight ω k of the k-th perspective can be determined by the following factors: image quality (such as image resolution, noise level, etc.), confidence in feature point tracking (such as the stability of the optical flow algorithm in this perspective), and observation conditions of the perspective for the three-dimensional points (such as the angle between the perspective and the movement direction, the more perpendicular the observation effect, the higher the accuracy). This weight is dynamically adjusted according to the following strategy: the weight of low-resolution or high-noise images is lower, the weight of perspectives with unstable matching is lower, and the weight of perspectives close to perpendicular to the movement direction is higher.

[0153] On the other hand, the embodiment of the present invention also provides a measurement system for the three-dimensional flow velocity of cut tobacco flowing in a cut tobacco drying cavity. The system includes a processor configured to execute the measurement method for the three-dimensional flow velocity of cut tobacco as described above.

[0154] To simulate the flow condition of cut tobacco in the cavity, a simple experimental platform was built. The experimental device is as shown in Figure 7 (a), and it mainly consists of a square column with a length of 0.2 m, a width of 0.15 m, and a height of 1.5 m. Three measurement windows of different sizes are provided on one side of the square column as shown in Figure 7(b) is used to observe and measure the flow of cut tobacco. A transfer hole is designed at the bottom of the square column for connecting a hose, and the other end of the hose is connected to a blower to provide air flow drive. Through this device, the flow characteristics of cut tobacco in the cavity can be studied in detail.

[0155] The cut tobacco is in a high-speed motion state in the cavity. To accurately measure its motion speed, an Imaging Source high-frame-rate camera was selected for the experiment, and its fastest frame rate can reach 300 frames per second. Figure 8 Shows the progress of the speed measurement experiment. The experimental steps are as follows: (1) Start the blower and send air into the cavity to make the cut tobacco flow. (2) Place the speed measurement camera 5 cm away from the measurement window to capture the motion of the cut tobacco. (3) Continuously capture for a period of time and save the video. (4) Measure the motion speed of the cut tobacco by analyzing the video frame by frame. By adjusting the rotation speed of the blower and the opening degree of the valve, the gas flow rate entering the experimental bench can be adjusted, thereby changing the air flow velocity in the cavity. During the experiment, a small amount of cut tobacco was put into the square column through the feeding port. When the cut tobacco density is low, it can be considered that the flow speed of the cut tobacco is equal to the gas flow speed. This method can not only accurately capture the details of high-speed motion but also provide sufficient data support for subsequent analysis. The high-frame-rate shooting technology ensures that the subtle motion of the cut tobacco is captured in an extremely short time, thus providing accurate and reliable speed measurement results.

[0156] The experiment collected a one-minute video as the original data for speed measurement analysis. Figure 9 Shows the flow image of a single cut tobacco, where the green dots mark the coordinate positions of the characteristic points of the cut tobacco in each frame. By recording the coordinate positions of the characteristic points in each frame, the motion rate of each characteristic point can be calculated. The specific method is as follows: (1) Mark the characteristic points in each frame and record their coordinate positions. (2) Calculate the motion speed between each frame by comparing the displacements of the characteristic points in consecutive frames. (3) Since there are 6 frames of images, the speeds at 5 moments can be calculated (the speed is calculated from the displacement and time interval between adjacent two frames). (4) Finally, calculate the average value of the speeds at these five moments to obtain the average flow speed of the cut tobacco. Figure 3 Shows the flow speed measurement of a single cut tobacco in the cavity, verifying the feasibility of the optical flow speed measurement algorithm in cut tobacco speed measurement. To further simulate the real measurement situation, a large amount of cut tobacco was added to the cavity and a speed measurement experiment was carried out. Figure 10 Shows the flow images of multiple cut tobaccos. In the case of multiple cut tobaccos, the speed measurement process is more complex because it is impossible to comprehensively measure the speed of all cut tobaccos. Therefore, the cut tobaccos with more obvious characteristic points were selected in the experiment, their coordinates were recorded, and by analyzing the motion trajectories of these characteristic points, the motion speed of each characteristic point was calculated. Finally, the average value of these speeds was calculated to obtain the overall flow speed of the cut tobacco. This method can not only reflect the overall motion trend of the cut tobacco but also improve the accuracy and efficiency of the measurement.

[0157] To further improve the measurement accuracy, multiple high-speed camera devices were also set up in the experiment to achieve synchronous acquisition of multi-angle images. The optical flow method was used to calculate the flow velocity of the cut tobacco, and through three-dimensional reconstruction and multi-angle velocity fusion, the three-dimensional flow velocity distribution of the cut tobacco flowing in the cut tobacco drying chamber was obtained. This method provides high-precision and real-time flow velocity measurement, providing strong technical support for optimizing the cut tobacco drying process and improving the quality of cut tobacco.

[0158] Through the above technical solutions, the embodiments of the present invention provide a method and system for measuring the three-dimensional flow velocity of cut tobacco flowing in a cut tobacco drying chamber. The method and system capture image sequences of the flowing cut tobacco from different angles by arranging high-speed camera devices at multiple angles in the cut tobacco drying chamber, and use the optical flow algorithm to calculate the velocity vectors at each viewing angle, so as to obtain the motion information of the cut tobacco at different viewing angles. Further, through multi-angle data fusion, the three-dimensional flow velocity distribution of the cut tobacco is accurately constructed. Compared with the prior art, this multi-angle optical flow measurement method greatly improves the accuracy and spatial resolution of flow velocity measurement, can provide more detailed and reliable flow information than traditional methods, comprehensively reflects the flow state of the cut tobacco in three-dimensional space, and provides accurate data support for optimizing process parameters.

[0159] The embodiments of the present invention adopt the optical flow method as the core measurement technology, and capture images externally through high-speed camera devices, realizing a completely non-contact measurement. Compared with the prior art, this non-contact measurement method effectively avoids interfering with the flow of the cut tobacco, ensuring the accuracy of the measurement results and the durability of the equipment.

[0160] The system provided by the embodiments of the present invention can output the three-dimensional flow velocity data of the cut tobacco in real time, providing immediate process parameter feedback for production operators. Through the processor, the measurement results can be displayed and stored in the database in real time for subsequent analysis and process optimization. This real-time monitoring function helps to adjust process parameters such as wind speed, temperature, and humidity in a timely manner during the cut tobacco drying process to maintain the stability of the cut tobacco quality. In addition, the storage and analysis of historical data can also provide valuable basis for long-term process improvement.

[0161] The speed measurement method provided by the embodiments of the present invention can accurately grasp the flow velocity distribution of cut tobacco in the cut tobacco drying cavity, and then can accurately control and adjust the air flow velocity and distribution in the cut tobacco drying cavity to ensure that the cut tobacco is evenly heated and fully dried, thereby maintaining the balance of the physical and chemical properties and aroma substances of the cut tobacco. It effectively improves the final quality of the cut tobacco and improves the consistency and taste of the product. At the same time, by accurately measuring and optimizing the control of the cut tobacco flow velocity during the cut tobacco drying process, the efficiency of the cut tobacco drying process can be improved, and unnecessary energy waste can be reduced. For example, by optimizing the flow velocity and direction of the air flow, more effective heat transfer can be achieved, reducing the cut tobacco drying time and energy consumption. This not only reduces the production cost, but also meets the requirements of energy conservation and environmental protection, and helps to improve the economic benefits and social responsibility image of the enterprise.

[0162] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0163] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 the functions specified in one block or multiple blocks.

[0164] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means implements the functions specified in Figure 1 one process or multiple processes and / or blocks Figure 1 the functions specified in one block or multiple blocks.

[0165] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one process Figure 1 one process or multiple processes and / or blocks Figure 1 steps of the functions specified in one block or multiple blocks.

[0166] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and memory.

[0167] The memory may include non-permanent memory in the form of computer-readable media, random access memory (RAM) and / or non-volatile memory such as read-only memory (ROM) or flash memory (flash RAM). The memory is an example of computer-readable media.

[0168] Computer-readable media includes permanent and non-permanent, removable and non-removable media that can store information by any method or technology. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette tapes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transitory media such as modulated data signals and carrier waves.

[0169] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, commodity or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, method, commodity or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, commodity or device comprising the element.

[0170] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included within the scope of the claims of the present application.

Claims

1. A method for measuring the three-dimensional flow velocity of tobacco shreds flowing in a drying chamber, characterized in that: include: Acquire multi-angle images of flowing tobacco; Preprocessing the multi-angle image and using the processed image as the original image; Using a feature point detection algorithm to detect the original image, detect all tobacco shreds and corresponding multiple feature points in the original image, and obtain a tobacco shreds-feature point data set; The three-dimensional flow velocity of the multiple feature points corresponding to each tobacco shred in the tobacco shred-feature point data set is calculated, and the three-dimensional flow velocity of each tobacco shred is obtained according to the three-dimensional flow velocity of the multiple feature points corresponding to each tobacco shred.

2. The method for measuring the three-dimensional flow velocity of shredded tobacco flowing in a drying chamber according to claim 1, characterized in that: Acquire multi-angle images of flowing tobacco, including: A high-speed camera device with multiple shooting angles is preset to shoot the flowing tobacco at multiple angles; The images captured by each high-speed camera device are acquired with the same frame rate and synchronized shooting time to obtain the multi-angle images.

3. The method for measuring the three-dimensional flow velocity of shredded tobacco flowing in a drying chamber according to claim 1, characterized in that: Preprocessing the multi-angle images includes: Grayscale operation, denoising operation and enhancement operation are performed on the multi-angle image.

4. The method for measuring the three-dimensional flow velocity of shredded tobacco flowing in a drying chamber according to claim 1, characterized in that: The original image is detected using a feature point detection algorithm to detect all tobacco shreds and corresponding multiple feature points in the original image, and a tobacco shred-feature point data set is obtained, including: According to formula (1), the original image is convolved with the Gaussian kernel to obtain each layer of Gaussian blurred image. L(x,y,σ)=G(x,y,σ)*I(x,y), (1) Wherein, L(x, y, σ) is the Gaussian blurred image at scale σ, G(x, y, σ) is the Gaussian kernel function, I(x, y) is the original image, σ is the scale parameter used to control the degree of Gaussian blur, and the Gaussian kernel function is the following formula (2), According to formula (3), the difference image is obtained: D(x,y,σ)=L(x,y,kσ)-L(x,y,σ), (3) Where D(x,y,σ) is the difference image at scale σ, L(x,y,kσ) is the Gaussian blurred image at scale kσ, and k is a constant used to control the ratio between adjacent scales; Compare each pixel point in the differential image with the values ​​of its 26 adjacent pixel points in space and scale. If the value of the point is the maximum value or the minimum value among the values ​​of its 26 adjacent pixel points, the point is a feature point. According to formula (4), the exact position of the feature point is solved: in, is the displacement vector of the feature point, H is the Hessian matrix, which is used to describe the second-order derivative information around the feature point, is the gradient vector of the difference image; According to formula (5), the main direction of the feature point is calculated: Among them, L x is the image gradient in the x direction of the Gaussian blurred image, L y is the image gradient of the Gaussian blurred image in the y direction; According to the main direction of the feature point, the pixel area around the feature point is divided to generate a feature descriptor with direction invariance.

5. The method for measuring the three-dimensional flow velocity of tobacco shreds flowing in a drying chamber according to claim 1, characterized in that: Calculating the three-dimensional flow velocity of a plurality of feature points corresponding to each tobacco shred in the tobacco shred-feature point data set, and obtaining the three-dimensional flow velocity of each tobacco shred according to the three-dimensional flow velocity of the plurality of feature points corresponding to each tobacco shred, including: Obtaining an undetected tobacco shred in the tobacco shred-feature point data set as the tobacco shred to be detected; Determining whether the characteristic points of the tobacco to be tested flow in one direction; In the case where it is determined that the characteristic points of the tobacco to be tested flow in one direction, an optical flow algorithm is used to obtain the speeds of multiple characteristic points of the tobacco to be tested in two-dimensional directions, and the average speed of the multiple characteristic points of the tobacco to be tested in each direction is calculated as the three-dimensional flow velocity of the tobacco to be tested; In the case where it is determined that the characteristic points of the tobacco to be tested do not flow in one direction, a three-dimensional velocity vector is used to obtain the velocities of multiple characteristic points of the tobacco to be tested in three-dimensional directions, and an average velocity of the multiple characteristic points of the tobacco to be tested in each direction is calculated as the three-dimensional flow velocity of the tobacco to be tested; Determining whether there is any undetected tobacco in the tobacco-feature point data set; If it is determined that there is still undetected tobacco in the tobacco-feature point data set, returning to the step of obtaining an undetected tobacco in the tobacco-feature point data set as the tobacco to be tested; When it is determined that there is no undetected tobacco in the tobacco-feature point data set, the calculated three-dimensional flow velocity is used as the corresponding three-dimensional flow velocity of each tobacco.

6. The method for measuring the three-dimensional flow velocity of shredded tobacco flowing in a drying chamber according to claim 5, characterized in that: Determining whether the characteristic point of the tobacco to be tested flows in one direction includes: Determine whether the proportion of the velocity component in the dominant direction of the characteristic point of the tobacco to be tested to the total velocity is higher than n%; When it is determined that the velocity component of the dominant direction of the characteristic point of the tobacco to be tested accounts for a proportion of the total velocity that is higher than n%, the characteristic point of the tobacco to be tested flows in one direction; When it is determined that the velocity component of the dominant direction of the characteristic point of the tobacco to be tested accounts for no more than n% of the total velocity, the characteristic point of the tobacco to be tested does not flow in one direction.

7. The method for measuring the three-dimensional flow velocity of shredded tobacco flowing in a drying chamber according to claim 5, characterized in that: The optical flow algorithm is used to obtain the speed of multiple feature points of the tobacco to be tested in two-dimensional directions, including: According to formula (6), the basic equation of optical flow is constructed: I(x,y,t)=I(x+Δx,y+Δy,t+Δt), (6) Where I(x,y,t) is the image brightness function; According to formula (7), the optical flow constraint equation is determined: Among them, u is the component of the optical flow velocity vector in the x direction, v is the component of the optical flow velocity vector in the y direction, is the rate of change of image brightness over time, is the gradient of image brightness in the x direction, is the gradient of image brightness in the y direction; Calculate the gradient of the original image in the x direction at the feature point and the gradient in the y direction Calculate the rate of change of brightness over time based on the difference between the two frames Constructing an optical flow constraint equation group for a plurality of pixel points in a local window of the original image; The optical flow constraint equations are minimized using the least squares method, and the velocity vector of the characteristic point of the tobacco to be tested is determined according to formula (8) to formula (10). Among them, A is the brightness gradient matrix, b is the negative vector of the temporal gradient, is the gradient of pixel I1 in the x direction, is the gradient of pixel I1 in the y direction, is the gradient of pixel I2 in the x direction, is the gradient of pixel I2 in the y direction, is pixel I n The gradient in the x direction, is pixel I n The gradient in the y direction, is the rate of change of pixel I1 over time, is the rate of change of pixel I2 over time, is pixel I n Rate of change over time.

8. The method for measuring the three-dimensional flow velocity of shredded tobacco flowing in a drying chamber according to claim 5, characterized in that: The three-dimensional velocity vector is used to obtain the velocity of multiple characteristic points of the tobacco to be tested in the three-dimensional direction, including: Determine the three-dimensional position of the characteristic point P of the tobacco to be tested; The velocity vector of the characteristic point of the tobacco to be tested is calculated according to formula (11): Wherein, ΔP is the three-dimensional position change of the characteristic point P of the tobacco to be tested in the continuous frames, and Δt is the time difference between the continuous frames; According to formula (12), the multi-view velocity vectors are fused to obtain the three-dimensional flow velocity of the tobacco. Among them, V k,x is the x-direction component of the k-th viewing velocity vector, V k,y is the y-direction component of the k-th viewing velocity vector, V k,z is the z-direction component of the k-th viewing velocity vector, ω k is the weight of the k-th view.

9. The method for measuring the three-dimensional flow velocity of shredded tobacco flowing in a drying chamber according to claim 8, characterized in that: Determining the three-dimensional position of the characteristic point P of the tobacco to be tested includes: According to formula (13), the three-dimensional position of the characteristic point of the tobacco to be tested is determined. Wherein, P is the characteristic point of the tobacco to be tested, and its three-dimensional point position is (x, y, z), argmin P Indicates the value of P with the smallest sum of squared errors among all possible P, p i is the position of the 2D image point observed at viewing angle i, P i is the projection matrix of view i, P i,3 is the projection matrix P i The third line, p j is the position of the 2D image point observed at viewing angle j, P j is the projection matrix of view j, P j,3 is the projection matrix P j The third row is used for normalization.

10. A system for measuring the three-dimensional flow velocity of tobacco shreds flowing in a drying chamber, characterized in that: The system comprises a processor configured to execute the method according to any one of claims 1 to 9.

Images