Hairiness quality on-line detection device, method and sorting method in cone yarn conveying process

Abstract

The present application relates to a kind of sliver quality on-line detection device, method and sorting method in cone conveying process, sliver is photographed by being set on the side of cone conveying path movable camera, the sliver surface generatrix tangent direction sliver image is obtained, and then sliver length index is obtained by the processing of sliver image to work computer, the sliver quality on the surface of sliver is directly evaluated to distinguish the quality of sliver and timely control sliver conveying device different quality sliver is classified and stored, realize sliver quality on-line detection, evaluation, sorting in conveying process are all completed, sliver quality detection and sorting efficiency are greatly improved.

Description

Technical Field

[0001] This invention belongs to the field of yarn hairiness detection technology, specifically relating to an online detection device, method, and sorting method for hairiness quality during the yarn bobbin conveying process. Background Technology

[0002] Yarn cones are the final stage of the spinning process. Multiple yarn tubes are connected and wound together to form a larger package of yarn for warping, twisting, weft winding, dyeing, weft yarn on shuttleless looms, and knitting. Therefore, strictly controlling the quality of the yarn on the yarn cones is a necessary means to ensure the quality of subsequent processes. However, it is difficult to inspect the quality of the yarn cones during production. Currently, the common method is for inspectors to randomly sample the yarn cones, cut a portion of the yarn, and then test the hairiness of this portion. The quality of the entire batch of yarn cones is then judged based on the test results. This inspection method is relatively crude, has low accuracy, and cannot accurately evaluate the individual quality of each yarn cone. This results in the problem that some defective products are mixed with good products and enter the market.

[0003] If the traditional testing methods described above are used to test the quality of each yarn package, each yarn package needs to be sampled, which will greatly affect production efficiency, increase testing costs, and extend the testing cycle. Moreover, each yarn package sample needs to be numbered separately, and after the test is completed, the product location needs to be traced back and transferred (usually to the defective product area or changed to a different specification). This requires a huge storage capacity, which greatly increases the company's production costs.

[0004] Patent application number 2015102373400 discloses a yarn bobbin conveying device. This device uses a gripper to carry the yarn bobbins and transfer them to a storage line for storage. This conveying device can serve multiple yarn bobbin production lines simultaneously, thus serving as a yarn bobbin aggregation point. Developing a device that works in conjunction with this conveying device and can perform online hairiness quality detection on the yarn bobbins during the conveying process will greatly reduce detection costs and improve detection efficiency. Summary of the Invention

[0005] The primary technical problem to be solved by this invention is to provide an online detection device for yarn hairiness quality during the yarn bobbin conveying process, thereby addressing the current technical issues of high sample collection difficulty, high cost, and reduced production efficiency in the online yarn hairiness quality detection process during production.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: an online detection device for hairiness quality during the yarn bobbin conveying process, comprising a guide column parallel to one side of the yarn bobbin conveying path, the upstream end of the guide column being fixedly connected to a fixedly set rear end plate, the downstream end of the guide column being fixedly connected to a fixedly set front end plate, a rear slider and a front slider being slidably connected on the guide column, the front slider being located downstream of the rear slider, the front slider being able to slide only along the axial direction of the guide column, a camera mount being connected to the rear slider, a camera being fixedly connected to the camera mount opposite to the lower edge of the yarn bobbin, and at least two facing... The yarn package gripper has guide holes, each with a guide rod slidably inserted into it. The outer ends of all guide rods are connected to a guide wheel frame, on which a guide wheel is rotatably connected. The axle of the guide wheel is vertically positioned on the outside of the yarn package gripper, with part of the guide wheel facing the end face of the yarn package gripper. Each guide rod is fitted with a coil spring, and one end of each guide rod located within the guide hole has a limiting device to prevent it from disengaging from the guide hole. The front and rear sliders are connected to each other via a connecting rod. An elastic reset device that pulls the rear slider upwards is connected between the rear end plate and the rear slider. The camera is electrically connected to an industrial control computer.

[0007] As a preferred embodiment, the guide post is a grooved cylinder with an axially extending groove on its surface. The front slider is sleeved on the guide post, and a limiting protrusion inserted into the groove is fixedly connected to the front slider. The limiting protrusion slides in cooperation with the groove. The rear slider is rotatably slidably connected to the guide post. A guide plate is fixedly installed on the side of the guide post facing away from the yarn feeding path. The guide plate has multiple guide grooves arranged at equal intervals along the upstream and downstream directions of the yarn. Each guide groove rises or falls at the same height in the vertical direction, forming a discontinuous stepped shape. Any two adjacent guide grooves are connected to each other by an inclined connecting groove. The rear slider is fixedly... A guide rod extending into the guide groove is fixedly connected. A rotating shaft inserted into the guide groove is connected to one end of the guide rod facing the guide groove. A roller is rotatably connected to the rotating shaft and rolls within the guide groove. The guide groove and the connecting groove have equal widths. The roller and the guide groove are fitted with a clearance fit. A proximity sensor is embedded in the inner wall of the middle or upper-middle part of the upstream guide groove. Each of the other guide grooves has a proximity sensor at its upstream end. The proximity sensors are communicatively connected to an industrial control computer and send detection signals to it. The camera is electrically connected to the industrial control computer and controlled by it. Both ends of the connecting rod are connected to the rear slider and the front slider via universal joints.

[0008] As a preferred embodiment, a bracket extending to the other side of the yarn bobbin movement path is fixedly connected to the rear slider. The bracket is located below the yarn bobbin conveying path. A background plate is fixedly connected to the end of the bracket away from the rear slider. The background plate faces the camera directly, and the lower edge of the yarn bobbin is located between the camera and the background plate.

[0009] As a preferred embodiment, the limiting member is a flexible cable.

[0010] As a preferred embodiment, the elastic resetter is a tension spring or a pull rope door closer.

[0011] The further technical problem to be solved by the present invention is to provide a method for online non-destructive detection of yarn hairiness, so as to solve the technical problems of low detection efficiency and poor accuracy of hairiness detection.

[0012] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for online non-destructive detection of yarn bobbin hair, wherein the above-mentioned online detection device is used to take pictures of the surface of the conveyed yarn bobbin, to obtain the hair image of at least one generatrix on the surface of the yarn bobbin, and then the obtained image is processed to calculate the amount of hair per unit length on the surface of the yarn bobbin along its generatrix direction.

[0013] As a preferred embodiment, the image processing procedure specifically includes the following steps:

[0014] a. A Gaussian filter is used to suppress noise in each feather image to obtain a denoised image;

[0015] b. Perform image convolution processing on the denoised image using the discrete differential operator to detect edges in both the horizontal and vertical directions. The horizontal kernel Gx and the vertical kernel Gy are both 3×3 matrices, and their expressions are as follows:

[0016] , ;

[0017] Let I represent the denoised image, and Gx' and Gy' represent the horizontal and vertical image grayscale values, respectively. The formula for convolving any denoised image using the above convolution kernels Gx and Gy is as follows:

[0018] ;

[0019] ;

[0020] After the above process, we obtain grayscale images of the horizontal and vertical edge information. Combining these two grayscale images yields a grayscale image containing edge information in both the horizontal and vertical directions. The grayscale value G of each point in the grayscale image is:

[0021] ;

[0022] c. Perform first-order adaptive dual-limit segmentation on the above grayscale image, specifically including the following steps:

[0023] c1. Select a point z(x,y) at the top left corner of the image. Combine the pixel values ​​of z and its 8 neighboring points to form a 3×3 matrix. If there are no pixels in the neighborhood of point z, fill the corresponding position with 0. The matrix is ​​shown below:

[0024] ;

[0025] c2. Calculate the gradient Zn of point z with respect to its 8 neighborhoods. The gradient calculation formula is shown below:

[0026] ;

[0027] The above formula yields eight gradient values ​​Z1, Z2, ..., Z8 in the neighborhood of point z. Let Z be the average value of Z1, Z2, ..., Z8. The calculation formula is as follows:

[0028] ;

[0029] Traverse all pixels of the grayscale image to obtain a gradient matrix with the same number of rows and columns as the grayscale image pixels;

[0030] c3. Perform double limit detection on the image using iterative calculation. The steps are as follows: Step 1, in the gradient matrix, select the initial gradient T0 as the limit, which is half of the maximum and minimum gradient values.

[0031] Step 2: Divide the image into edge and non-edge parts based on T0. The parts with gradient values ​​higher than T0 are edge parts, and the parts with gradient values ​​lower than T0 are non-edge parts. The percentage of pixels in the edge part is W1, and the average gradient is U1. The percentage of pixels in the non-edge part is W2, and the average gradient is U2. The average gradient of all gradients in the gradient matrix is ​​U. Establish the function:

[0032] + ;

[0033] Step 3: Calculate the new limit value T1 according to the above formula;

[0034] Step 4: Repeat steps 2 and 3 until the limit value T no longer changes, and record this value as Te;

[0035] Step 5: Select a limit value Te as the high limit for dual-limit segmentation and 0.4Te as the low limit. Traverse the image gradient matrix and divide the gradient matrix into three parts according to the dual limits. If the gradient value is higher than the high limit, set the pixel value of that point to 255. If the gradient value is between the high and low limits, observe the gradient values ​​in its 8 neighborhoods. If there is a gradient value higher than the high limit, set the pixel value of that point to 255; otherwise, set it to zero. If the gradient value of a point is lower than the low limit, set the pixel value of that point to zero. Finally, obtain the binary image of the feathers.

[0036] d. Use pixel counting method to count feathers: count the number of pixels with a pixel value of 255 in the feather image obtained in step c, which is the number of pixels occupied by the feather edge. By calibrating the actual length represented by each pixel under the system resolution, the actual length of the feather edge can be calculated. Since the feather has two edges, take half of it as the actual length N of the feather.

[0037] e. Set the yarn hairiness length index Hj, which represents the total amount of hairiness per unit length of the yarn bobbin. Assume the yarn bobbin is parallel to the top and bottom edges of the camera viewfinder. Therefore, the actual length represented by the number of pixels in the length direction of the camera resolution is the length M of the roll edge bobbin. The ratio of these two is the yarn hairiness length index Hj, calculated using the following formula:

[0038] .

[0039] The technical problem that this invention further aims to solve is to provide a sorting method in the yarn bobbin conveying process, so as to solve the technical problems of low detection efficiency, long detection cycle and high detection cost caused by the additional intermediate processes such as transfer and storage in the yarn bobbin hair quality detection process.

[0040] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a finished product quality control method in the yarn package conveying process, which uses the above-mentioned online hairiness quality detection device and the online non-destructive hairiness detection method of the yarn package to perform online non-destructive hairiness detection of the yarn package during the conveying process, and sends the detection results to the industrial control computer that controls the yarn package conveying device. The industrial control computer evaluates the detection results and makes a result of qualified or unqualified. The industrial control computer controls the yarn package conveying device to convey the yarn package with unqualified hairiness quality to the unqualified area, and to convey the qualified yarn package to the qualified area.

[0041] As a preferred option, the evaluation method is as follows:

[0042] A statistical database of yarn hairiness of the same specification is pre-established, and the test results of each yarn package of the same specification are recorded to form a sample group. Then, according to the formula... Calculate the outlier value of the hairiness length index of the current tested yarn package. When D≥9, the tested yarn package is considered to be abnormal. The industrial control computer controls the yarn package conveying device to transport the yarn package to the non-conforming area. In the above formula, D is the evaluation score, x is the hairiness length index Hj, μ is the average value of the test index Hj, and S2 is the variance of the test index Hj.

[0043] As another preferred option, the evaluation method is as follows: based on the yarn package specifications, a corresponding hair length index standard and a reasonable deviation rate are preset. The industrial control computer compares the currently detected yarn package hair length index with the hair length index standard of the same specification yarn package and calculates the deviation rate. If the deviation rate exceeds the reasonable deviation rate, the yarn package is recorded as unqualified; otherwise, it is recorded as qualified.

[0044] The beneficial effects of this invention are as follows: This invention uses a camera that can follow and move along one side of the yarn package conveying path to capture images of the yarn package surface in the direction of the tangent of the generatrix. Then, the image of the hairy texture is processed by an industrial control computer to obtain the hairy texture length index, which evaluates the quality of the hairy texture on the surface of the yarn package. During the data acquisition and detection process, it will not affect the production process of the yarn package, reducing the difficulty of data acquisition. Moreover, it can obtain clear images and accurate detection results. Furthermore, the image processing speed of the industrial control computer or computer is fast, and the detection efficiency is high, which can obtain the detection results in a very short time. Then, it can directly allocate storage points for yarn packages of different qualities during the yarn package conveying process.

[0045] This invention further utilizes first-order adaptive double-limit segmentation processing, particularly gradient matrix double-limit segmentation, in the processing of feather images. This effectively extracts feather edge details, removes noisy edges, achieves better image segmentation results, improves the accuracy of feather length calculation, and enhances the accuracy of online feather quality detection.

[0046] This invention further distinguishes between good and bad yarn quality by directly evaluating the quality of the yarn hairiness of the tested yarn packages and promptly controls the yarn package conveying device to classify and store yarn packages of different qualities. This enables the quality inspection and classification of yarn packages to be completed during the conveying process, greatly improving the efficiency of yarn package quality inspection and classification. Attached Figure Description

[0047] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, wherein:

[0048] Figure 1 This is a top view of the specific structure of the invention;

[0049] Figure 2 yes Figure 1 AA section view in the middle;

[0050] Figure 3 yes Figure 2 BB section view in the middle;

[0051] Figure 4 This is a schematic diagram of the structure of the guide plate described in this invention;

[0052] Figure 5 It is an image of the surface fuzz of the yarn package captured by a camera;

[0053] Figure 6 yes Figure 5 The grayscale image shown is the result of denoising and convolution of the feather image.

[0054] Figure 7 yes Figure 6 The grayscale image shown is a binary image of the feathers after first-order adaptive double-limit segmentation.

[0055] Figures 1-4 In the middle: 1. Guide post, 2. Rear end plate, 3. Front end plate, 4. Rear slider, 5. Front slider, 6. Camera mount, 7. Camera, 8. Industrial computer, 9. Guide hole, 10. Guide rod, 11. Guide wheel frame, 12. Guide wheel, 13. Coil spring, 14. Limiting component, 15. Connecting rod, 16. Elastic resetter, 17. Slide groove, 18. Limiting protrusion, 19. Guide plate, 20. Guide groove, 21. Connecting groove, 22. Guide rod, 23. Rotating shaft, 24. Roller, 25. Universal joint, 26. Bracket, 27. Background plate, 28. Proximity sensor, 29. Yarn bobbin, 30. Gripper. Detailed Implementation

[0056] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0057] Example 1:

[0058] like Figures 1-4The online detection device for hairiness quality during the yarn bobbin conveying process shown includes a guide post 1 parallel to one side of the yarn bobbin 29 conveying path. The upstream end of the guide post 1 is fixedly connected to a fixed rear end plate 2, and the downstream end of the guide post 1 is fixedly connected to a fixed front end plate 3. Both the front end plate 3 and the rear end plate 2 are fixedly installed and can be fixedly connected to the top of a building or to a column or other fixed object. A rear slider 4 and a front slider 5 are slidably connected to the guide post 1. The front slider 5 is located downstream of the rear slider 4 and can only slide along the axial direction of the guide post 1. A camera mount 6 is connected to the rear slider 4, and a camera 7 opposite to the lower edge of the yarn bobbin is fixedly connected to the camera mount 6. The front slider 4 has at least two guide holes 9 facing the yarn bobbin gripper. A guide rod 10 is slidably inserted into each guide hole 9. The outer ends of all guide rods 10 are connected to a guide wheel frame 11, and a guide wheel is rotatably connected to the guide wheel frame 11. 12. The axle of the guide wheel 12 is vertically positioned outside the gripper 30 used for conveying the yarn package 29, and part of the guide wheel 12 is opposite to the end face of the yarn package gripper 30. Each guide rod 10 is fitted with a coil spring 13. One end of each guide rod 10 located in the guide hole 9 is provided with a limiting member 14 to restrict its disengagement from the guide hole 9. In this embodiment, the limiting member 14 is a flexible cable. Of course, a tension spring or other commonly used structural form can also be used, as long as it does not affect the axial sliding of the guide rod 10 and can restrict its disengagement from the guide hole 9. The front slider 5 and the rear slider 4 are connected to each other by a connecting rod 15. An elastic resetter 16 is connected between the rear end plate 2 and the rear slider 4 to pull the rear slider 4 to slide upstream. The elastic resetter 16 can be a tension spring or a pull rope door closer. In this embodiment, a tension door closer is used. When the tension is pulled out, it is elastic and can pull its free end back. The camera 7 is electrically connected to the industrial control computer 8. In use, the images captured by the camera are sent to the industrial control computer 8, which performs image processing. The industrial control computer 8 can be installed on other fixed objects or on the ground and is connected to the camera 7 via a communication line.

[0059] During use, the elasticity of the coil spring 13 needs to be appropriately greater than the tension of the elastic resetter 16. When the guide wheel 12 abuts against the front end face of the gripper 30, the elastic force of the coil spring 13 enables the gripper 30 to push the guide wheel 12, the front slider 5, the rear slider 4, and the gripper 30 to move synchronously. After the front slider 5 abuts against the front end plate 3, it is blocked by the front end plate 3. The force of the gripper 30 on the guide wheel 12 causes the coil spring 13 to contract, and the guide wheel 12 is squeezed out of the front end of the gripper 30. The gripper 30 continues to move forward, while the front slider 5 and the rear slider 4 are reset upstream under the reset force of the elastic resetter 16, preparing for the detection of the yarn on the next gripper 30.

[0060] In this embodiment, the guide post 1 is preferably a grooved cylinder. An axially extending groove 17 is formed on the surface of the guide post 1. The front slider 5 is sleeved on the guide post 1, and a limiting protrusion 18 inserted into the groove 17 is fixedly connected to the front slider 5. The limiting protrusion 18 slides in cooperation with the groove 17. The rear slider 4 is rotatably slidably connected to the guide post 1. A guide plate 19 is fixedly provided on the side of the guide post 1 facing away from the yarn feeding path. Multiple guide grooves 20 are formed on the guide plate 19, arranged at equal intervals along the upstream and downstream directions of the yarn. Each guide groove 20 rises or falls at the same height in the vertical direction, forming a discontinuous stepped shape. Any two adjacent guide grooves 20 are connected to each other by an inclined connecting groove 21. A guide rod 22 extending towards the guide groove 20 is fixedly connected to the rear slider 4. One end of the guide groove 20 is connected to a rotating shaft 23 inserted into the guide groove 20. A roller 24 is rotatably connected to the rotating shaft 23. The roller 24 is rolled and embedded in the guide groove 20. The guide groove 20 and the connecting groove 21 have the same width. The roller 24 is clearance-fitted with the guide groove 20. A proximity sensor 28 is embedded in the inner wall of the middle or upper middle part of the upstream guide groove 20. A proximity sensor 28 is respectively provided at the upstream end of each of the other guide grooves. The proximity sensor 28 can be a photoelectric sensor or a magnetic field sensor. The proximity sensor 28 is communicatively connected to the industrial control computer 8 and sends detection signals to the industrial control computer 8. The camera 7 is electrically connected to the industrial control computer 8 and is controlled by the industrial control computer 8. In order to accommodate the rotation of the rear slider 4, the two ends of the connecting rod 15 in this embodiment are connected to the rear slider 4 and the front slider 5 respectively through universal joints 25.

[0061] In actual production, to accommodate the tilt of roller 24, its diameter is designed to be smaller, and the width of its surface is also smaller than the depth of guide groove 20, ensuring that the roller can still roll within guide groove 20 even when tilted. Generally, the weight of the camera 7 end is greater, causing roller 24 to roll close to the top surface of guide groove 20. Guided by guide groove 20 and connecting groove 21, roller 24 is driven to rotate around guide post 1, which in turn causes camera 7 to swing. This allows the camera to capture images of the hairline on different generatrices of the yarn package 29 during synchronized movement with the yarn package 29, increasing the number of samples to be detected and further improving detection accuracy.

[0062] In this embodiment, a bracket 26 extending to the other side of the yarn bobbin movement path is preferably fixedly connected to the rear slider 4. The bracket 26 is located below the yarn bobbin conveying path. A background plate 27 is fixedly connected to the end of the bracket 26 away from the rear slider 4. The background plate 27 faces the camera 7, and the lower edge of the yarn bobbin is located between the camera 7 and the background plate 27. The background plate 27 can improve the clarity of the feathers in the image, reduce noise, and eliminate the influence of complex environments on the detection results. The background plate 27 can be selected to emit light or not emit light as needed. In practical applications, supplementary lights can also be added to the bracket 26 or the rear slider 4 as needed.

[0063] The working process of this embodiment is as follows: Figures 1-4 As shown, in the initial state, the rear slider 4 is pulled by the elastic reset device 16 and is located at one end of the rear end plate 2, in contact with the rear end plate 2. Alternatively, a buffer pad can be set between the rear end plate 2 and the rear slider 4, and the roller 24 remains stationary in the upstream guide groove 20. When the gripper 30 moves with the yarn package 29 to abut against the guide wheel 12, the guide wheel 12 moves along with the front end of the gripper 30, causing the front slider 5 to move downstream along the guide post 1 in sync with the gripper 30. The front slider 5 drives the rear slider 4, camera 7, background plate 27, and roller 24 on the rear slider to move downstream in sync via the connecting rod 15. The roller 24 moves along the guide groove 20. When the roller 24 passes any of the proximity sensors 28, the proximity sensor 28 sends a signal to the industrial control computer 8. The industrial control computer 8 then controls the camera 7 to take one or more photos. During the synchronous movement of the gripper 30 and the camera 7, the roller 24 rolls into the guide groove 20 at different heights in sequence, causing the camera 7 to change different shooting angles and take multi-angle photos of the yarn package to obtain images of the hairiness on different generatrices of the yarn package, which are then sent to the industrial control computer.

[0064] When the current slider 5 moves to abut against the front plate 3, the front slider 5 can no longer move forward. At this time, the guide wheel 12 is squeezed, causing the coil spring 13 to contract. The guide wheel 12 disengages from the front face of the gripper 30. After the guide wheel 12 disengages from the front face of the gripper 30, the pushing force of the gripper 30 on the guide wheel 12 and the front slider 5 disappears. Under the traction force of the elastic resetter 16, the rear slider 4 and the front slider 5 reset upstream.

[0065] During the reset process, the roller 24 and each guide groove 20 are reset upstream. During the reset process, they will pass through each proximity sensor 28 again. The industrial control computer can determine whether the camera 7 is in a state of synchronous following with the yarn package 29 or in a reset state according to the order of receiving signals from each sensor. If it is in a reset state, the camera 7 will not be operated or controlled.

[0066] The industrial control computer performs image processing based on the received yarn fluff image to finally obtain the yarn fluff quality of the inspected yarn package 29.

[0067] Example 2:

[0068] like Figures 1-7 As shown, a method for online non-destructive detection of yarn bobbin hairs involves using the online detection device described in Embodiment 1 to photograph the surface of the conveying yarn bobbin 29, acquiring images of the hairs on at least one generatrix of the yarn bobbin 29 surface. Figure 5 (as shown), and then image processing is performed on the obtained image to calculate the amount of hair per unit length on the surface of the yarn package 29 along its generatrix direction.

[0069] In this embodiment, the image processing process specifically includes the following steps:

[0070] a. A Gaussian filter is used to suppress noise in each feather image to obtain a denoised image;

[0071] b. Perform image convolution processing on the denoised image using the discrete differential operator to detect edges in both the horizontal and vertical directions. The horizontal kernel Gx and the vertical kernel Gy are both 3×3 matrices, and their expressions are as follows:

[0072] , ;

[0073] Let I represent the denoised image, and Gx' and Gy' represent the horizontal and vertical image grayscale values, respectively. The formula for convolving any denoised image using the above convolution kernels Gx and Gy is as follows:

[0074] ;

[0075] ;

[0076] After the above process, we obtain grayscale images of the horizontal and vertical edge information. Combining these two grayscale images yields a grayscale image containing edge information in both the horizontal and vertical directions, as shown below. Figure 6 As shown, the grayscale value G of each point in the grayscale image is:

[0077] ;

[0078] Compared to traditional edge extraction operators, the discrete differential operator proposed in this embodiment can reduce edge loss. Given the large number of hairy edges in yarn bobbins, this operator can retain more hairy details.

[0079] c. Perform first-order adaptive dual-limit segmentation on the above grayscale image, specifically including the following steps:

[0080] c1. Select a point z(x,y) at the top left corner of the image. Combine the pixel values ​​of z and its 8 neighboring points to form a 3×3 matrix. If there are no pixels in the neighborhood of point z, fill the corresponding position with 0. The matrix is ​​shown below:

[0081] ;

[0082] c2. Calculate the gradient Zn of point z with respect to its 8 neighborhoods. The gradient calculation formula is shown below:

[0083] ;

[0084] The above formula yields eight gradient values ​​Z1, Z2, ..., Z8 in the neighborhood of point z. Let Z be the average value of Z1, Z2, ..., Z8. The calculation formula is as follows:

[0085] ;

[0086] Traverse all pixels of the grayscale image to obtain a gradient matrix with the same number of rows and columns as the grayscale image pixels;

[0087] c3. Perform double limit detection on the image using iterative calculation. The steps are as follows: Step 1, in the gradient matrix, select the initial gradient T0 as the limit, which is half of the maximum and minimum gradient values.

[0088] Step 2: Divide the image into edge and non-edge parts based on T0. The parts with gradient values ​​higher than T0 are edge parts, and the parts with gradient values ​​lower than T0 are non-edge parts. The percentage of pixels in the edge part is W1, and the average gradient is U1. The percentage of pixels in the non-edge part is W2, and the average gradient is U2. The average gradient of all gradients in the gradient matrix is ​​U. Establish the function:

[0089] + ;

[0090] Step 3: Calculate the new limit value T1 according to the above formula;

[0091] Step 4: Repeat steps 2 and 3 until the limit value T no longer changes, and record this value as Te;

[0092] Step 5: Select a limit value Te as the high limit for dual-limit segmentation, and select 0.4Te as the low limit. The choice of the low limit is based on engineering experience and actual experimental results. Traverse the image gradient matrix and divide the pixels into three parts according to the dual limits: if the gradient of a pixel is higher than the high limit, set its pixel value to 255; if the gradient of a pixel is between the high and low limits, observe the gradient in its 8 neighborhoods. If there is a gradient higher than the high limit, set its pixel value to 255; otherwise, set it to zero. If the gradient of a pixel is lower than the low limit, set its pixel value to zero. This yields the binary image of the feathers.

[0093] The proposed gradient matrix dual-limit segmentation method is highly sensitive to edges, effectively extracting feathery edge details and removing noisy edges, thus achieving better image segmentation results.

[0094] d. Use pixel counting method to count feathers: count the number of pixels with a pixel value of 255 in the feather image obtained in step c, which is the number of pixels occupied by the feather edge. By calibrating the actual length represented by each pixel under the system resolution, the actual length of the feather edge can be calculated. Since the feather has two edges, take half of it as the actual length N of the feather.

[0095] e. Set the yarn hairiness length index Hj, which represents the total amount of hairiness per unit length of the yarn bobbin. Assume the yarn bobbin is parallel to the top and bottom edges of the camera viewfinder. Therefore, the actual length represented by the number of pixels in the length direction of the camera resolution is the length M of the roll edge bobbin. The ratio of these two is the yarn hairiness length index Hj, calculated using the following formula:

[0096] .

[0097] Example 3:

[0098] The sorting method during the yarn bobbin conveying process employs the online hairiness quality detection device described in Example 1 and the online non-destructive hairiness detection method described in Example 2 to perform online non-destructive hairiness detection on the yarn bobs during the conveying process. The detection results are sent to the industrial control computer that controls the yarn bobbin conveying device. The industrial control computer evaluates the detection results and determines whether the results are qualified or unqualified. The industrial control computer controls the yarn bobbin conveying device to convey yarn bobs with unqualified hairiness quality to the unqualified area and to convey qualified yarn bobs to the qualified area.

[0099] The evaluation method in this embodiment is as follows:

[0100] A statistical database of yarn hairiness of the same specification is pre-established, and the test results of each yarn package of the same specification are recorded to form a sample group. Then, according to the formula... Calculate the outlier value of the hairiness length index of the current tested yarn package. When D≥9, the tested yarn package is considered to be abnormal. The industrial control computer controls the yarn package conveying device to transport the yarn package to the non-conforming area. In the above formula, D is the evaluation score, x is the hairiness length index Hj, μ is the average value of the test index Hj, and S2 is the variance of the test index Hj.

[0101] In actual production, the evaluation method can also be as follows: based on the yarn package specifications, a corresponding hair length index standard and a reasonable deviation rate are preset. The industrial control computer compares the currently detected yarn package hair length index with the hair length index standard of the same specification yarn package and calculates the deviation rate. If the deviation rate exceeds the reasonable deviation rate, the yarn package is recorded as unqualified; otherwise, it is recorded as qualified.

[0102] The above embodiments are merely illustrative of the principles and effects of the present invention, as well as some examples of its application, and are not intended to limit the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept of the present invention, and these modifications and improvements are all within the scope of protection of the present invention.

Claims

1. An online detection device for the hairiness quality during the yarn bobbin conveying process, characterized in that, The device includes a guide post (1) parallel to one side of the yarn bobbin conveying path. The upstream end of the guide post (1) is fixedly connected to a fixed rear end plate (2), and the downstream end of the guide post (1) is fixedly connected to a fixed front end plate (3). A rear slider (4) and a front slider (5) are slidably connected on the guide post (1). The front slider (5) is located downstream of the rear slider (4) and can only slide along the axial direction of the guide post (1). A camera mount (6) is connected to the rear slider (4), and a camera (7) is fixedly connected to the camera mount (6) opposite to the lower edge of the yarn bobbin. At least two guide holes (9) facing the yarn bobbin gripper are opened on the front slider (5), and a guide rod (10) is slidably inserted into each guide hole (9). The outer ends of the guide rods (10) are connected to a guide wheel frame (11), and a guide wheel (12) is rotatably connected on the guide wheel frame (11). The axle of the guide wheel (12) is vertically set on the outside of the yarn package gripper, and part of the guide wheel (12) is opposite to the end face of the yarn package gripper. Each guide rod (10) is fitted with a coil spring (13). One end of each guide rod (10) located in the guide hole (9) is provided with a limiting piece (14) to restrict it from leaving the guide hole (9). The front slider (5) and the rear slider (4) are connected to each other through a connecting rod (15). An elastic resetter (16) that pulls the rear slider (4) to slide upstream is connected between the rear end plate (2) and the rear slider (4). The camera (7) is electrically connected to the industrial control computer (8). The guide post (1) is a grooved cylinder. An axially extending groove (17) is provided on the surface of the guide post (1). The front slider (5) is sleeved on the guide post (1) and a limiting protrusion (18) inserted into the groove (17) is fixedly connected to the front slider (5). The limiting protrusion (18) and the groove (17) are slidably engaged. The rear slider (4) is rotatably slidably connected to the guide post (1). A guide plate (19) is fixedly provided on the side of the guide post (1) facing away from the yarn conveying path. The guide plate (19) has multiple guide grooves (20) arranged at equal intervals along the upstream and downstream directions of the yarn. Each guide groove (20) rises or falls at the same height in the vertical direction, forming a discontinuous stepped shape. Any two adjacent guide grooves (20) are connected to each other by an inclined connecting groove (21). A guide rod extending towards the guide groove (20) is fixedly connected to the rear slider (4). (22) One end of the guide rod (22) facing the guide groove (20) is connected to a rotating shaft (23) inserted into the guide groove (20). A roller (24) is rotatably connected on the rotating shaft (23). The roller (24) is rolled and embedded in the guide groove (20). The width of the guide groove (20) and the connecting groove (21) are equal. The roller (24) is clearance-fitted with the guide groove (20). A proximity sensor (28) is embedded on the inner wall of the middle or upper middle part of the upstream guide groove (20). A proximity sensor (28) is respectively set at the upstream end of each of the other guide grooves. The proximity sensor (28) is connected to the industrial control computer (8) and sends a detection signal to the industrial control computer (8). The camera (7) is electrically connected to the industrial control computer (8) and is controlled by the industrial control computer (8). The two ends of the connecting rod (15) are connected to the rear slider (4) and the front slider (5) respectively through the universal joint (25).

2. The online detection device for yarn hairiness quality during the yarn bobbin conveying process according to claim 1, characterized in that, A bracket (26) extending to the other side of the yarn bobbin movement path is fixedly connected to the rear slider (4). The bracket (26) is located below the yarn bobbin conveying path. A background plate (27) is fixedly connected to one end of the bracket (26) away from the rear slider (4). The background plate (27) is directly opposite the camera (7), and the lower edge of the yarn bobbin is located between the camera (7) and the background plate (27).

3. The online detection device for yarn hairiness quality during the yarn bobbin conveying process according to claim 1, characterized in that, The limiting component (14) is a flexible cable.

4. The online detection device for yarn hairiness quality during the yarn bobbin conveying process according to claim 1, characterized in that, The elastic resetter (16) is a tension spring or a pull rope door closer.

5. A method for online non-destructive detection of yarn hairiness, characterized in that, The online detection device described in any one of claims 1 to 4 is used to photograph the surface of the conveyed yarn package to obtain a hairy image on at least one generatrix of the yarn package surface. Then, the obtained image is processed to calculate the amount of hairy material per unit length on the yarn package surface along its generatrix direction.

6. The method for online non-destructive detection of yarn hairiness according to claim 5, characterized in that, The image processing procedure specifically includes the following steps: a. A Gaussian filter is used to suppress noise in each feather image to obtain a denoised image; b. Perform image convolution processing on the denoised image using the discrete differential operator to detect edges in both the horizontal and vertical directions. The horizontal kernel Gx and the vertical kernel Gy are both 3×3 matrices, and their expressions are as follows: ， ； Let I represent the denoised image, and Gx' and Gy' represent the horizontal and vertical image grayscale values, respectively. The formula for convolving any denoised image using the above convolution kernels Gx and Gy is as follows: ； ； After the above process, we obtain grayscale images of the horizontal and vertical edge information. Combining these two grayscale images yields a grayscale image containing edge information in both the horizontal and vertical directions. The grayscale value G of each point in the grayscale image is: ； c. Perform first-order adaptive dual-limit segmentation on the above grayscale image, specifically including the following steps: c1. Select a point z(x,y) at the top left corner of the image. Combine the pixel values ​​of z and its 8 neighboring points to form a 3×3 matrix. If there are no pixels in the neighborhood of point z, fill the corresponding position with 0. The matrix is ​​shown below: ； c2. Calculate the gradient Zn of point z with respect to its 8 neighborhoods. The gradient calculation formula is shown below: ； The above formula yields eight gradient values ​​Z1, Z2, ..., Z8 in the neighborhood of point z. Let Z be the average value of Z1, Z2, ..., Z8. The calculation formula is as follows: ； Traverse all pixels of the grayscale image to obtain a gradient matrix with the same number of rows and columns as the grayscale image pixels; c3. Perform double limit detection on the image using iterative calculation. The steps are as follows: Step 1, in the gradient matrix, select the initial gradient T0 as the limit, which is half of the maximum and minimum gradient values. Step 2: Divide the image into edge and non-edge parts based on T0. The parts with gradient values ​​higher than T0 are edge parts, and the parts with gradient values ​​lower than T0 are non-edge parts. The percentage of pixels in the edge part is W1, and the average gradient is U1. The percentage of pixels in the non-edge part is W2, and the average gradient is U2. The average gradient of all gradients in the gradient matrix is ​​U. Establish the function: ； Step 3: Calculate the new limit value T1 according to the above formula; Step 4: Repeat steps 2 and 3 until the limit value T no longer changes, and record this value as Te; Step 5: Select a limit value Te as the high limit value for dual-limit segmentation and select 0.4Te as the low limit value. Traverse the image gradient matrix and divide the pixels into three parts according to the dual limits. If the gradient of a pixel is higher than the high limit value, set its pixel value to 255. If the gradient of a pixel is between the high and low limits, observe the gradient in its 8 neighborhoods. If there is a gradient higher than the high limit value, set its pixel value to 255. Otherwise, set it to zero. If the gradient of a pixel is lower than the low limit value, set the pixel value to zero. Obtain the binary image of the feathers. d. Use pixel counting method to count feathers: count the number of pixels with a pixel value of 255 in the feather image obtained in step c, which is the number of pixels occupied by the feather edge. By calibrating the actual length represented by each pixel under the system resolution, the actual length of the feather edge can be calculated. Since the feather has two edges, take half of it as the actual length N of the feather. e. Set the yarn hairiness length index Hj, which represents the total amount of hairiness per unit length of the yarn bobbin. Assume the yarn bobbin is parallel to the top and bottom edges of the camera viewfinder. Therefore, the actual length represented by the number of pixels in the length direction of the camera resolution is the length M of the roll edge bobbin. The ratio of these two is the yarn hairiness length index Hj, calculated using the following formula: 。 7. A sorting method during the yarn bobbin conveying process, characterized in that, The yarn bobbins during the conveying process are inspected online using the online bobbin quality detection device as described in any one of claims 1 to 4 and the online non-destructive bobbin detection method as described in claim 5 or 6. The detection results are sent to the industrial control computer that controls the yarn bobbin conveying device. The industrial control computer evaluates the detection results and determines whether the yarn bobbins are qualified or unqualified. The industrial control computer controls the yarn bobbin conveying device to convey yarn bobbins with unqualified bobbin quality to the unqualified area and to convey qualified yarn bobbins to the qualified area.

8. The sorting method according to claim 7, characterized in that, The evaluation method is as follows: A statistical database of yarn hairiness of the same specification is pre-established, and the test results of each yarn package of the same specification are recorded to form a sample group. Then, according to the formula... Calculate the outlier value of the hairiness length index of the current tested yarn package. When D≥9, the tested yarn package is considered to be abnormal. The industrial control computer controls the yarn package conveying device to transport the yarn package to the non-conforming area. In the above formula, D is the evaluation score, x is the hairiness length index Hj, μ is the average value of the test index Hj, and S2 is the variance of the test index Hj.

9. The sorting method according to claim 8, characterized in that, The evaluation method is as follows: based on the yarn package specifications, a corresponding hair length index standard and a reasonable deviation rate are preset. The industrial control computer compares the currently detected yarn package hair length index with the hair length index standard of the same specification yarn package and calculates the deviation rate. If the deviation rate exceeds the reasonable deviation rate, the yarn package is recorded as unqualified; otherwise, it is recorded as qualified.

Images