Double-needle bar warp knitting machine broken yarn detection device based on machine vision and detection method thereof

Abstract

The invention discloses a double-needle bar warp knitting machine broken yarn detection device based on machine vision, which comprises cross beams arranged on the front side and the rear side of a guide bar of a double-needle bar warp knitting machine, a sliding table guide rail is arranged on the cross beams, one end of a guide rail body of the sliding table guide rail is connected with a driving motor, an image acquisition module is fixed on a sliding table, and the double-needle bar warp knitting machine broken yarn detection device further comprises a position sensor, the position sensor is triggered when all the yarns are clearly presented in the knitting areas of the front needle bar and the rear needle bar, the image acquisition module is connected with a computer through a wire, the position sensor, the driving motor and the computer are jointly connected with a controller through wires, and the controller is further connected with a control box of the double-needle-bar warp knitting machine through a wire. The problem that in the prior art, a detection component is complex in structure and difficult to install on the two-needle bar warp knitting machine is solved, and real-time detection of broken yarn can be achieved. The invention further discloses a broken yarn detection method of the double-needle bar warp knitting machine based on machine vision.

Description

Technical Field

[0001] The present invention belongs to the technical field of textile broken yarn detection equipment, and relates to a broken yarn detection device for a double needle bed warp knitting machine based on machine vision. The present invention also relates to a broken yarn detection method for a double needle bed warp knitting machine based on machine vision. Background Art

[0002] During the production process of a double needle bed warp knitting machine, due to high speed, multiple layers of yarns, and mutual occlusion of yarns in different layers, it is very difficult to detect broken yarns with existing methods once they occur, resulting in a large number of defective products. There are mainly two existing detection methods. One is to use an infrared laser detection device, which determines whether the yarn breaks by using the occlusion or reflection characteristics of an object to infrared laser. When the yarn does not break, the signal received by the signal receiving end of the infrared laser detection device is stable. When the yarn breaks, the broken yarn will occlude or change the propagation path of the laser, and the signal received by the signal receiving end of the infrared laser detection device changes, and it is determined that the yarn breaks. This detection method has a low detection accuracy and has large false detection and missed detection problems, and can only be used as an auxiliary. The other is to use the method of machine vision, which collects and analyzes the fabric surface image through a single or multiple cameras to determine whether there is a broken yarn, rather than directly detecting broken yarns in real time during the machine knitting process. This method is currently mostly used for single needle bed warp knitting machines, and there is no mature product for the detection of double needle bed warp knitting machines. Coupled with the complex structure, it is not conducive to the intelligent transformation of the equipment.

[0003] In addition, the patent with application number and publication number CN108318496A proposes an in-line fabric defect detection device for a warp knitting machine, which controls the on-off state of the camera through a cam mechanism, dynamically adjusts the number of cameras participating in the work according to the fabric width, and uses multiple flattening roller units to flatten the fabric to reduce the interference of wrinkles on imaging. The detected image is wirelessly transmitted to the terminal device to analyze the type of defect. This method detects the fabric that has already been produced, rather than directly detecting broken yarns in real time during the machine knitting process. In addition, this method is only suitable for single needle bed warp knitting machines. Due to the complex structure and limited space of double needle bed warp knitting machines, such detection equipment cannot be installed on double needle bed warp knitting machines.

[0004] Therefore, how to detect the broken yarn problem of a double needle bed warp knitting machine in real time, comprehensively, quickly and accurately, so as to avoid a large amount of waste of raw materials, and at the same time reduce labor and save time and effort is a key problem that urgently needs to be solved in the production process of double needle bed warp knitting machines at present. Summary of the Invention

[0005] The purpose of the present invention is to provide a broken yarn detection device for a double needle bed warp knitting machine based on machine vision, which solves the problems of complex structure of the detection components and difficult installation on a double needle bed warp knitting machine in the prior art, and can realize real-time detection of broken yarns.

[0006] Another object of the present invention is to provide a broken yarn detection method for a double needle bed warp knitting machine based on machine vision.

[0007] The technical solution adopted by the present invention is that a broken yarn detection device for a double needle bed warp knitting machine based on machine vision includes cross beams arranged on the front and rear sides of the comb of the double needle bed warp knitting machine. A slide rail is arranged on the cross beam. One end of the rail body of the slide rail is connected with a driving motor. The driving motor drives the slide of the slide rail to move left and right on its rail body. An image acquisition module is fixed on the slide. A position sensor is also included. The position sensor is used for: when all yarns are clearly presented in the knitting areas of the front and rear needle beds, the position sensor is triggered. The image acquisition module is connected to a computer through a wire. The position sensor, the driving motor, and the computer are all connected to a controller through wires. The controller is also connected to the control box of the double needle bed warp knitting machine through a wire.

[0008] Preferably, the installation position of the position sensor is: when the comb swings back and forth and horizontally moves left and right to the position where all yarns are clearly presented in the knitting areas of the front and rear needle beds, the position sensor identifies the position of the comb at this time; or when the needle bed reciprocates to the position where all yarns are clearly presented in the knitting areas of the front and rear needle beds, the position sensor identifies the position of the needle bed at this time.

[0009] Preferably, the position sensor adopts one of a photoelectric switch, an inductive proximity switch, a Hall switch, a wire-drawing type position switch, and an angular displacement sensor.

[0010] Preferably, the position sensor adopts a photoelectric switch module. The photoelectric switch module includes a fixed mounting bracket mounted on the comb fixing frame. A photoelectric switch is mounted on the fixed mounting bracket. The photoelectric switch is connected to the controller.

[0011] Preferably, connecting brackets are respectively and fixedly arranged at the left and right ends of the frame of the double needle bed warp knitting machine. The two connecting brackets are respectively located on the front and rear sides of the comb. The two cross beams are respectively fixed at the other ends of the corresponding connecting brackets. The two cross beams are on the front and rear sides of the comb and extend from the leftmost end to the rightmost end of the comb. One or two slides are slidably arranged on the rail body of each slide rail. The driving motor drives the slide to move on the rail body of the slide rail, so as to realize full coverage of photographing the yarns in the knitting areas of the front and rear needle beds by the image acquisition module.

[0012] Preferably, the image acquisition module includes a mounting bracket fixed on the slide. A camera facing the yarns in the knitting areas of the front and rear needle beds is fixed on the mounting bracket. The camera is electrically connected to the computer through a wire; A light source and a fan are also fixed on the mounting bracket. The light source is used to illuminate the yarns in the knitting areas of the front and rear needle beds to provide a bright environment for the camera to take pictures. The fan is used to cool the light source. The light source and the fan are both electrically connected to the controller through wires.

[0013] The second technical solution adopted by the present invention is a double needle bed warp knitting machine yarn breakage detection method based on machine vision, which adopts the above-mentioned double needle bed warp knitting machine yarn breakage detection device based on machine vision, and is specifically implemented according to the following steps: Step 1, after the double needle bed warp knitting machine is working, the controller controls the driving motor to start, and the slide moves left and right on the guide rail body of the slide rail. When all the yarns are clearly presented in the front and rear needle bed knitting areas, the position sensor is triggered, and the controller controls the camera to take pictures; Step 2, the camera transmits the collected yarn image of the weaving area to the computer in real time; Step 3, the computer determines whether there is yarn breakage based on the collected yarn image; Step 4: If yarn breakage is detected, the computer sends a stop signal to the controller, and the controller controls the double needle bed warp knitting machine to stop running, otherwise it continues to run normally.

[0014] Preferably, step 3 is specifically: Step 3.1, the computer receives the yarn image and then uses the image capture algorithm to capture the yarn image of the weaving area, captures the yarn image of the weaving area, retains the complete yarn, and removes useless image information; Step 3.2, the image captured in step 3.1 is enhanced by using histogram equalization and mean filtering; Among them, histogram equalization is to perform grayscale histogram equalization processing on the image intercepted in step 3.1 to calculate the final grayscale level, specifically:

[0015] Among them, Round to the nearest integer, which is the final grayscale. is the mapping function, Represents grayscale levels from 0 to The cumulative probability of k is the original gray level of the image. In the range, is the probability of the original gray level i appearing:

[0016] in, is the total number of pixels in the image, is the number of pixels with gray level i; Then the image after grayscale histogram equalization is processed by mean filtering, specifically: Define a The filter kernel traverses the image after grayscale histogram equalization pixel by pixel, calculates the average value of all pixels in the window, and assigns the average value to the center pixel. , its filtered value It is:

[0017] Among them, ; An enhanced image after mean filtering is obtained; Step 3.3: Determine the yarn tilt angle and calculate the stitch pitch pixel number according to the yarn tilt angle; Step 3.4: According to the yarn tilt angle determined in Step 3.3, use the gray integral projection algorithm to calculate the corresponding projection curve at the yarn tilt angle for the image after image enhancement in Step 3.2. Each peak in the projection curve represents a yarn. Then calculate the distance between adjacent peaks in the projection curve. When the distance between two adjacent peaks in the projection curve is greater than or equal to the stitch pitch pixel number in Step 3.3, it means that the distance between the peaks is too large, exceeding the normal distance between two yarns, that is, there is no yarn between two needles, that is, there is a broken yarn; otherwise, it means that the distance between the yarns is normal and there is no broken yarn.

[0018] Preferably, the specific method for determining the yarn tilt angle in Step 3.3 is: Pre-collect a group of M images without broken yarns. Then, after image cropping, histogram equalization, and mean filtering for each image according to the methods in Steps 3.1 - 3.2, determine an angle range ( ), and use the gray integral projection algorithm to calculate the projection curves of each image at , , ,... ; For any angle , , obtain the projection curves of M images at the angle , that is, obtain M projection curves at any angle . Then calculate the average value of the peaks of the M projection curves at any angle , and respectively obtain the corresponding average values at the angles , , ,... , a total of . Then, take the angle corresponding to the largest average value among the average values as the yarn tilt angle; The specific method for calculating the stitch pitch pixel number according to the yarn tilt angle is: According to the obtained yarn tilt angle, the gray integral projection algorithm is used to calculate the projection curve of each image at the yarn tilt angle, obtaining M projection curves. Calculate the distance between adjacent two peaks in each projection curve. Assuming there are b adjacent peaks in the corresponding projection curve, then b distances are obtained. Then calculate the average value of the b distances corresponding to each projection curve, obtaining M average values. Then, take the average of the M average values to obtain the value, which is the stitch pixel number.

[0019] Preferably, for an image with n rows m and I columns, i and j ), the gray value of the image at the position ( i and j ) is, then the projection curve of the image at any angle θ is as follows: .

[0020] The beneficial effects of the present invention are: The present invention detects the position of the periodic moving parts associated with the movement of the yarn at the knitting mouth through a position sensor. When all the yarns are clearly presented on the front needle bed, it triggers the image acquisition module to collect images. The image acquisition module moves left and right on the guide rail body of the slide rail through the slide table to achieve full coverage of the knitting areas of the front and back needle beds. At the same time, it takes images of the yarn at this time under external triggering and transmits them to the computer to perform real-time broken yarn detection using the gray integral projection algorithm. When a broken yarn is detected, it will send a shutdown signal to the machine. The present invention realizes real-time and comprehensive broken yarn detection of a double-needle bed warp knitting machine by the back-and-forth movement of 2 to 4 cameras, and ingeniously uses the existing structure of the warp knitting machine to install this detection device, with a simple mechanical structure, low cost, and no impact on the daily operation of workers. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 is a schematic structural diagram of the broken yarn detection device for a double-needle bed warp knitting machine based on machine vision according to the present invention; Figure 2 is a schematic structural diagram of the photoelectric switch module in the broken yarn detection device for a double-needle bed warp knitting machine based on machine vision according to the present invention; Figure 3 is a schematic installation structure diagram of the cross beam and the slide rail in the broken yarn detection device for a double-needle bed warp knitting machine based on machine vision according to the present invention; Figure 4 is a schematic structural diagram of the image acquisition module in the broken yarn detection device for a double-needle bed warp knitting machine based on machine vision according to the present invention; Figure 5 is a diagram showing the distribution of multi-layer yarns in the knitting areas of the front and back needle beds obtained by shooting in the embodiment of the present invention; Figure 6 is the flow chart of the broken yarn detection method for double needle bed warp knitting machines based on machine vision of the present invention; Figure 7 is the flow chart of broken yarn judgment in the broken yarn detection method for double needle bed warp knitting machines based on machine vision of the present invention; Figure 8 is the detection flow chart of the gray integral projection algorithm in the broken yarn detection method for double needle bed warp knitting machines based on machine vision of the present invention.

[0022] In the figure: 1. Frame, 2. Guide bar, 3. Guide bar fixing frame, 4. Photoelectric switch module, 5. Connecting bracket, 6. Cross beam, 7. Slide table guide rail, 8. Slide table, 9. Image acquisition module, 10. Driving motor, 11. Knitting area of front and back needle beds; 41. Fixed installation bracket, 42. Photoelectric switch; 91. Light source, 92. Fan, 93. Camera, 94. Optical lens, 95. Installation bracket. Specific embodiments

[0023] The following is a detailed description in combination with specific embodiments.

[0024] Embodiment 1 The broken yarn detection device for double needle bed warp knitting machines based on machine vision of the present invention has a structure as Figure 1 shown, including a cross beam 6 arranged on both the front and rear sides of the guide bar 2 of the double needle bed warp knitting machine. A slide table guide rail 7 is arranged on the cross beam 6. One end of the guide rail body of the slide table guide rail 7 is connected with a driving motor 10. The driving motor 10 drives the slide table 8 of the slide table guide rail 7 to move left and right on its guide rail body. An image acquisition module 9 is fixed on the slide table 8. It also includes a position sensor, which is used for: when all the yarns are clearly presented in the knitting area 11 of the front and back needle beds, the position sensor is triggered. The image acquisition module 9 is connected to a computer through a wire. The position sensor, the driving motor 10, and the computer are all connected to a controller through wires together. The controller is also connected to the control box of the double needle bed warp knitting machine through a wire.

[0025] The working principle of this embodiment is as follows: Since the double needle bed warp knitting machine is a periodic process, at a certain moment in a cycle, the camera takes a photo of the yarn at the fabric mouth. At this moment, clear images of several layers of yarn can be taken in the same photo. The computer analyzes this image to determine whether there is a broken yarn in this photo. The role of the position sensor is to detect the above moment. The controller sends a photo-taking signal to the image acquisition module 9 to take a photo of the yarn at the fabric mouth at this moment. Every time the position sensor gives a photo-taking signal, the image acquisition module 9 takes a photo.

[0026] Embodiment 2 On the basis of Embodiment 1, the position sensor for detecting the photographing moment can be installed at different positions. Any periodic moving part on the equipment associated with the movement of the yarn can become the part detected by the position sensor. When the part moves to a certain position, it is the photographing moment. The moving parts that the position sensor can detect include the guide bar, the needle bed that swings back and forth, etc. The photographing moment can be determined by measuring the position of the guide bar on the left and right; the photographing moment can also be determined by measuring the position of the needle bed that swings back and forth.

[0027] Therefore, the installation position of the position sensor is as follows: when the guide bar 2 swings back and forth and horizontally moves left and right to the position where all the yarns are clearly presented in the knitting area 11 of the front and rear needle beds, the position sensor identifies the position of the guide bar 2 at this time; or when the needle bed swings back and forth to the position where all the yarns are clearly presented in the knitting area 11 of the front and rear needle beds, the position sensor identifies the position of the needle bed at this time.

[0028] In the present invention, the method of measuring the position of the guide bar is preferably adopted to determine the photographing moment. Its advantages are that it is more convenient, reliable to measure the position of the guide bar, and the mechanism is simple.

[0029] Embodiment 3 On the basis of Embodiment 2, the position sensor adopts one of a photoelectric switch, an inductive proximity switch, a Hall switch, a wire-drawing type position switch, and an angular displacement sensor. The present invention preferably adopts a photoelectric switch, and its advantages are flexible and convenient to install on a warp knitting machine. Specifically: The position sensor adopts a photoelectric switch module 4, such as Figure 2 shown. The photoelectric switch module 4 includes a fixed mounting bracket 41 mounted on the guide bar fixing frame 3. A photoelectric switch 42 is mounted on the fixed mounting bracket 41, and the photoelectric switch 42 is connected to a controller.

[0030] In this embodiment, the position sensor adopts a photoelectric switch module 4. The photoelectric switch module 4 is fixed on the guide bar fixing frame 3. Since the weaving is a periodic operation, the states of the yarns in the knitting areas 11 of the front and rear needle beds are different at different times, and the positions of the guide bar moving left and right are also different. The photoelectric switch 42 is used to sense the position of the guide bar 2. When it is determined that all the yarns are clearly presented in the front and rear needle bed knitting areas, the photoelectric switch 42 is triggered. The computer controls that since the weaving is a periodic operation, the states of the yarns in the knitting areas 11 of the front and rear needle beds are different at different times, and the positions of the guide bar moving left and right are also different. The photoelectric switch 42 is used to sense the position of the guide bar 2. When it is determined that all the yarns are clearly presented in the front and rear needle bed knitting areas, the photoelectric switch 42 triggers the camera 93 to start taking pictures.

[0031] Regarding the position of the optoelectronic switch module 4, it is not limited to being installed on the comb fixed frame 3, and there are various installation methods: ① It can be installed on the left and right comb fixed frames 3 respectively; ② It can be directly installed at the end of the comb 2; ③ The shaft swings back and forth, and it is directly installed on the shaft. The installation position only needs to ensure that when all the yarns are clearly presented in the knitting area of the front and rear needle beds, the optoelectronic switch 42 is triggered.

[0032] Embodiment 4 On the basis of Embodiment 3, connection brackets 5 are respectively fixedly arranged at the left and right ends of the double-needle bed warp knitting machine frame 1. The two connection brackets 5 are respectively located on the front and rear sides of the comb 2. The two cross beams 6 are respectively fixed at the other ends of the corresponding connection brackets 5. The two cross beams 6 are on the front and rear sides of the comb 2 and extend from the leftmost end to the rightmost end of the comb 2, as Figure 3 shown. One or two sliders 8 are slidably arranged on the guide rail body of each slider guide rail 7. The driving motor 10 drives the slider 8 to move on the guide rail body of the slider guide rail 7, so as to realize the full coverage of the yarn in the knitting area 11 of the front and rear needle beds by the image acquisition module 9 taking pictures.

[0033] In this embodiment, the image acquisition module is installed on the slider 8. The slider 8 can be driven by the driving motor 10, and the image acquisition module 9 slides left and right on the guide rail body of the slider guide rail 7 along with the slider 8, so as to realize the full coverage of the yarn in the knitting area 11 of the front and rear needle beds by the camera taking pictures. The slider guide rail is installed on the cross beam, and both ends of the cross beam are fixed on the frame 1 of the double-needle bed warp knitting machine. At least one slider guide rail needs to be installed on the cross beam. The purpose of using the cross beam is to provide an installation base for the slider guide rail, strengthen the slider guide rail, and reduce the deformation of the slider guide rail. If the strength of the slider guide rail is sufficient, the cross beam can also be not used.

[0034] As Figure 4 shown, the image acquisition module 9 includes a mounting bracket 95 fixed on the slider 8. A camera 93 facing the yarn in the knitting area 11 of the front and rear needle beds is fixed on the mounting bracket 95. The camera 93 is electrically connected to the computer through a wire, and an optical lens is installed on the camera 93.

[0035] A light source 91 and a fan 92 are also fixed on the mounting bracket 95. The light source 91 is used to illuminate the yarn in the knitting area 11 of the front and rear needle beds to provide a bright environment for the camera 92 to take pictures. The fan 92 is used to cool the light source 91 to ensure that the light source can work stably for a long time. Both the light source 91 and the fan 92 are electrically connected to the controller through wires.

[0036] The light source can be an LED light source or other light sources. It is preferably an LED light source, and its advantages are high brightness, small power, and stable and reliable. The color of the light source can be white light, red light, blue light or other colors of light. It is preferably white light and red light, and its advantage is that the yarn image in the camera image is clearer and more prominent, which is convenient for analysis.

[0037] Example 5 The double needle bed warp knitting machine yarn breakage detection method based on machine vision of the present invention adopts the double needle bed warp knitting machine yarn breakage detection device based on machine vision in Example 4, and its process is as follows: Figure 6 and 8 As shown, the specific steps are as follows: Step 1, after the double needle bed warp knitting machine is working, the controller controls the driving motor 10 to start, and the slide 8 moves left and right on the guide rail body of the slide rail 7. Since weaving is a periodic operation process, the yarn states of the front and rear needle bed knitting areas 11 are different at different times. When all the yarns are clearly presented in the front and rear needle bed knitting areas 11, the position sensor is triggered, and the controller controls the camera 93 to take pictures; Step 2, the camera 93 transmits the collected yarn image of the weaving area to the computer in real time, such as Figure 5 As shown, it is the distribution of multiple layers of yarn in the front and rear needle bed knitting area 11, and the oblique lines with different inclination angles in the figure represent different layers of yarn; Step 3, the computer determines whether there is yarn breakage based on the collected yarn image; Step 4: If yarn breakage is detected, the computer sends a stop signal to the controller, and the controller controls the double needle bed warp knitting machine to stop running, otherwise it continues to run normally.

[0038] Example 6 Based on Example 5, the process of step 3 is as follows Figure 7 As shown, specifically: Step 3.1, the computer receives the yarn image and then uses the image capture algorithm to capture the yarn image of the weaving area, captures the yarn image of the weaving area, retains the complete yarn, and removes useless image information; Step 3.2, the image captured in step 3.1 is enhanced by using histogram equalization and mean filtering; Among them, histogram equalization is to perform grayscale histogram equalization on the original image, converting its non-uniform distribution into an approximately uniform distribution, so that the details of the dark and bright areas can be clearly presented. The image intercepted in step 3.1 is processed by grayscale histogram equalization to calculate the final grayscale level, specifically:

[0039] Among them, Round to the nearest integer, which is the final grayscale. is the mapping function, Represents grayscale levels from 0 to The cumulative probability of k is the original gray level of the image. Within the range, The probability of the original gray level i appearing:

[0040] Among them, is the total number of pixels in the image, is the number of pixels with gray level i; Then, the image after gray histogram equalization is subjected to mean filtering. Mean filtering is a linear spatial filtering technique used in image processing to remove noise and achieve image smoothing. Its core idea is to replace the central pixel value with the average value of the pixels in the neighborhood, thereby suppressing noise and blurring details. Specifically: Define a filter kernel, traverse the image after gray histogram equalization pixel by pixel, calculate the average value of all pixels in the window, and assign this average value to the central pixel. For the central pixel , its filtered value is:

[0041] Among them, ; Obtain the enhanced image after mean filtering; Step 3.3, determine the yarn tilt angle, and calculate the stitch pixel number according to the yarn tilt angle; Among them, determining the yarn tilt angle specifically is: Pre-collect a group of normal images without broken yarn, a total of M images. Then, after intercepting the images, performing histogram equalization, and mean filtering on each image according to the methods in Steps 3.1 - 3.2, determine an angle range ( ) according to prior knowledge, and respectively use the gray integral projection algorithm to calculate the projection curves of each image at , , ,... ; For any angle , , obtain the projection curves of M images at the angle , that is, obtain M projection curves at any angle . Then calculate the average value of the peaks of the M projection curves at any angle , and respectively obtain the corresponding average values at the angles , , ,... , a total of . Then, take the angle corresponding to the largest average value among the average values as the yarn tilt angle; Among them, for an image n rowsm The image of the column, I ( i , j ) is the gray value of the image at the position ( i , j ). Then the projection curve of the image at any angle θ is as follows: ; Specifically, calculating the stitch pitch pixels according to the yarn tilt angle is as follows: According to the obtained yarn tilt angle, use the gray integral projection algorithm to calculate the projection curve of each image at the yarn tilt angle, obtain M projection curves, calculate the distance between adjacent two peaks in each projection curve. Assuming that there are b adjacent peaks in the corresponding projection curve, then b distances are obtained. Then calculate the average value of the b distances corresponding to each projection curve to obtain M average values. Then average the M average values to obtain the value, which is the stitch pitch pixels; Step 3.4, according to the yarn tilt angle determined in step 3.3, use the gray integral projection algorithm to calculate the corresponding projection curve at the yarn tilt angle for the image after image enhancement in step 3.2. Each peak in the projection curve represents a yarn. Then calculate the distance between adjacent peaks in the projection curve. When the distance between two adjacent peaks in the projection curve is greater than or equal to the stitch pitch pixels in step 3.3, it means that the distance between the peaks is too large, exceeding the normal distance between two yarns, that is, there is no yarn between two needles, that is, there is a broken yarn; otherwise, it means that the distance between the yarns is normal and there is no broken yarn.

[0042] Embodiment 7 On the basis of Embodiment 6, the position sensor senses the left and right positions of the comb 2 through the photoelectric switch 42. Then step 1 is specifically as follows: After the double needle bed warp knitting machine works, the computer controls the driving motor 10 to start. The slide 8 moves left and right on the guide rail body of the slide guide rail 7. Since weaving is a periodic operation process, the yarn states in the knitting areas 11 of the front and rear needle beds are different at different times. The photoelectric switch 42 senses the left and right positions of the comb 2. When all the yarns are clearly presented in the knitting areas 11 of the front and rear needle beds, the photoelectric switch 42 is triggered, and the computer controls the camera 93 to take pictures.

[0043] Embodiment 8 On the basis of Embodiment 6, when using the method of the present invention for judgment for the first time, it is necessary to determine the yarn tilt angle and the stitch pitch pixels according to the image without broken yarn according to the steps of step 3.3, and save them in the computer after obtaining. In the subsequent judgment process, step 3.3 does not need to be carried out, and the saved data can be directly used. If a new product variety is replaced, it is necessary to obtain the corresponding yarn tilt angle and stitch pitch pixels again according to the method of step 3.

[0044] Example 9 On the basis of Example 8, in step 3.3, a group of M normal images without broken yarn are pre - collected, where M≥50.

[0045] Example 10 On the basis of Example 9, the slide rail of the present invention adopts a synchronous belt linear slide module.

Claims

1. A double needle bed warp knitting machine yarn breakage detection device based on machine vision, characterized in that: The invention comprises a crossbeam (6) arranged on both sides of the front and rear of a comb bar (2) of a double needle bed warp knitting machine, a slide rail (7) being arranged on the crossbeam (6), one end of a guide rail body of the slide rail (7) being connected to a drive motor (10), the drive motor (10) driving a slide (8) of the slide rail (7) to move left and right on its guide rail body, an image acquisition module (9) being fixed on the slide (8), and a position sensor being further provided, the position sensor being used for: when all yarns are clearly presented in the front and rear needle bed knitting areas (11), the position sensor is triggered, the image acquisition module (9) is connected to a computer via a wire, the position sensor, the drive motor (10) and the computer are all connected to a controller via a wire, and the controller is also connected to a control box of the double needle bed warp knitting machine via a wire.

2. The double needle bed warp knitting machine yarn breakage detection device based on machine vision according to claim 1, characterized in that: The installation position of the position sensor is: when the comb (2) swings back and forth and moves left and right to a position where all yarns are clearly present in the front and rear needle bed knitting areas (11), the position sensor identifies the position of the comb (2) at this time; or when the needle bed swings back and forth to a position where all yarns are clearly present in the front and rear needle bed knitting areas (11), the position sensor identifies the position of the needle bed at this time.

3. The double needle bed warp knitting machine yarn breakage detection device based on machine vision according to claim 2, characterized in that: The position sensor adopts one of a photoelectric switch, an inductive proximity switch, a Hall switch, a pull-wire position switch, and an angular displacement sensor.

4. The double needle bed warp knitting machine yarn breakage detection device based on machine vision according to claim 3, characterized in that: The position sensor adopts a photoelectric switch module (4), the photoelectric switch module (4) comprising a fixed mounting bracket (41) mounted on a comb fixing bracket (3), a photoelectric switch (42) being mounted on the fixed mounting bracket (41), and the photoelectric switch (42) being connected to the controller.

5. The double needle bed warp knitting machine yarn breakage detection device based on machine vision according to claim 4, characterized in that: A connecting bracket (5) is fixedly provided at the left and right ends of a double needle bed warp knitting machine frame (1), the two connecting brackets (5) are respectively located at the front and rear sides of a comb bar (2), the two cross beams (6) are respectively fixed at the other ends of the corresponding connecting brackets (5), the two cross beams (6) are located at the front and rear sides of the comb bar (2) and extend from the leftmost end to the rightmost end of the comb bar (2), one or two slides (8) are slidably provided on the guide rail body of each slide rail (7), and the drive motor (10) drives the slide (8) to move on the guide rail body of the slide rail (7), so that the image acquisition module (9) can fully cover the yarn in the front and rear needle bed knitting areas (11) by taking pictures.

6. The double needle bed warp knitting machine yarn breakage detection device based on machine vision according to claim 5, characterized in that: The image acquisition module (9) comprises a mounting bracket (95) fixed on the slide (8), a camera (93) facing the yarn in the front and rear needle bed knitting areas (11) is fixed on the mounting bracket (95), and the camera (93) is electrically connected to the computer via a wire; A light source (91) and a fan (92) are also fixed to the mounting bracket (95); the light source (91) is used to illuminate the yarns in the front and rear needle bed knitting areas (11) to provide a bright environment for the camera (92) to take pictures; the fan (92) is used to cool the light source (91); and both the light source (91) and the fan (92) are electrically connected to the controller via wires.

7. A method for detecting yarn breakage in a double needle bed warp knitting machine based on machine vision, characterized in that: The double needle bed warp knitting machine yarn breakage detection device based on machine vision as claimed in claim 6 is implemented specifically according to the following steps: Step 1, after the double needle bed warp knitting machine is in operation, the controller controls the driving motor (10) to start, and the slide (8) moves left and right on the guide rail body of the slide rail (7). When all the yarns are clearly presented in the front and rear needle bed knitting areas (11), the position sensor is triggered, and the controller controls the camera (93) to take a picture; Step 2, the camera (93) transmits the collected yarn image of the weaving area to the computer in real time; Step 3, the computer determines whether there is yarn breakage based on the collected yarn image; Step 4: If yarn breakage is detected, the computer sends a stop signal to the controller, and the controller controls the double needle bed warp knitting machine to stop running, otherwise it continues to run normally.

8. The method for detecting yarn breakage in a double needle bed warp knitting machine based on machine vision according to claim 7, characterized in that: The step 3 is specifically as follows: Step 3.1, the computer receives the yarn image and then uses the image capture algorithm to capture the yarn image of the weaving area, captures the yarn image of the weaving area, retains the complete yarn, and removes useless image information; Step 3.2, the image captured in step 3.1 is enhanced by using histogram equalization and mean filtering; Among them, histogram equalization is to perform grayscale histogram equalization processing on the image intercepted in step 3.1 to calculate the final grayscale level, specifically: Among them, Round to the nearest integer, which is the final grayscale. is the mapping function, Represents grayscale levels from 0 to The cumulative probability of k is the original gray level of the image. Within the range, The original gray level i Probability of occurrence: in, is the total number of pixels in the image, is the number of pixels with gray level i; Then the image after grayscale histogram equalization is processed by mean filtering, specifically: Define a The filter kernel traverses the image after grayscale histogram equalization pixel by pixel, calculates the average value of all pixels in the window, and assigns the average value to the center pixel. , its filtered value for: in, ; Get the enhanced image after mean filtering; Step 3.3, determining the yarn inclination angle, and calculating the number of stitch length pixels according to the yarn inclination angle; Step 3.4, according to the yarn inclination angle determined in step 3.3, the grayscale integral projection algorithm is used to calculate the corresponding projection curve under the yarn inclination angle for the image after image enhancement in step 3.

2. Each peak in the projection curve represents a yarn, and then the distance between adjacent peaks in the projection curve is calculated. When the distance between two adjacent peaks in the projection curve is greater than or equal to the number of needle spacing pixels in step 3.3, it means that the distance between the peaks is too large, exceeding the normal distance between two yarns, that is, there is no yarn between the two needles, that is, there is a yarn break; otherwise, it means that the distance between the yarns is normal and there is no yarn break.

9. The method for detecting yarn breakage in a double needle bed warp knitting machine based on machine vision according to claim 8, characterized in that: The yarn inclination angle is determined in step 3.3 as follows: A set of normal images without yarn breakage is collected in advance, a total of M images, and then each image is subjected to image interception, histogram equalization, and mean filtering according to the methods in steps 3.1-3.2, and an angle range is determined according to prior knowledge ( ), respectively, the grayscale integral projection algorithm is used to calculate each image in , , ,,, The projection curve below: For any angle , , get M images at angles The projection curve is at any angle. Get M projection curves, and then calculate any angle The average value of the peak values ​​of the next M projection curves is obtained at the angle , , ,,, The corresponding average values ​​are Then, The angle corresponding to the largest average value among the average values ​​is taken as the yarn inclination angle; The specific calculation of the number of pixels of stitch length according to the yarn inclination angle is: According to the acquired yarn inclination angle, the grayscale integral projection algorithm is used to calculate the projection curve of each image at the yarn inclination angle to obtain M projection curves, and the distance between two adjacent peaks in each projection curve is calculated. Assuming that the corresponding projection curve has b adjacent peaks, b distances are obtained, and then the average value of the b distances corresponding to each projection curve is calculated to obtain M average values, and then the M average values ​​are averaged to obtain the value, which is the number of stitch pixels.

10. The method for detecting yarn breakage in a double needle bed warp knitting machine based on machine vision according to claim 9, characterized in that: For a n OK m Column images, I ( i , j ) is the image at position ( i , j ), the image is at any angle θ The projection curve below is as follows: 。

Images