A towel loom automatic control method and system based on computer vision

Abstract

The present invention discloses a towel loom automatic control method and system based on computer vision, and relates to the field of towel loom control technology. The method comprises the following steps: installing a counter and a yarn sensor inside the towel loom to respectively obtain the actual motor speed and the yarn feeding amount as analysis data sets, and using the analysis data sets to judge the relationship between the motor speed and the yarn feeding amount; using the judgment result to analyze the transmission noise value generated by different motor speeds and yarn feeding amounts, and setting the motor speed value corresponding to the minimum noise value as the initial control value; using an infrared imager to obtain an image of the towel under the initial control value condition; using computer vision technology to perform feature detection operations on the image to be detected, and performing a similarity comparison with a preset towel standard to judge whether the towel has defects; if there are defects, adjusting the initial control value. The present invention minimizes the transmission noise by optimizing the motor speed, which may reduce the wear on the machine while ensuring production efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of towel loom control, and in particular to an automatic control method and system for a towel loom based on computer vision. Background Art

[0002] A towel loom is a textile machine used to produce towels. Through a specific weaving method, the yarn forms a special terry structure, thereby producing towels with good water absorbency and soft touch. The towel loom is mainly composed of a frame, a weaving system, a drive system, a tension control device, a yarn feeding system, a control system, etc. These components work together to ensure that the towel loom can efficiently and stably produce high-quality towel products.

[0003] When the towel loom is in use, its drive control system is a key component to ensure the efficient and precise operation of the machine. The drive system consists of a motor and a transmission device (such as belts, gears, etc.), which is responsible for driving the operation of the loom. Through the integrated drive control system, the towel loom can achieve high efficiency and high quality production while reducing failure rate and maintenance costs.

[0004] However, the automatic control system in the prior art cannot reduce the noise generated by the motor drive when in use, which may easily lead to a situation where the transmission noise caused by the motor vibration and yarn feeding friction is relatively large after the motor speed requirement, thereby reducing the comfort level within the production range. At the same time, after the control command is set, it is impossible to determine whether the finished product meets the preset standard value, resulting in a decrease in the overall quality of the product. Summary of the invention

[0005] In order to solve the above problems, the present invention proposes a towel loom automatic control method and system based on computer vision, which realizes the purpose of automatic control monitoring and adjustment of motor speed and yarn feeding amount.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] In a first aspect, the present invention provides a towel loom automatic control method based on computer vision, the towel loom automatic control method based on computer vision comprises the following steps:

[0008] S1. Install a counter and a yarn sensor inside the towel loom to obtain the actual motor speed and the yarn feeding amount as analysis data sets, and use the analysis data sets to determine the relationship between the motor speed and the yarn feeding amount;

[0009] S2, using the judgment result to analyze the transmission noise values ​​generated by different motor speeds and yarn feeding amounts, and setting the motor speed value corresponding to the minimum noise value as the initial control value of the towel loom;

[0010] S3, using an infrared imager to obtain an image of the towel made under the initial control value condition and preprocessing it to obtain an image of the towel to be detected;

[0011] S4, using computer vision technology to perform feature detection operations on the image to be detected, and performing similarity comparison with a preset towel standard to determine whether the towel has defects;

[0012] S5. If there is a defect, return to step S2 to adjust the initial control value. If there is no defect, continue to execute according to the initial control value.

[0013] Preferably, using the judgment result to analyze the transmission noise values ​​generated by different motor speeds and yarn feeding amounts, and setting the motor speed value corresponding to the minimum noise value as the initial control value of the towel loom includes the following steps:

[0014] S21, obtaining a direct proportional relationship between the motor speed and the yarn feeding amount, setting motor speed values ​​under different values, and solving the actual yarn feeding amount by using the direct proportional relationship;

[0015] S22, establishing a physical model including the motor, yarn feeding components and sound level meter functions, simulating the transmission noise value generated by the motor vibration and yarn feeding friction;

[0016] S23, inputting different motor speed values ​​and yarn feeding amounts into the physical model to output transmission noise values, and collecting all transmission noise values ​​to form a judgment database;

[0017] S24, using an optimization algorithm to select the minimum noise value and the motor speed value corresponding to it, and setting the motor speed value as the initial control value of the motor in the towel loom.

[0018] Preferably, establishing a physical model including the motor, the yarn feeding component and the sound level meter function, and simulating the transmission noise value generated by the motor vibration and the yarn feeding friction comprises the following steps:

[0019] S221, sampling the motion instances of the motor and the yarn feeding component, simulating the motion trajectory of the towel loom in the production process, and representing the towel loom by using the motion trajectory and edge marking method;

[0020] S222, selecting a grid area as the initial point of the physical model, and dividing the grid area into three groups of medium, high and low using the motion trajectory;

[0021] S223, using an adaptive partitioning algorithm to divide the three groups of areas, constructing a geometric model corresponding to the grid area, and defining the sound level meter position and noise coefficient based on the geometric model to obtain a physical model;

[0022] S224, using a physical model to simulate the vibration of the motor when it is working and the friction force generated by the guide component and the tension control component of the yarn feeding component during the yarn feeding process;

[0023] S225. Use a sound level meter to measure the noise value generated by the combination of vibration and friction to determine the propagation path and intensity of the noise.

[0024] Preferably, using an adaptive partitioning algorithm to divide the three groups of areas, constructing a geometric model corresponding to the grid area, and defining the sound level meter position and the noise coefficient based on the geometric model to obtain a physical model includes the following steps:

[0025] S2231, marking the vertices, edges and faces of the three groups of regions as fixed points, fixed edges and fixed faces, and setting the three groups of fixed points, fixed edges and fixed faces as triangle vertices, triangle faces and triangle edges respectively;

[0026] S2232, traversing the triangle vertices and assigning the sampled motion trajectory points to the corresponding vertices, and updating the activities of the fixed points in the three groups of regions according to the relationship between the average trajectory mobility of the fixed points and the preset value;

[0027] S2233, traverse the triangular face to obtain the number of fixed points and edge markings contained in the triangular face, update the activity of the triangular face and the triangular edge, and use the partitioning method to update the position of the fixed point and the corresponding edge point;

[0028] S2234, setting the activity value of the fixed point to the average value of the activity value of the corresponding edge point, and subtracting the preset value from the activity value of the fixed point to determine the relationship between the difference and the set value, and repeating steps S2232 to S2234 if the activity value is less than the set value;

[0029] S2235, if the activity value is less than the set value, the activity value is set to a preset value, and a grid region structure is generated according to the preset value to generate a geometric model corresponding to the grid region;

[0030] S2236. Introduce the sound level meter and noise coefficient into the geometric model and define them in combination to obtain a complete physical model.

[0031] Preferably, selecting the minimum noise value and the motor speed value corresponding to it by using an optimization algorithm, and setting the motor speed value as the initial control value of the motor in the towel loom comprises the following steps:

[0032] S241, dividing the search area for the noise value, and using the confidence search method to determine the optimal search area boundary to achieve a global area containing the noise value and the optimal value;

[0033] S242, constructing a steering model optimization search path adapted to the boundary characteristics of the best search area, and taking the minimum noise value and its corresponding motor speed value as the optimization target;

[0034] S243, using an optimization algorithm to optimize the steering model to obtain an optimal search path, and selecting a minimum noise value and a motor speed value;

[0035] S244, writing the motor speed value result into the towel loom control program and setting it as the motor initial control value, so that the corresponding pulse frequency outputted by it controls the motor.

[0036] Preferably, the optimization algorithm is used to optimize the steering model to obtain the best search path, and the selection of the minimum noise value and the motor speed value includes the following steps:

[0037] S2431, setting the search scale, search path length, maximum number of iterations, search line spacing range and spacing step according to the steering model;

[0038] S2432, using an optimization algorithm to generate initial values ​​of the traversal order of the initial row and the operating row, and determining the best traversal order under the condition of determining the starting row according to the search line spacing;

[0039] S2433, determining the total search path and the search probability according to the optimal traversal order, and repeatedly iterating within the maximum number of iterations using a non-dominated sorting method until the set search scale is reached to obtain the optimal total path;

[0040] S2434, changing the search line spacing range according to the search spacing step, repeating steps S2432 and S2434 until the search line spacing range upper limit is reached;

[0041] S2435. Use non-dominated sorting to obtain the minimum search path length under different search line spacings to obtain the optimal search path, and output the corresponding minimum noise value and motor speed value according to the optimal search path.

[0042] Preferably, using computer vision technology to perform feature detection on the image to be detected, and performing similarity comparison with a preset towel standard to determine whether the towel has defects includes the following steps:

[0043] S41, performing an extreme value detection operation on the image to be detected in the scale space, locating the feature key points of the image to determine the image gradient direction to form feature description points;

[0044] S42, using Euclidean distance to match feature description points in pairs to form matching blocks, removing redundant matching blocks through morphology, extracting feature points in the matching blocks and performing component analysis operations;

[0045] S43, purifying the matched feature points, and locating the feature positions in the image to be detected using the distribution of the feature points;

[0046] S44, obtaining a preset towel standard and performing a similarity comparison with a feature position of the image to be detected, and judging whether the towel has defects according to the similarity result.

[0047] Preferably, obtaining a preset towel standard and comparing the feature positions of the image to be detected for similarity, and judging whether the towel has defects according to the similarity result includes the following steps:

[0048] S441, defining a set of ellipses centered on the feature position on the image to be detected and the standard towel image respectively p With ellipse q , initialize the ellipse p The boundary of the ellipse p The coordinate position of the pole;

[0049] S442, traverse the rows between the poles according to the coordinate positions of the poles and calculate the rows and ellipses p The coordinates of the two intersection points of the boundary, which belong to the row and are located in the ellipse p All points within the boundary;

[0050] S443, Initialize the ellipse q The boundary of the ellipse q The pole coordinate position, and according to the pole coordinate position, traverse the rows between the poles and calculate the row and the ellipse q The coordinates of the two intersection points of the boundary, which belong to the row and are located in the ellipse q All points within the boundary;

[0051] S3444, using affine transformation to transform the ellipse p With ellipse q All points within the boundary are normalized to the same number of position points;

[0052] S3445, using an affine invariant block similarity algorithm to respectively calculate the similarity between the two ellipse position points, and adding the similarity values ​​of all points to determine whether the towel has a defect;

[0053] S3446. If the similarity value is greater than the preset value, it indicates that the towel has defects. If the similarity value is less than the preset value, it indicates that the towel has no defects.

[0054] Preferably, the similarity algorithm is expressed as:

[0055] ;

[0056] In the formula, T Represents the similarity result, Represents an ellipse p The center point of D u Indicates uThe structure tensor of the position points, E u Indicates u The affine transformation result of the position points is Represents an ellipse q The center point of u Indicates the number of location points, Indicates the first k The variance of the image points is t Gaussian function, represents the image point on the image to be detected, k express Any point on .

[0057] In a second aspect, the present invention further provides a towel loom automatic control system based on computer vision, the towel loom automatic control system based on computer vision comprises a data collection and analysis module, a loom control setting module, a towel image processing module, a defect detection and judgment module and an adjustment and optimization execution module;

[0058] Among them, the data collection and analysis module, the loom control setting module, the towel image processing module, the defect detection and judgment module and the adjustment and optimization execution module are connected in sequence;

[0059] A data collection and analysis module is used to install a counter and a yarn sensor inside the towel loom to obtain the actual motor speed and the yarn feeding amount as analysis data sets, and use the analysis data sets to determine the relationship between the motor speed and the yarn feeding amount;

[0060] A loom control setting module is used to analyze the transmission noise values ​​generated by different motor speeds and yarn feeding amounts using the judgment results, and set the motor speed value corresponding to the minimum noise value as the initial control value of the towel loom;

[0061] A towel image processing module, used to obtain a towel image made under the initial control value condition by using an infrared imager and perform preprocessing to obtain an image of the towel to be detected;

[0062] A defect detection and judgment module is used to perform feature detection operations on the image to be detected using computer vision technology, and to perform a similarity comparison with a preset towel standard to determine whether the towel has defects;

[0063] The adjustment and optimization execution module is used to return to the loom control setting module to adjust the initial control value if there is a defect, and continue to execute according to the initial control value if there is no defect.

[0064] The beneficial effects of the present invention are:

[0065] 1. The present invention minimizes transmission noise by optimizing the motor speed, thereby reducing wear on the machine while ensuring production efficiency, and uses computer vision technology for feature detection and defect judgment to more accurately identify product defects and improve the overall quality of the product. It uses automated control to monitor and adjust the motor speed and yarn feed amount, reduce manual intervention, and improve the continuity and efficiency of the production process.

[0066] 2. The present invention establishes a direct proportional relationship between the motor speed and the yarn feeding amount, so that the yarn feeding amount can be accurately controlled, and the transmission noise under different operating conditions is simulated. The minimum noise value and its corresponding motor speed are selected by the optimization algorithm, which can ensure operation at the lowest noise level and improve the working environment. At the same time, reducing vibration and friction can reduce equipment wear and extend the service life of the equipment.

[0067] 3. The present invention utilizes computer vision technology to perform feature detection and compares it with preset standards to ensure the quality and production efficiency of finished towels, and locates the image through the distribution of feature points of the image to be detected, thereby ensuring that the key areas of detection are associated with the key quality standards of the towels, thereby improving the accuracy of defect detection, and quickly determining whether the towels meet the preset quality standards through similarity comparison, thereby achieving efficient quality control. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0069] Figure 1 is a flow chart of a towel loom automatic control method based on computer vision according to an embodiment of the present invention;

[0070] Figure 2 The present invention is a block diagram of a computer vision-based automatic control system for a towel loom according to an embodiment of the present invention.

[0071] In the figure:

[0072] 1. Data collection and analysis module; 2. Loom control setting module; 3. Towel image processing module; 4. Defect detection and judgment module; 5. Adjustment and optimization execution module. DETAILED DESCRIPTION

[0073] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0074] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used in the present invention have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0075] It should be noted that the terms used herein are only for describing specific embodiments, and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that the terms "include" and "have" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0076] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0077] See also Figure 1 The present invention provides a towel loom automatic control method based on computer vision, and the towel loom automatic control method based on computer vision comprises the following steps:

[0078] S1. A counter and a yarn sensor are installed inside the towel loom to obtain the actual motor speed and the yarn feeding amount as analysis data sets, and the relationship between the motor speed and the yarn feeding amount is determined by using the analysis data sets.

[0079] In this embodiment, a counter and a yarn sensor are installed inside the towel loom to respectively obtain the actual motor speed and the yarn feeding amount as analysis data sets, and the relationship between the motor speed and the yarn feeding amount is determined by using the analysis data sets, including the following steps:

[0080] A counter is installed on the motor to measure and record the motor speed in real time, and a yarn sensor is installed on the yarn feeding system to monitor and record the yarn feeding amount. The loom is started, and the counter and yarn sensor are allowed to record the motor speed and yarn feeding amount data in a time series manner.

[0081] The collected data is transmitted to the data processor, the collected data is preprocessed, and the relationship between the motor speed and the yarn feeding amount is analyzed using a statistical learning method. The analysis results are interpreted to analyze the relationship between the motor speed and the yarn feeding amount, and the linear correlation between the two is determined.

[0082] S2. Analyze the transmission noise values ​​generated by different motor speeds and yarn feeding amounts using the judgment result, and set the motor speed value corresponding to the minimum noise value as the initial control value of the towel loom.

[0083] In this embodiment, the following steps are included to analyze the transmission noise values ​​generated by different motor speeds and yarn feeding amounts using the judgment results, and to set the motor speed value corresponding to the minimum noise value as the initial control value of the towel loom:

[0084] S21, obtaining the direct proportional relationship between the motor speed and the yarn feeding amount, setting the motor speed value under different values, and solving the actual yarn feeding amount by using the direct proportional relationship.

[0085] Specifically, obtaining the direct proportional relationship between the motor speed and the yarn feeding amount, setting the motor speed value under different values, and using the direct proportional relationship to solve the actual yarn feeding amount includes the following steps:

[0086] According to the relationship between the motor speed and the yarn feeding amount, a linear equation model is used to determine the direct proportional relationship between the two, the motor speed value is set within a fixed range, and the actual yarn feeding amount corresponding to the motor speed value is obtained using the direct proportional relationship.

[0087] S22. Establish a physical model including the motor, yarn feeding components and sound level meter functions to simulate the transmission noise value generated by the motor vibration and yarn feeding friction.

[0088] Specifically, establishing a physical model including the motor, yarn feeding components and sound level meter functions, and simulating the transmission noise value generated by the motor vibration and yarn feeding friction includes the following steps:

[0089] S221, sampling the motion instances of the motor and the yarn feeding component, simulating the motion trajectory of the towel loom in the production process, and representing the towel loom by using the motion trajectory and edge marking method;

[0090] S222, selecting a grid area as the initial point of the physical model, and dividing the grid area into three groups of medium, high and low using the motion trajectory;

[0091] S223, using an adaptive partitioning algorithm to divide the three groups of areas, constructing a geometric model corresponding to the grid area, and defining the sound level meter position and noise coefficient based on the geometric model to obtain a physical model;

[0092] Among them, the three groups of areas are divided by using an adaptive partitioning algorithm, a geometric model corresponding to the grid area is constructed, and the sound level meter position and noise coefficient are defined based on the geometric model to obtain a physical model, which includes the following steps:

[0093] S2231, marking the vertices, edges and faces of the three groups of regions as fixed points, fixed edges and fixed faces, and setting the three groups of fixed points, fixed edges and fixed faces as triangle vertices, triangle faces and triangle edges respectively;

[0094] S2232, traversing the triangle vertices and assigning the sampled motion trajectory points to the corresponding vertices, and updating the activities of the fixed points in the three groups of regions according to the relationship between the average trajectory mobility of the fixed points and the preset value;

[0095] S2233, traverse the triangular face to obtain the number of fixed points and edge markings contained in the triangular face, update the activity of the triangular face and the triangular edge, and use the partitioning method to update the position of the fixed point and the corresponding edge point;

[0096] S2234, setting the activity value of the fixed point to the average value of the activity value of the corresponding edge point, and subtracting the preset value from the activity value of the fixed point to determine the relationship between the difference and the set value, and repeating steps S2232 to S2234 if the activity value is less than the set value;

[0097] S2235, if the activity value is less than the set value, the activity value is set to a preset value, and a grid region structure is generated according to the preset value to generate a geometric model corresponding to the grid region;

[0098] S2236. Introduce the sound level meter and noise coefficient into the geometric model and define them in combination to obtain a complete physical model.

[0099] S224, using a physical model to simulate the vibration of the motor when it is working and the friction force generated by the guide component and the tension control component of the yarn feeding component during the yarn feeding process;

[0100] S225. Use a sound level meter to measure the noise value generated by the combination of vibration and friction to determine the propagation path and intensity of the noise.

[0101] S23, inputting different motor speed values ​​and yarn feeding amounts into the physical model to output transmission noise values, and collecting all transmission noise values ​​to form a judgment database.

[0102] Specifically, inputting different motor speed values ​​and yarn feeding amounts into the physical model to output transmission noise values, and collecting all transmission noise values ​​to form a judgment database includes the following steps:

[0103] The range and distribution of motor speed and yarn feed amount are determined and input into the physical model to simulate the towel loom under different motor speeds and yarn feed amounts.

[0104] The sound level meter in the physical model is used to obtain the noise value generated by the motor vibration and the yarn feeding friction, and all the preset simulation values ​​are traversed until all are input into the physical model to obtain the noise value.

[0105] All noise values ​​are collected and integrated to form a noise value judgment database to facilitate the subsequent minimum value optimization operation.

[0106] S24, using an optimization algorithm to select the minimum noise value and the motor speed value corresponding to it, and setting the motor speed value as the initial control value of the motor in the towel loom.

[0107] Specifically, the optimization algorithm is used to select the minimum noise value and the motor speed value corresponding to it, and the motor speed value is set as the initial control value of the motor in the towel loom, which includes the following steps:

[0108] S241, dividing the search area for the noise value, and using the confidence search method to determine the optimal search area boundary to achieve a global area containing the noise value and the optimal value;

[0109] S242, constructing a steering model optimization search path adapted to the boundary characteristics of the best search area, and taking the minimum noise value and its corresponding motor speed value as the optimization target;

[0110] S243, using an optimization algorithm to optimize the steering model to obtain an optimal search path, and selecting a minimum noise value and a motor speed value;

[0111] Among them, the optimization algorithm is used to optimize the steering model to obtain the best search path, and the selection of the minimum noise value and the motor speed value includes the following steps:

[0112] S2431, setting the search scale, search path length, maximum number of iterations, search line spacing range and spacing step according to the steering model;

[0113] S2432, using an optimization algorithm to generate initial values ​​of the traversal order of the initial row and the operating row, and determining the best traversal order under the condition of determining the starting row according to the search line spacing;

[0114] S2433, determining the total search path and the search probability according to the optimal traversal order, and repeatedly iterating within the maximum number of iterations using a non-dominated sorting method until the set search scale is reached to obtain the optimal total path;

[0115] S2434, changing the search line spacing range according to the search spacing step, repeating steps S2432 and S2434 until the search line spacing range upper limit is reached;

[0116] S2435. Use non-dominated sorting to obtain the minimum search path length under different search line spacings to obtain the optimal search path, and output the corresponding minimum noise value and motor speed value according to the optimal search path.

[0117] S244, writing the motor speed value result into the towel loom control program and setting it as the motor initial control value, so that the corresponding pulse frequency outputted by it controls the motor.

[0118] Therefore, by establishing a direct proportional relationship between the motor speed and the yarn feeding amount, the yarn feeding amount can be accurately controlled, and the transmission noise under different operating conditions is simulated. The minimum noise value and its corresponding motor speed are selected by the optimization algorithm to ensure operation at the lowest noise level.

[0119] S3. Using an infrared imager to obtain an image of the towel made under the initial control value condition and preprocessing it to obtain an image of the towel to be tested.

[0120] In this embodiment, the infrared imager is used to obtain the towel image made under the initial control value condition and preprocess it to obtain the towel image to be detected, which includes the following steps:

[0121] Adjust the parameters of the infrared imager according to the characteristics of the towel and the detection requirements, including resolution, focal length, and detection range, to ensure that the imager is accurately calibrated.

[0122] The towel loom is modified according to the initial control value, and the loom is operated to manufacture towels under the set initial control value. After the manufacturing is completed, the towels are photographed using an infrared imager.

[0123] The collected infrared images are preprocessed by denoising, contrast enhancement, and grayscale adjustment operations, and it is determined whether the preprocessed images are clear. If not, the preprocessing is repeated.

[0124] The pre-processed images are stored in appropriate formats and directories, and each image is marked, including the shooting time and control value parameters for easy recording.

[0125] S4. Use computer vision technology to perform feature detection operations on the image to be detected, and compare the similarity with the preset towel standard to determine whether the towel has defects.

[0126] In this embodiment, the following steps are used to perform feature detection on the image to be detected using computer vision technology and to compare the similarity with the preset towel standard to determine whether the towel has defects:

[0127] S41. Performing an extreme value detection operation on the image to be detected in the scale space, locating the feature key points of the image, determining the image gradient direction to form feature description points.

[0128] Specifically, performing an extreme value detection operation on the image to be detected in the scale space, locating the feature key points of the image and determining the image gradient direction to form feature description points includes the following steps:

[0129] The original image is processed by Gaussian blur to generate images of different scales to form a scale space. Images of different scales represent different degrees of blur.

[0130] The difference between adjacent scale images is determined in the scale space to form a differential scale space, and local maximum and minimum points are searched in the differential scale space.

[0131] The detected extreme points are precisely located, the positions of the extreme points are assigned based on the direction of the local gradient of the image, and based on the neighborhood information of each key point, a feature descriptor is generated to describe the position and scale information of the key point.

[0132] S42, using Euclidean distance to perform pairwise matching of feature description points to form matching blocks, and removing redundant matching blocks through morphology, extracting feature points within the matching blocks and performing component analysis operations.

[0133] Specifically, the feature description points are matched in pairs using the Euclidean distance to form matching blocks, and redundant matching blocks are removed by morphology, and the feature points in the matching blocks are extracted to perform component analysis operations, including the following steps:

[0134] Using Euclidean distance as the similarity metric, the nearest neighbor description point is found for each feature description point in the target image, the matching feature point pairs are organized into matching blocks, and the matching blocks are subjected to morphological analysis.

[0135] The matching blocks are further screened and verified to ensure the geometric consistency between the matching blocks, and representative feature points are extracted from the matching blocks based on the distribution and attributes of the feature points in the matching blocks.

[0136] S43, purifying the matched feature points, and locating the feature positions in the image to be detected using the distribution of the feature points.

[0137] Specifically, purifying the matched feature points and locating the feature positions in the image to be detected by using the distribution of the feature points includes the following steps:

[0138] The most representative and distinctive points are selected from the matching feature points using the geometric constraint test method and purified. The purified feature points are clustered to identify the main feature areas in the image.

[0139] Analyze and purify the distribution of feature points in the image, including the density and distribution pattern of feature points. Identify the key areas in the image by analyzing the spatial distribution of feature points. Determine the feature positions in the image to be detected based on the distribution and clustering results of feature points.

[0140] S44, obtaining a preset towel standard and performing a similarity comparison with a feature position of the image to be detected, and judging whether the towel has defects according to the similarity result.

[0141] Specifically, obtaining a preset towel standard and comparing the feature positions of the image to be detected for similarity, and judging whether the towel has defects according to the similarity result includes the following steps:

[0142] S441, defining a set of ellipses centered on the feature position on the image to be detected and the standard towel image respectively pWith ellipse q , initialize the ellipse p The boundary of the ellipse p The coordinate position of the pole;

[0143] S442, traverse the rows between the poles according to the coordinate positions of the poles and calculate the rows and ellipses p The coordinates of the two intersection points of the boundary, which belong to the row and are located in the ellipse p All points within the boundary;

[0144] S443, Initialize the ellipse q The boundary of the ellipse q The pole coordinate position, and according to the pole coordinate position, traverse the rows between the poles and calculate the row and the ellipse q The coordinates of the two intersection points of the boundary, which belong to the row and are located in the ellipse q All points within the boundary;

[0145] S3444, using affine transformation to transform the ellipse p With ellipse q All points within the boundary are normalized to the same number of position points;

[0146] S3445, using an affine invariant block similarity algorithm to respectively calculate the similarity between the two ellipse position points, and adding the similarity values ​​of all points to determine whether the towel has a defect;

[0147] S3446. If the similarity value is greater than the preset value, it indicates that the towel has defects. If the similarity value is less than the preset value, it indicates that the towel has no defects.

[0148] Among them, the expression of the similarity algorithm is:

[0149] ;

[0150] In the formula, T Represents the similarity result, Represents an ellipse p The center point of D u Indicates u The structure tensor of the position points, E u Indicates u The affine transformation result of the position points is Represents an ellipse q The center point of u Indicates the number of location points, Indicates the first k The variance of the image points is t Gaussian function, represents the image point on the image to be detected,k express Any point on .

[0151] Therefore, computer vision technology is used to perform feature detection and compare it with preset standards to ensure the quality and production efficiency of finished towels. The image is positioned through the distribution of feature points of the image to be detected, which can ensure that the key areas of detection are associated with the key quality standards of the towels, thereby improving the accuracy of defect detection. Through similarity comparison, it can be quickly determined whether the towels meet the preset quality standards, thereby achieving efficient quality control.

[0152] S5. If there is a defect, return to step S2 to adjust the initial control value. If there is no defect, continue to execute according to the initial control value.

[0153] Therefore, it is possible to ensure quality problems in the production process. When problems are found, the control value is returned to the initial setting and the motor speed is adjusted to improve product quality.

[0154] See also Figure 2 The present invention also provides a towel loom automatic control system based on computer vision, which includes a data collection and analysis module 1, a loom control setting module 2, a towel image processing module 3, a defect detection and judgment module 4 and an adjustment and optimization execution module 5;

[0155] Among them, the data collection and analysis module 1, the loom control setting module 2, the towel image processing module 3, the defect detection and judgment module 4 and the adjustment and optimization execution module 5 are connected in sequence;

[0156] The data collection and analysis module 1 is used to install a counter and a yarn sensor inside the towel loom to obtain the actual motor speed and the yarn feeding amount as analysis data sets, and use the analysis data sets to determine the relationship between the motor speed and the yarn feeding amount;

[0157] The loom control setting module 2 is used to analyze the transmission noise values ​​generated by different motor speeds and yarn feeding amounts using the judgment results, and set the motor speed value corresponding to the minimum noise value as the initial control value of the towel loom;

[0158] Towel image processing module 3, used to obtain the towel image made under the initial control value condition by using an infrared imager and perform preprocessing to obtain the towel image to be detected;

[0159] Defect detection and judgment module 4 is used to perform feature detection operation on the image to be detected by using computer vision technology, and to compare the similarity with the preset towel standard to determine whether the towel has defects;

[0160] The adjustment and optimization execution module 5 is used to return to the loom control setting module 2 to adjust the initial control value if there is a defect, and continue to execute according to the initial control value if there is no defect.

[0161] In summary, with the help of the above technical solutions of the present invention, the present invention minimizes the transmission noise by optimizing the motor speed, thereby being able to reduce the wear on the machine while ensuring production efficiency, and using computer vision technology for feature detection and defect judgment to more accurately identify product defects, improve the overall quality of the product, automatically control monitoring and adjust the motor speed and yarn feeding amount, reduce manual intervention, and improve the continuity and efficiency of the production process. The present invention establishes a direct proportional relationship between the motor speed and the yarn feeding amount, so that the yarn feeding amount can be accurately controlled, and the transmission noise under different operating conditions is simulated. The optimization algorithm is used to select the minimum noise value and its corresponding motor speed, which can ensure operation at the lowest noise level and improve the working environment. At the same time, reducing vibration and friction can reduce the wear of the equipment and extend the service life of the equipment. The present invention uses computer vision technology to perform feature detection and compare with preset standards to ensure the quality and production efficiency of the finished towel, and locates the image through the distribution of feature points of the image to be detected, which can ensure that the key area of ​​the detection is associated with the key quality standard of the towel, improve the accuracy of defect detection, and quickly judge whether the towel meets the preset quality standard through similarity comparison, thereby achieving efficient quality control.

[0162] Those skilled in the art will appreciate that the units, i.e., algorithm steps, of the various examples described in conjunction with this embodiment can be implemented in electronic hardware or in a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0163] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A towel loom automatic control method based on computer vision, characterized in that: The computer vision-based towel loom automatic control method comprises the following steps: S1. Install a counter and a yarn sensor inside the towel loom to obtain the actual motor speed and the yarn feeding amount as analysis data sets, and use the analysis data sets to determine the relationship between the motor speed and the yarn feeding amount; S2, using the judgment result to analyze the transmission noise values ​​generated by different motor speeds and yarn feeding amounts, and setting the motor speed value corresponding to the minimum noise value as the initial control value of the towel loom; S3, using an infrared imager to obtain an image of the towel made under the initial control value condition and preprocessing it to obtain an image of the towel to be detected; S4, using computer vision technology to perform feature detection operations on the image to be detected, and performing similarity comparison with a preset towel standard to determine whether the towel has defects; S5. If there is a defect, return to step S2 to adjust the initial control value. If there is no defect, continue to execute according to the initial control value; The method of analyzing the transmission noise values ​​generated by different motor speeds and yarn feeding amounts using the judgment results and setting the motor speed value corresponding to the minimum noise value as the initial control value of the towel loom comprises the following steps: S21, obtaining a direct proportional relationship between the motor speed and the yarn feeding amount, setting motor speed values ​​under different values, and solving the actual yarn feeding amount by using the direct proportional relationship; S22, establishing a physical model including the motor, yarn feeding components and sound level meter functions, simulating the transmission noise value generated by the motor vibration and yarn feeding friction; S23, inputting different motor speed values ​​and yarn feeding amounts into the physical model to output transmission noise values, and collecting all transmission noise values ​​to form a judgment database; S24, using an optimization algorithm to select the minimum noise value and the motor speed value corresponding to it, and setting the motor speed value as the initial control value of the motor in the towel loom; The physical model including the motor, the yarn feeding component and the sound level meter is established to simulate the transmission noise value generated by the motor vibration and the yarn feeding friction, and the following steps are included: S221, sampling the motion instances of the motor and the yarn feeding component, simulating the motion trajectory of the towel loom in the production process, and representing the towel loom by using the motion trajectory and edge marking method; S222, selecting a grid area as the initial point of the physical model, and dividing the grid area into three groups of medium, high and low using the motion trajectory; S223, using an adaptive partitioning algorithm to divide the three groups of areas, constructing a geometric model corresponding to the grid area, and defining the sound level meter position and noise coefficient based on the geometric model to obtain a physical model; S224, using a physical model to simulate the vibration of the motor when it is working and the friction force generated by the guide component and the tension control component of the yarn feeding component during the yarn feeding process; S225, using a sound level meter to evaluate the noise value generated by the combination of vibration and friction to determine the propagation path and intensity of the noise; The method of selecting the minimum noise value and the motor speed value corresponding to the minimum noise value by using the optimization algorithm and setting the motor speed value as the initial control value of the motor in the towel loom comprises the following steps: S241, dividing the search area for the noise value, and using the confidence search method to determine the optimal search area boundary to achieve a global area containing the noise value and the optimal value; S242, constructing a steering model optimization search path adapted to the boundary characteristics of the best search area, and taking the minimum noise value and its corresponding motor speed value as the optimization target; S243, using an optimization algorithm to optimize the steering model to obtain an optimal search path, and selecting a minimum noise value and a motor speed value; S244, writing the motor speed value result into the towel loom control program and setting it as the motor initial control value, so that the corresponding pulse frequency outputted by it controls the motor.

2. The computer vision-based automatic control method for a towel loom according to claim 1, characterized in that: The optimization algorithm is used to optimize the steering model to obtain the best search path, and the selection of the minimum noise value and the motor speed value includes the following steps: S2431, setting the search scale, search path length, maximum number of iterations, search line spacing range and spacing step according to the steering model; S2432, using an optimization algorithm to generate initial values ​​of the traversal order of the initial row and the operating row, and determining the best traversal order under the condition of determining the starting row according to the search line spacing; S2433, determining the total search path and the search probability according to the optimal traversal order, and repeatedly iterating within the maximum number of iterations using a non-dominated sorting method until the set search scale is reached to obtain the optimal total path; S2434, changing the search line spacing range according to the search spacing step, repeating steps S2432 and S2434 until the search line spacing range upper limit is reached; S2435. Use non-dominated sorting to obtain the minimum search path length under different search line spacings to obtain the optimal search path, and output the corresponding minimum noise value and motor speed value according to the optimal search path.

3. The computer vision-based automatic control method for a towel loom according to claim 1, characterized in that: The method of using computer vision technology to perform feature detection on the image to be detected and performing similarity comparison with a preset towel standard to determine whether the towel has defects includes the following steps: S41, performing an extreme value detection operation on the image to be detected in the scale space, locating the feature key points of the image to determine the image gradient direction to form feature description points; S42, using Euclidean distance to perform pairwise matching of feature description points to form matching blocks, and removing redundant matching blocks through morphology, extracting feature points in the matching blocks and performing component analysis operations; S43, purifying the matched feature points, and locating the feature positions in the image to be detected using the distribution of the feature points; S44, obtaining a preset towel standard and performing a similarity comparison with a feature position of the image to be detected, and judging whether the towel has defects according to the similarity result.

4. The computer vision-based automatic control method for a towel loom according to claim 3, characterized in that: The method of obtaining a preset towel standard and comparing the characteristic positions of the image to be detected for similarity, and judging whether the towel has defects according to the similarity result includes the following steps: S441, defining a set of ellipses centered on the feature position on the image to be detected and the standard towel image respectively p With ellipse q , initialize the ellipse p The boundary of the ellipse p The coordinate position of the pole; S442, traverse the rows between the poles according to the coordinate positions of the poles and calculate the rows and ellipses p The coordinates of the two intersection points of the boundary, which belong to the row and are located in the ellipse p All points within the boundary; S443, Initialize the ellipse q The boundary of the ellipse q The pole coordinate position, and according to the pole coordinate position, traverse the rows between the poles and calculate the row and the ellipse q The coordinates of the two intersection points of the boundary, which belong to the row and are located in the ellipse q All points within the boundary; S444, using affine transformation to transform the ellipse p With ellipse q All points within the boundary are normalized to the same number of position points; S445, using an affine invariant block similarity algorithm to respectively calculate the similarity between the two ellipse position points, and adding the similarity values ​​of all the points to determine whether the towel has a defect; S446. If the similarity value is greater than the preset value, it indicates that the towel has defects. If the similarity value is less than the preset value, it indicates that the towel has no defects.

5. The computer vision-based automatic control method for a towel loom according to claim 4, characterized in that: The expression of the similarity algorithm is: ； In the formula, T Represents the similarity result; Represents an ellipse p The center point of D u Indicates u The structure tensor of the position points; E u Indicates u Affine transformation result of position points; Represents an ellipse q The center point of u Indicates the number of location points; Indicates the first k The variance of the image points is t Gaussian function of Represents an image point on the image to be detected; k express Any point on .

6. A towel loom automatic control system based on computer vision, used to implement the towel loom automatic control method based on computer vision according to any one of claims 1 to 5, characterized in that: The towel loom automatic control system based on computer vision includes a data collection and analysis module, a loom control setting module, a towel image processing module, a defect detection and judgment module, and an adjustment and optimization execution module; Wherein, the data collection and analysis module, the loom control setting module, the towel image processing module, the defect detection and judgment module and the adjustment and optimization execution module are connected in sequence; The data collection and analysis module is used to install a counter and a yarn sensor inside the towel loom to respectively obtain the actual motor speed and the yarn feeding amount as analysis data sets, and use the analysis data sets to determine the relationship between the motor speed and the yarn feeding amount; The loom control setting module is used to analyze the transmission noise values ​​generated by different motor speeds and yarn feeding amounts using the judgment results, and set the motor speed value corresponding to the minimum noise value as the initial control value of the towel loom; The towel image processing module is used to obtain the towel image made under the initial control value condition by using the infrared imager and perform preprocessing to obtain the towel image to be detected; The defect detection and judgment module is used to perform feature detection operations on the image to be detected using computer vision technology, and to perform similarity comparison with a preset towel standard to determine whether the towel has defects; The adjustment and optimization execution module is used to return to the loom control setting module to adjust the initial control value if there is a defect, and continue to execute according to the initial control value if there is no defect.

Images