On-line detecting device, detecting method and regulating method for yarn hairiness of sizing machine

Abstract

The present invention relates to an on-line detecting device, a detecting method and a regulating method for yarn hairiness of a sizing machine. The present invention detects the yarn in the sizing process through an image acquisition device, an image processing module and an information feedback module, obtains the real-time yarn surface hairiness index, and gives real-time feedback to the operator. The operator can adjust the process parameters of the sizing machine according to the feedback on the yarn surface hairiness index information during the sizing process, thereby improving the sizing quality.

Description

Technical Field

[0001] The present invention relates to an on-line detecting device, a detecting method and a regulating method for yarn hairiness of a sizing machine. Background Art

[0002] A sizing machine is a device for sizing yarns. After sizing, the strength and abrasion resistance of the yarns are improved, and the amount of hairiness is reduced, making them more suitable for weaving. The sizing machine successively includes a creel, a sizing vat, a drying chamber and a cloth beam creel from its upstream to downstream. The original yarn unwinds from the creels on the creel, is sized in the sizing vat and then enters the drying chamber for drying, and is then wound up by the cloth beams on the cloth beam creel.

[0003] Reducing the hairiness on the surface of the yarn is one of the important objectives of sizing. Detecting the amount of hairiness of the sized yarn, evaluating the change of yarn hairiness in different processes during the sizing process, and comparing the difference in the amount of yarn hairiness at different lateral positions of the warp sheets on the sizing machine are of great significance for evaluating the sizing quality of the sizing machine and timely responding to regulation. At present, the common detection of the amount of hairiness on the surface of the yarn is mainly carried out by manually or using specific tools to detect the static yarn hairiness, which is a cumbersome process, with low efficiency and high labor costs. Moreover, the yarn samples are taken from the cloth beams after the sizing machine has finished working, and the obtained hairiness information only comes from the yarns after sizing is completely finished. Therefore, the hairiness information of the yarns at each stage during the sizing work cannot be obtained, which belongs to off-line detection and cannot meet the needs of on-line detection and real-time regulation. Such a "small sample" method results in lagging and incomplete hairiness information, and has limited reference significance for the control of the sizing process of the sizing machine.

[0004] At present, the regulation of the sizing machine mainly adopts a reverse regulation method, that is, after sizing a batch of yarns, the hairiness of the yarns on the cloth beam is detected, and then the process parameters such as the pressing force, the concentration of the sizing solution, and the machine speed during the sizing of the next batch of yarns are deduced in reverse. Such a correction method cannot change the already completed yarns, and it is difficult to control the correction accuracy of the whole batch of yarns for subsequent sizing. Another method is to conduct a trial run first, that is, sizing the front section of a batch of yarns, then performing a breaking hairiness detection on the sized and dried yarns, and then adjusting the sizing process parameters of the sizing machine according to the detection results. This may require repeated detections, with low detection efficiency and slow feedback speed. The detected yarns have limited guiding significance for the sizing process of the yarns being sized, and ultimately cannot effectively improve the sizing quality. Summary of the Invention

[0005] The first technical problem to be solved by the present invention is to provide an on-line detecting device for yarn hairiness of a sizing machine, so as to solve the technical problems of lagging detection results, low detection efficiency, and inability to guide and correct the sizing process of the current sizing machine.

[0006] To solve the above technical problems, the technical solution adopted by the present invention is: an on-line yarn hairiness detection device for a sizing machine, including at least one set of image acquisition devices installed on the yarn movement path downstream of the sizing tank of the sizing machine, a set of image processing modules, and a set of information feedback modules. The image acquisition devices are communicatively connected to the image processing modules and send the acquired yarn images to the image processing modules. The image processing modules process and calculate the yarn images to obtain hairiness index data, and the feedback modules are communicatively connected to the image processing modules to feedback the processing results of the image processing modules; each set of image acquisition devices includes at least one camera for acquiring yarn images. The camera faces the yarn and takes pictures of the yarn.

[0007] As a preferred solution, there are multiple sets of the image acquisition devices, and the groups of image acquisition devices are arranged in sequence along the upstream and downstream directions of the yarn path. Each set of image acquisition devices includes multiple cameras, and the multiple cameras within the same set are arranged in sequence along the arrangement direction of each yarn in the yarn sheet. Each camera correspondingly acquires images of at least one yarn.

[0008] As a preferred solution, there is also at least one set of image acquisition devices for acquiring yarn images provided between the sizing tank and the warp beam rack. This image acquisition device is also communicatively connected to the image processing module and sends the acquired yarn images to the image processing module.

[0009] As a preferred solution, the above on-line yarn hairiness detection device for a sizing machine further includes a background board for providing a background for the image acquisition devices. The background boards correspond to the image acquisition devices one by one or one background board corresponds to one set of image acquisition devices.

[0010] As a preferred solution, the above on-line yarn hairiness detection device for a sizing machine further includes a light source for providing illumination for the image acquisition devices.

[0011] The further technical problem to be solved by the present invention is: to provide an on-line yarn hairiness detection method for a sizing machine based on the above on-line yarn hairiness detection device for a sizing machine, so as to solve the technical problems of lagging detection results, low detection efficiency, and inability to guide and correct the sizing process on the current sizing machine.

[0012] To solve the above technical problems, the technical solution adopted by the present invention is: an on-line yarn hairiness detection method for a sizing machine, using the above on-line yarn hairiness detection device for a sizing machine to detect the yarn during the sizing process, obtaining the surface hairiness of the yarn after sizing or before and after sizing and immediately feeding it back through the information feedback module. The specific steps are as follows:

[0013] 1). Use each camera to take pictures of the yarn within its shooting range, obtain the yarn images and send them to the image processing module;

[0014] 2), the image processing module numbers each received yarn image according to different image information sources for the camera position and sequence;

[0015] 3), the image processing module uses an image analysis algorithm to calculate the numbered yarn image to obtain the hairiness index of the yarn in the image;

[0016] 4), the hairiness index obtained by the processing of the image processing module is fed back through the information feedback module;

[0017] In step 3), the image analysis algorithm adopted by the image processing module specifically includes the following steps:

[0018] i). Yarn trunk segmentation:

[0019] Let I represent the yarn image, and the Otsu method can be directly used to binarize it to obtain the binary image bw of the yarn trunk core ;

[0020] ii). Yarn trunk angle and size measurement:

[0021] a. Angle:

[0022] First, extract all the connected regions in the binary image bw of the yarn trunk core . Let represent the m-th connected region among them, and solve the following formula (1) to calculate the main linear trajectory of the position where each connected region is located, which is determined by the two parameters a m , b m ;

[0023]

[0024] Thus, the angle θ of the yarn c can be calculated by the following formula (2):

[0025]

[0026] where card() is used to calculate the number of pixel points in the connected region , and M is the total number of connected regions;

[0027] b. Diameter:

[0028] Along θ c +π / 2, that is, the direction perpendicular to the yarn, search and record the number of consecutive points with a value of 1 in the binary image bw of the yarn trunk core , and find its average value to obtain the average pixel diameter d of the yarn;

[0029] c. Length:

[0030] Adopt the rectangular shape assumption, and use the binary image bw of the yarn trunkcore The total number S of the medium 1 values is divided by the average pixel diameter d of the yarn, and then the total length L of the yarn pixels in the image can be obtained;

[0031] iii). Image enhancement:

[0032] According to the yarn backbone binary image bw core Determine the positions of the yarn backbones, and calculate the average pixel values at these positions in the original yarn image I Then, the enhancement parameter γ is calculated through the following formula (3):

[0033]

[0034] where k is a calibration target parameter, and preferably, its value is 0.5. Thus, the original yarn image I can be enhanced with the parameter γ as follows:

[0035] I v = I γ ;

[0036] iv). Hairiness segmentation:

[0037] a. Decision vector calculation:

[0038] First, select a series of alternative thresholds as follows:

[0039]

[0040] where τ0 is the lower threshold parameter, and its value is 0.1; τ is the threshold step, and its value is 0.05. Then, use each alternative threshold to segment the enhanced image I v , and count the number of edge pixels in the binary image, which is recorded as the vector v e , as the decision vector;

[0041] b. Threshold selection:

[0042] Define the high threshold t h as: Determine the positions of the yarn backbones in the yarn backbone binary image bw core , and calculate the average pixel values at these positions in the enhanced yarn image I v ;

[0043] Define the low threshold t l as: The threshold at the maximum second-order gradient of the decision vector v e ;

[0044] c. Double-threshold segmentation:

[0045] For the enhanced image I v , use the binary image bw h obtained by high-threshold segmentation, and use the binary image bw obtained by low-threshold segmentationl , take bw l in the bw h All the connected regions adjacent to or covering the connected regions within, that is, the final hairiness segmentation result bw hair ;

[0046] v), Hairiness index calculation:

[0047] Calculate the edges in the hairiness segmentation result bw hair Count the total number of its pixels and divide by 2, then divide by the total length L of the yarn, that is, the hairiness index H in a yarn image e (Total hairiness length on the surface of the yarn per unit length).

[0048] As a preferred solution, use sequential images for training. For the sequential images of the yarn collected by each camera position, preprocess the first n images, and count the yarn angle θ c , average yarn diameter d, yarn length L in the image, enhancement parameter γ, high threshold t h , low threshold t l and other index means, which are used for the image analysis algorithm of the subsequent images in the sequence. Reduce the computational complexity of the image analysis algorithm and ensure the real-time processing ability of the image analysis algorithm. Preferably, n can take the value of 10.

[0049] As a preferred solution, the image processing module makes a horizontal comparison of the total hairiness length per unit length of the surface of each yarn photographed by multiple cameras within the same group of image acquisition devices downstream of the sizing vat, and feeds back the comparison results through the information feedback module.

[0050] Let represent the average hairiness index (total hairiness length on the surface of the yarn per unit length) statistically calculated for the sequential images of the yarn photographed by the i-th of multiple cameras within the same group of image acquisition devices in a time period T. Calculate the standard deviation of all , and divide this by its mean value, then the coefficient of variation of is obtained. Use the statistical characteristics such as the range, maximum value, minimum value, and coefficient of variation of as the evaluation index for the lateral quality uniformity of the sheet of yarn, and continuously record these indexes calculated within multiple time periods T, which can be plotted into a curve graph for the on-machine technicians to evaluate and make feedback adjustments.

[0051] As a preferred solution, any camera takes pictures of the yarn within its shooting range at intervals in chronological order and sends them to the image processing module. The image processing module numbers the yarn images taken by the same camera longitudinally along the time axis and processes each yarn image to obtain the total amount of hairiness per unit length of the yarn in each image, and arranges the results in chronological order in a coordinate system, connecting them with continuous multi-segment lines to form a hairiness curve graph, and feeding back the hairiness total amount values arranged in chronological order and the hairiness curve graph through the information feedback module.

[0052] The further technical problem to be solved by the present invention is: to provide an on-line regulation method for a sizing machine based on the above-mentioned on-line detection device for yarn hairiness of a sizing machine, so as to solve the technical problem that the current sizing machine cannot regulate the quality of the yarn sizing process.

[0053] To solve the above technical problems, the technical solution adopted by the present invention is: an on-line regulation method for a sizing machine based on the above-mentioned on-line detection device for yarn hairiness of a sizing machine, including the following specific steps:

[0054] a. Regularly detect the yarn in the sizing process by using the on-line detection device for yarn hairiness of the sizing machine to obtain the hairiness curve of the yarn at the same position on the time axis;

[0055] b. Adjust at least one of the three parameters of the pressing force, the vehicle speed and the slurry concentration of the sizing machine according to the change of the hairiness curve to reduce the change range of the hairiness amount of the yarn after sizing.

[0056] The beneficial effects of the present invention are as follows: First, the on-line detection device for yarn hairiness of the sizing machine described in the present invention detects the yarn in the sizing process through the image acquisition device, the image processing module and the information feedback module, obtains the real-time hairiness amount index of the yarn surface, and feeds it back to the operator in real time. The operator can adjust the process parameters of the sizing machine according to the feedback on the hairiness amount index information of the yarn surface during the sizing process, thereby improving the sizing quality.

[0057] Secondly, the on-line detection method for yarn hairiness of the sizing machine described in the present invention uses the above-mentioned on-line detection device for yarn hairiness of the sizing machine, and through means such as image acquisition, image processing, data acquisition and statistical processing, obtains the longitudinal comparison data map of the hairiness amount of the yarn surface at the same position along the time axis, the transverse comparison data of the hairiness amount of multiple yarns in the yarn sheet at the same position, and the comparison data of the total hairiness amount per unit length of the yarn sheet at different positions at the same moment. And feed it back to the operator in real time. Thus, sufficient reference data is provided for the operator to control the sizing effect, that is, the sizing quality, in the yarn sizing process. The operator can determine the adjustment range of the process parameters of the sizing machine according to the data comparison and changes, thereby improving the adjustment accuracy and further improving the sizing quality.

[0058] Furthermore, the online regulation method for the sizing machine of the present invention utilizes the above-mentioned online yarn hairiness detection device for the sizing machine. Through the comparison results and change processes of various data detected by the online yarn hairiness detection device for the sizing machine, corresponding adjustments are made to the yarn sizing process in a timely manner, even in advance, so as to perform real-time control over the yarn sizing process, eliminate the lag time of process regulation, reduce the sizing quality deviation amount in the yarn sizing process, and improve the sizing quality of the sized yarn.

[0059] Through the above technical solutions, the present invention achieves the technical effects of timely detection results, high detection efficiency, and the ability to provide real-time guidance and correction for the sizing process, which can greatly reduce the sizing quality deviation value, improve the consistency of the sizing quality of the entire yarn, and improve the sizing quality of the sizing machine for the yarn. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] The following further describes in detail the specific embodiments of the present invention with reference to the drawings, where:

[0061] Figure 1 is a schematic structural diagram of the specific structure of the online yarn hairiness detection device for the sizing machine of the present invention;

[0062] Figure 2 is Figure 1 the view in the direction of A in

[0063] Figure 3 is a yarn image captured by any one of the cameras;

[0064] Figure 4 is Figure 3 the result after the main trunk segmentation of the yarn image shown in

[0065] Figure 5 is Figure 3 the result after the adaptive enhancement of the image shown in

[0066] Figure 6 is Figure 5 the result after the low-threshold segmentation of the image shown in

[0067] Figure 7 is Figure 5 the result after the high-threshold segmentation of the image shown in

[0068] Figure 8 is Figure 5 the result after the double-threshold segmentation of the image shown in

[0069] Figure 9 is the curve of the number of edge pixel points changing with the threshold when segmenting the Figure 5 image shown in

[0070] Figure 10 is Figure 9The second-order gradient curve of the shown curve;

[0071] Figure 11 It is a curve graph showing the change of the surface hairiness value per unit length of the yarn over time obtained after processing the yarn image captured by any camera;

[0072] Figures 1 to 11 In the figure: 1. Sizing trough, 2. Image acquisition device, 201. Camera, 3. Image processing module, 4. Information feedback module, 5. Background board, 6. Light source, 7. Drying oven. Specific implementation mode

[0073] The following combines the attached drawings to describe in detail the specific implementation plan of the present invention.

[0074] Example 1:

[0075] As Figure 1 and Figure 2 shown, the on-line yarn hairiness detection device of the sizing machine includes three groups of image acquisition devices 2 installed on the yarn movement path downstream of the sizing trough 1 of the sizing machine, a set of image processing module 3 and a set of information feedback module 4. The three groups of image acquisition devices 2 are respectively communicatively connected to the image processing module 3 and send the captured yarn images to the image processing module 3. The image processing module 3 processes and calculates the yarn images to obtain the hairiness index data. The feedback module 4 is communicatively connected to the image processing module 3 and feeds back the processing results of the image processing module 3. Each group of image acquisition devices 2 includes four cameras 201 for acquiring yarn images. The cameras 201 face the yarn 100 and take pictures of the yarn 100.

[0076] In this embodiment, the three groups of image acquisition devices 2 are arranged in sequence along the upstream and downstream directions of the yarn path. One group of image acquisition devices 2 is located between the sizing trough 1 and the drying oven 7, and two groups of image acquisition devices 2 are located between the drying oven 7 and the warp beam 8. Each group of image acquisition devices 2 includes four cameras 201. The four cameras 201 in the same group are arranged in sequence along the arrangement direction of the yarns in the yarn sheet. Each camera 201 corresponds to acquire the images of at least two yarns.

[0077] In this embodiment, a background board 5 for providing a background for the image acquisition device 2 and a light source 6 for providing illumination for the image acquisition device 2 are preferably provided. The background board 5 can correspond to the image acquisition device 2 one by one or one background board 5 corresponds to a group of image acquisition devices 2. To simplify the structure, in this embodiment, one background board 5 corresponds to a group of image acquisition devices 2.

[0078] As a further optimization and improvement of this embodiment, a set of image acquisition devices 2 for acquiring yarn images is also provided between the sizing vat and the creel. The image acquisition devices 2 are also communicatively connected to the image processing module 3 and send the acquired yarn images to the image processing module 3.

[0079] After the on-line yarn hairiness detection device of the sizing machine in this embodiment is matched with the sizing machine, it can start to work. The specific working process is shown in Embodiment 2.

[0080] In this embodiment, by detecting the hairiness amount on the surface of the yarn after sizing or before and after sizing during the sizing process of the sizing machine and feeding it back to the operator, the operator can timely and accurately grasp the current working condition of the sizing machine and the real-time sizing quality, and adjust the sizing process stressfully according to the real-time change of the hairiness amount on the surface of the yarn, so as to improve the sizing quality of the yarn and ultimately improve the quality of the warp yarn.

[0081] Embodiment 2:

[0082] Reference Figures 1 to 11 As shown, the on-line yarn hairiness detection method of the sizing machine in this embodiment uses the on-line yarn hairiness detection device described in Embodiment 1 to detect the yarn during the sizing process, obtain the hairiness amount on the surface of the yarn after sizing or before and after sizing, and immediately feedback it through the information feedback module. The specific steps are as follows:

[0083] 1). Each camera is used to photograph the yarn within its shooting range, obtain the yarn image and send it to the image processing module; we number the four groups of image acquisition devices 2 in sequence along the upstream and downstream directions of the yarn: the one located at the most upstream is Group A, and the downstream ones are Group B, Group C, and Group D in sequence. The four cameras within any one group are sequentially marked as M1 to M4, and M is one of A, B, C, and D.

[0084] 2). The image processing module numbers the received yarn images according to different image information sources and in terms of camera position and sequence; for example, if the first yarn image taken by camera A1 is received, then the yarn image is numbered A1-01.

[0085] 3). The image processing module uses an image analysis algorithm to operate on the numbered yarn images to obtain the hairiness index of the yarn in the images.

[0086] 4). The information feedback module feeds back the hairiness index obtained by the processing of the image processing module.

[0087] In step 3), the image analysis algorithm adopted by the image processing module specifically includes the following steps:

[0088] i). Yarn trunk segmentation:

[0089] Let I represent the yarn image, and the Otsu method can be directly used to binarize it to obtain the binary image bw of the yarn trunk core , as Figure 3 and Figure 4 shown;

[0090] ii). Measurement of yarn trunk angle and size:

[0091] a. Angle:

[0092] First, extract all connected components in the binary image bw of the yarn trunk core . Let represent the m-th connected component among them, and solve the following formula (1) to calculate the main linear trajectory of the position of each connected component, which is determined by the two parameters a m , b m ;

[0093]

[0094] Thus, the angle θ of the yarn c can be calculated by the following formula (2):

[0095]

[0096] where card() is used to calculate the number of pixel points in the connected component , and M is the total number of connected components;

[0097] b. Diameter:

[0098] Along the direction of θ c +π / 2, that is, perpendicular to the yarn, search and record the number of consecutive points with a value of 1 in the binary image bw of the yarn trunk core , and the average value can be obtained to get the average pixel diameter d of the yarn;

[0099] c. Length:

[0100] To avoid possible errors in counting the number of yarns, in this embodiment, a rectangular shape assumption is adopted. Divide the total number S of 1 values in the binary image bw of the yarn trunk core by the average pixel diameter d of the yarn, and the total pixel length L of the yarn in the image can be obtained;

[0101] iii). Image enhancement:

[0102] As Figure 5 shown, to solve the problem of insufficient and unstable brightness of the images collected in real time on the sizing machine, the present invention proposes an adaptive image enhancement method. According to the position of the yarn trunk determined by the binary image bw of the yarn trunk core , count the average pixel values at these positions in the original yarn image I The enhancement parameter γ is calculated by the following formula (3):

[0103]

[0104] where k is the calibration target parameter, preferably taking the value of 0.5. Thus, the original yarn image I can be enhanced with the parameter γ as follows:

[0105] I v = I γ ;

[0106] iv), Hairiness segmentation:

[0107] a. Decision vector calculation:

[0108] As Figure 9 and Figure 10 shown, to solve the problem of uneven hairiness brightness, a series of alternative thresholds are first selected as follows:

[0109]

[0110] where τ0 is the threshold lower limit parameter, taking the value of 0.1; τ is the threshold step, taking the value of 0.05. Subsequently, each alternative threshold is used to segment the enhanced image I v , and the number of edge pixels in the binary image is statistically obtained and recorded as the vector v e , as the decision vector;

[0111] b. Threshold selection:

[0112] As the threshold decreases from large to small, the edges in the image will continuously increase. Initially, the increasing edges are hairiness edges, and then the noise edges in the background will gradually increase.

[0113] Some hairiness in the image has a low brightness, even as low as the level of some background noise. The present invention proposes a high-low threshold method for segmenting yarn hairiness. Among them:

[0114] Define the high threshold t h as: the average pixel value of the positions of the yarn main body determined by the binary image bw core of the yarn main body in the enhanced yarn image I v ;

[0115] Define the low threshold t l as: the threshold at the maximum second-order gradient of the decision vector v e ; The maximum second-order gradient means that the edges in the image increase sharply at this time, that is, the background begins to dominate the image segmentation result. Therefore, the hairiness can be generally segmented accurately at this time, and the noise exists but the influence is not yet severe.

[0116] c. Double-threshold segmentation:

[0117] For the enhanced image I v , a binary image bw is obtained by high-threshold segmentation h , a binary image bw is obtained by low-threshold segmentation l , take all the connected regions in bw l that are adjacent to or cover the connected regions within bw h , and the final hairiness segmentation result bw is obtained hair , as Figure 6 , Figure 7 and Figure 8 shown

[0118] v), Hairiness index calculation:

[0119] Calculate the edges in the hairiness segmentation result bw hair , count the total number of its pixels, divide by 2, and then divide by the total length L of the yarn, and the hairiness index H e (total length of hairiness on the surface of yarn per unit length) of a yarn image is obtained

[0120] As a further optimization of the above technical solution, in this embodiment, sequence images can also be used for training. For the yarn image sequence collected by each camera position, the first n images are pre-processed, and the yarn angle θ c , average yarn diameter d, yarn length L in the image, enhancement parameter γ, high threshold t h , low threshold t l and other index means are statistically calculated and used for the image analysis algorithm of the subsequent images in the sequence. The computational complexity of the image analysis algorithm is reduced, and the real-time processing ability of the image analysis algorithm is ensured. Preferably, n can take the value of 10

[0121] Furthermore, in this embodiment, an image processing module can also be used to horizontally compare the total length of hairiness on the surface of each unit length of yarn photographed by multiple cameras in the same group of image acquisition devices downstream of the sizing bath, and the comparison result is fed back through an information feedback module

[0122] As shown in Table 1, taking the image acquisition device 2 of group B as an example, the image processing module compares the total length of surface hairiness per unit length of each yarn photographed by multiple cameras 201 in the image acquisition device 2 of group B horizontally, and feeds back the comparison results through the information feedback module. The four cameras of group B take four hairiness pictures at the same time, which are numbered B1-01, B2-01, B3-01, and B4-01 respectively. Since each camera 201 in this embodiment takes images of four yarns, when processing each hairiness image, the total lengths of the hairiness on the inner surface per unit length of the four yarns can be summed for statistics, or the average value can be calculated for feedback, or feedback can be given separately. In this embodiment, the average value is preferably used for feedback, that is, after processing by the image processing module 3, the total length of the hairiness on the inner surface per unit length of the yarn in the B1-01 image is 8.4 mm, the total length of the hairiness on the inner surface per unit length of the yarn in the B1-02 image is 7.9 mm, the total length of the hairiness on the inner surface per unit length of the yarn in the B1-03 image is 7.7 mm, and the total length of the hairiness on the inner surface per unit length of the yarn in the B1-04 image is 8.2 mm.

[0123] Table 1: Transverse comparison of yarn hairiness quality taken simultaneously by each camera of the image acquisition device of group B

[0124] Yarn image number Unit length (cm) Average total length of yarn surface hairiness per unit length (mm) B1-01 1 8.41 B2-01 1 7.95 B3-01 1 7.74 B4-01 1 8.26

[0125] The operator sets the image processing module according to his needs, and can retrieve the statistical data in Table 1 and provide feedback on the information feedback module 4. The information feedback module 4 is mainly a display, and can also be combined with other output devices such as alarms and loudspeakers. The operator can understand the changes in the sizing effect of various parts of the sizing machine roller on the yarn in combination with the data information provided in Table 1 that is statistically analyzed in chronological order, and timely adjust the state of the sizing roller according to the changes in the sizing effect. For example, if there are more hairiness in the sizing roller locally, resulting in poor sizing quality of the yarn passing through this part, it will be reflected in the fact that the average total length of hairiness on the yarn surface per unit length of a certain hairiness image in Table 1 increases significantly. The operator can understand which part of the sizing roller has a problem based on the hairiness image number, and deal with it in time to avoid further or continuous abnormalities in the sizing quality of the yarn, resulting in a decrease in the sizing quality.

[0126] You can also It represents the average hairiness index (total length of hairiness on the yarn surface per unit length) of the yarn image sequence taken by the i-th camera in the same group of image acquisition devices in a time period T. Dividing this by its mean, we get The coefficient of variation of The statistical characteristics such as the range, maximum value, minimum value, coefficient of variation, etc. are used as evaluation indicators for the transverse quality uniformity of the yarn, and these indicators calculated within multiple time periods T are continuously recorded. They can be drawn into a curve graph for on-machine technicians to evaluate and make feedback adjustments.

[0127] As a further optimization of the present technical solution, in this embodiment, the time difference between the photos taken by the yarn image acquisition modules 2 of group A and group B is adjusted according to the moving speed of the yarn, and the two groups of yarn image acquisition modules 2 take the same section of yarn as the standard, that is, after the yarn image acquisition module 2 of group A takes the image of a certain section of the yarn, when the section of yarn moves to the yarn image acquisition module 2 of group B, the yarn image acquisition module 2 of group B takes the image of the section of yarn again. The shooting time of the yarn image acquisition modules 2 of groups B to D can be set at will. After the image processing module 3 accumulates the total length of surface hairiness per unit length of each yarn captured by multiple cameras 201 in the yarn image acquisition modules 2 of groups A, B, C, and D, the total length of surface hairiness of the yarn collected by each group of yarn image acquisition modules 2 is compared according to the upstream and downstream relationship, and the comparison results are fed back through the information feedback module, as shown in Table 2 below:

[0128] Table 2: Comparison of yarn hairiness quality captured by four sets of image acquisition devices

[0129] Hairiness image number Unit length (cm) Total length of yarn surface hairiness per unit length (mm) A-01 1 8.26 B-01 1 3.22 C-01 1 2.87 D-01 1 2.89

[0130] In Table 2, A represents the yarn image acquisition module 2 of group A, and A-01 represents the set of the first hairiness images taken by all cameras 201 in the yarn image acquisition module 2 of group A.

[0131] According to the above Table 2, the change of the yarn surface hairiness before and after sizing can be fed back to the operator. The image processing module 3 can also fit the functional relationship between the total length of the yarn surface hairiness per unit length before and after sizing based on a large amount of data, and according to this relationship, the total length of the yarn surface hairiness per unit length after sizing is calculated according to the change of the total length of the yarn surface hairiness per unit length before sizing, and according to the yarn moving speed, when the photographed section of the yarn reaches the sizing roller, the sizing parameters of the sizing machine are corrected to offset the change of the total length of the yarn surface hairiness per unit length before sizing on the total length of the yarn surface hairiness per unit length after sizing. This technical means is mainly used in the case where the total length of the yarn surface hairiness per unit length before sizing becomes longer. For the case where the total length of the yarn surface hairiness per unit length before sizing becomes shorter, the sizing parameters do not need to be corrected.

[0132] The above-mentioned means can effectively eliminate the lag problem of the traditional sizing detection method in guiding the process adjustment of the sizing machine.

[0133] Of course, you can also They respectively represent the average hairiness index (total length of yarn hairiness on the surface of unit length of yarn) statistically calculated from the yarn image sequences captured by multiple cameras in the front and rear groups of image acquisition devices on the sizing machine within a time period T. Calculate the difference between the two as the comparison and evaluation index for the yarn quality before and after the process treatment, and continuously record the calculated index within multiple time periods T. A curve graph can be plotted for the reference of on-machine technicians for feedback adjustment.

[0134] According to Table 2 above, the operator can also analyze the change in the hairiness amount on the surface of the yarn before and after the drying chamber 7, providing basic data for subsequent further technical improvement.

[0135] Such as Figure 11 shown, any camera 201 takes pictures of the yarn within its shooting range at intervals in chronological order and sends them to the image processing module 3. The image processing module 3 longitudinally sorts and numbers the yarn images taken by the same camera 20 according to the time axis and processes each yarn image to obtain the total amount of hairiness per unit length of the yarn in each image. When there are multiple yarns in one image, the image processing module 3 can feedback the total amount of hairiness per unit length of each yarn or the average value of the total amount of hairiness per unit length of multiple yarns according to the operator's setting. Then the image processing module 3 arranges the results in the coordinate system in chronological order, connects them with continuous multi-segment lines to form a hairiness amount curve graph, and feeds back the hairiness total amount values arranged in chronological order and the hairiness amount curve graph through the information feedback module.

[0136] This embodiment takes the camera B1 in the B group of image acquisition devices 2 as an example. Table 3 below is the total amount of hairiness per unit length of the yarn output from multiple hairiness images taken by the camera B1 over time. In this embodiment, the camera B1 simultaneously takes pictures of four yarns, and in this embodiment, the average value of the total length of yarn hairiness on the surface of two yarns per unit length is output.

[0137] Table 3: Data table of the hairiness quality of the yarn taken by the camera B1:

[0138] Hairiness image number Unit length (cm) Interval duration (s) Average total length of yarn surface hairiness per unit length (mm) B1-01 1 20 4.21 B1-02 1 20 4.43 B1-03 1 20 4.14 B1-04 1 20 3.85 B1-05 1 20 3.92 B1-06 1 20 3.73 B1-07 1 20 4.04 B1-08 1 20 4.16 B1-09 1 20 3.98 B1-10 1 20 4.10

[0139] Combined with Table 3 and Figure 11 , the operator can monitor the sizing quality of the yarn in real time and correct the continuously expanding deviation to ensure the sizing quality. If combined with the hairiness amount curve graphs of the sized yarn surface before sizing output by each camera in the A group of image acquisition base devices 2 at the same time, it is also possible to compare the influence degree of the process adjustment degree on the sizing effect by adjusting the sizing process, so as to form a set of precise pre-adjustment processes, pre-adjust the sizing process according to the increase value of the hairiness amount on the surface of the yarn before sizing, and use the adjustment of the sizing process to keep the hairiness amount on the surface of the sized yarn stable. Thus, the sizing quality can be effectively improved.

[0140] Embodiment 3:

[0141] An on-line regulation method for a sizing machine based on the above-mentioned on-line yarn hairiness detection device for a sizing machine, comprising the following specific steps:

[0142] a. Regularly detect the yarn during the sizing process by using the on-line yarn hairiness detection device for a sizing machine to obtain the hairiness amount curve of the yarn at the same position on the time axis, as Figure 8 shown; the detection method of the on-line yarn hairiness detection device for a sizing machine is as shown in Embodiment 2.

[0143] b. Adjust at least one of the three parameters of the pressing force, the vehicle speed, and the slurry concentration of the sizing machine according to the change of the hairiness amount curve, so as to reduce the change range of the yarn hairiness amount after sizing.

[0144] In this embodiment, the hairiness amount curve of Group A and the hairiness amount curve of Group B can be further compared, and the change of the hairiness amount curve of Group B can be predicted according to the hairiness amount curve of Group A. When the hairiness amount curve of Group A shows an obvious unilateral expansion beyond the safety margin, according to the running speed of the yarn, when the yarn with an increasing hairiness amount reaches the sizing vat 1 and starts sizing, adjust the sizing process to eliminate the influence of the increase in the yarn hairiness amount before sizing on the surface hairiness amount of the yarn after sizing, and improve the sizing quality.

[0145] The above embodiments only illustratively explain the principle and efficacy of the present invention and some applied embodiments, rather than limiting the present invention; it should be noted that for those of ordinary skill in the art, without departing from the inventive concept of the present invention, several deformations and improvements can be made, and these all belong to the protection scope of the present invention.

Claims

1. Method for on-line detecting hairiness of sizing machine yarns, characterized in that, Use the on-line yarn hairiness detection device of the sizing machine to detect the yarn during the sizing process, obtain the hairiness amount on the surface of the yarn after sizing or before and after sizing, and immediately feedback it through the information feedback module. The specific steps are as follows: 1). Use each camera to take pictures of the yarn within its shooting range, obtain the yarn image and send it to the image processing module; 2). The image processing module numbers the position and sequence of each received yarn image according to different image information sources; 3). The image processing module uses an image analysis algorithm to operate on the numbered yarn image to obtain the hairiness index of the yarn in the image; 4). The information feedback module feeds back the hairiness index obtained by the processing of the image processing module; In step 3), the image analysis algorithm adopted by the image processing module specifically includes the following steps: ⅰ). Yarn trunk segmentation: Let I denote the yarn image, and it can be directly binarized by Otsu's method to obtain the binary image bw of the yarn trunk core ; ⅱ). Measurement of yarn trunk angle and size: a. Angle: First, extract all the connected components in the binary image bw of the yarn trunk core Let represent the m-th connected component among them, and solve the following formula (1) to calculate the main linear trajectory of the position where each connected component is located, which is determined by the two parameters a m and b m ; Thus, the angle θ of the yarn c can be calculated by the following formula (2): where card() is used to calculate the number of pixels in the connected component , M is the total number of connected components; b. Diameter: Along θ c +π / 2, that is, in the direction perpendicular to the yarn, search and record the number of consecutive points with a value of 1 in the binary image bw of the yarn main body core and obtain the average value to get the average pixel diameter d of the yarn; c. Length: Using the rectangular shape assumption, the binary image bw of the yarn trunk core divides the total number S of 1 values in it by the average pixel diameter d of the yarn, and then the total length L of the yarn pixels in the image can be obtained; ⅲ). Image enhancement: According to the binary image bw of the yarn backbone core Determine the positions of the yarn backbone, and count the average pixel values at these positions in the original yarn image I Thus, calculate the enhancement parameter γ through the following formula (3): Where k is the calibration target parameter, and its value is 0.

5. Thus, the original yarn image I can be enhanced with parameter γ as follows: I v = I γ ; ⅳ). Hairiness segmentation: a. Decision vector calculation: First, select a series of alternative thresholds as follows: where τ0 is the lower threshold parameter with a value of 0.1; τ is the threshold step with a value of 0.

05. Subsequently, each candidate threshold is used to segment and enhance the image I v , and the number of edge pixels in the binary image is statistically obtained and recorded as a vector v e , which serves as the decision vector; b. Threshold selection: Define the high threshold t h as: the binary image bw of the yarn backbone core Determine the positions of the yarn backbone, and count the average pixel values at these positions in the yarn enhanced image I v among them; Define the low threshold t l as: the threshold at the maximum of the second-order gradient of the decision vector v e ; c. Double-threshold segmentation: For the enhanced image I v , a binary image bw is obtained by high-threshold segmentation h , a binary image bw is obtained by low-threshold segmentation l , take all the connected components in bw l that are adjacent to or cover the connected components within bw h , and the final feather segmentation result bw is obtained hair ; ⅴ). Hairiness index calculation: Calculate the hairiness segmentation result bw hair For the edges in it, count the total number of pixels and divide it by 2, then divide by the total length L of the yarn, and the hairiness index H in a yarn image can be obtained e ; The on-line yarn hairiness detection device of the sizing machine includes at least one set of image acquisition devices (2) installed on the yarn movement path downstream of the sizing machine sizing tank (1), a set of image processing modules (3) and a set of information feedback modules (4). The image acquisition device (2) is communicatively connected to the image processing module (3) and sends the acquired yarn image to the image processing module (3). The image processing module processes and calculates the yarn image to obtain the hairiness amount index data. The feedback module (4) is communicatively connected to the image processing module (3) and feeds back the processing result of the image processing module (3); each set of image acquisition devices (2) includes at least one camera (201) for acquiring yarn images. The camera (201) faces the yarn and takes pictures of the yarn.

2. The on-line detecting method for yarn hairiness of a sizing machine according to claim 1, characterized in that Training is carried out using a sequence of images. For the sequence of yarn images collected at each camera position, the first n images are pre-processed, and the yarn angle θ c , the average diameter d of the yarn, the length L of the yarn in the image, the enhancement parameter γ, the high threshold t h , and the low threshold t l are averaged. These averages are used in the image analysis algorithm for subsequent images in the sequence.

3. The method for on-line detection of yarn hairiness of a sizing machine according to claim 1, characterized in that, The image processing module makes a horizontal comparison of the total hairiness length per unit length of the surface of each yarn photographed by multiple cameras within the same set of image acquisition devices downstream of the sizing tank, and feeds back the comparison result through the information feedback module.

4. The method for on-line detection of yarn hairiness of a sizing machine according to claim 1, characterized in that, Any camera takes pictures of the yarn within its shooting range at intervals in chronological order and sends them to the image processing module. The image processing module numbers the yarn images taken by the same camera longitudinally according to the time axis and processes each yarn image to obtain the total hairiness amount per unit length of the yarn in each image. The results are arranged in chronological order in a coordinate system and connected by continuous multi-segment lines into a hairiness amount curve graph. The total hairiness amount values arranged in chronological order and the hairiness amount curve graph are fed back through the information feedback module.

5. The method for on-line detection of yarn hairiness of a sizing machine according to claim 1, characterized in that, The image acquisition devices (2) are in multiple groups, and the image acquisition devices (2) in each group are arranged in sequence along the upstream and downstream directions of the yarn path. Each group of image acquisition devices (2) includes multiple cameras (201). The multiple cameras (201) within the same group are arranged in sequence along the arrangement direction of each yarn in the yarn sheet, and each camera (201) corresponds to acquiring images of at least one yarn.

6. The on-line detecting method for yarn hairiness of a sizing machine according to claim 5, characterized in that, At least one group of image acquisition devices (2) for acquiring yarn images is further provided between the sizing bath and the warp beam rack. This image acquisition device (2) is also communicatively connected to the image processing module (3) and sends the acquired yarn images to the image processing module (3).

7. The on-line detecting method for yarn hairiness of a sizing machine according to any one of claims 1, 5, and 6, characterized in that It further includes a background board (5) for providing a background for the image acquisition device (2). The background board (5) corresponds one-to-one with the image acquisition device (2) or one background board (5) corresponds to one group of image acquisition devices (2).

8. The on-line hairiness detection method for sizing machine yarn according to claim 7, characterized in that It further includes a light source (6) for providing illumination for the image acquisition device (2).

Images