An online automatic fabric flower arranging method based on machine vision

Abstract

The present invention discloses an online automatic fabric pattern rectifying method based on machine vision, comprising the following steps: S1: Collecting a full-width image of the running fabric, and obtaining the fabric pattern area through image processing; S2: Performing operations on the fabric pattern area to obtain the width, height and number of the pattern repeat patterns in the fabric pattern area; S3: Setting a pattern template in the fabric pattern area, and using the pattern template to detect the pattern position of the fabric pattern area to obtain the position coordinates of the pattern; S4: According to the position coordinates of the pattern, obtaining the pattern deformation amount of the pattern; S5: According to the pattern deformation amount of the pattern, correcting the fabric pattern deformation through the weft straightening device of the weft straightening machine. The present invention can accurately calculate the pattern deformation amount for fabrics with pattern patterns, especially for fabrics such as lace with pattern repeat patterns or batik fabrics with background patterns in the case of pattern inclination or bending, and has good pattern rectifying effect and high quality.

Description

Technical Field

[0001] The present invention relates to a method for fabric pattern rectification, in particular to an online automatic fabric pattern rectification method based on machine vision, belonging to the technical field of textile printing and dyeing processes. Background Art

[0002] Fabrics such as lace formed by knitting methods such as knotting, interlacing, and winding of yarns to form a repeating pattern of hollow flower shapes, during post-treatment processes such as washing, drying, stentering, or pre-shrinking, and during the printing process of batik fabrics with background patterns, due to the fabric being in a continuous traction state, affected by various mechanical movements and production operations, and problems such as uneven tension of each fabric guiding roller, the fabric shows distortions such as tilting and bending of the flower pattern and the flower pattern presenting an S-curve. The distortion of the flower pattern will affect the processing quality of subsequent processes and requires pattern rectification.

[0003] Existing fabric pattern rectification methods based on machine vision usually use industrial cameras to collect moving fabric images, use digital image feature extraction technology to extract the feature information of the fabric image, and rely on the fitting of several sampling points to obtain the flower shape deformation amount of the fabric image to be detected, and then use the rectification device of the weft straightener to rectify the flower shape deformation of the fabric, which has good adaptability and detection accuracy. However, for easily deformable fabrics such as lace with repeating flower patterns or the tilting or bending of the flower pattern of batik fabrics with background patterns, the existing pattern rectification methods only rely on the fitting of several sampling points, and the pattern rectification results often have certain deviations, and this deviation situation cannot be solved at all by the existing pattern rectification methods. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide an online automatic fabric pattern rectification method based on machine vision for fabrics with flower patterns, especially for fabrics such as lace with repeating flower patterns or batik fabrics with background patterns, which can accurately calculate the flower shape deformation amount, so as to achieve the purpose of accurately rectifying the tilting and bending of the fabric flower shape.

[0005] To solve the above technical problem, the present invention adopts such an online automatic fabric pattern rectification method based on machine vision, including the following steps:

[0006] S1: Collect the full-width image of the running fabric, and obtain the fabric flower shape area through image processing;

[0007] S2: Perform operations on the fabric flower shape area to obtain the width, height, and number of repeating flower patterns in the fabric flower shape area;

[0008] S3: Set a flower pattern template in the fabric flower pattern area, and use the flower pattern template to detect the position of the flower pattern in the fabric flower pattern area to obtain the position coordinates of the flower pattern;

[0009] S4: According to the position coordinates of the flower pattern, obtain the flower pattern deformation amount of the flower pattern;

[0010] S5: According to the flower pattern deformation amount of the flower pattern, correct the fabric flower pattern deformation through the weft straightening device of the weft straightening machine.

[0011] As a preferred implementation of the present invention, in step S1, an industrial camera collects the full-width image A of the running fabric, and conveys the collected full-width image A of the fabric to a central processor for image processing to obtain the fabric flower pattern area C. The width of the fabric flower pattern area C is C W , and the height is C H ; The industrial camera includes a line array camera or a area array camera, and the central processor includes a digital controller with a human-machine interface, or an embedded control system, or an industrial computer.

[0012] As a preferred implementation of the present invention, the conveying the collected full-width image A of the fabric to a central processor for image processing includes:

[0013] S1.1: Perform filtering and noise reduction processing on the full-width image A of the fabric to obtain the processed full-width image B of the fabric;

[0014] S1.2: Perform edge detection or brightness threshold segmentation processing on the filtered and noise-reduced full-width image B of the fabric to obtain the fabric flower pattern area C.

[0015] As a preferred implementation of the present invention, in step S1.1, the full-width image A of the fabric is filtered and noise-reduced through a filter. The filter includes a mean filter, a median filter, a low-pass filter, a Gaussian filter in the spatial domain filter, or the filter includes a wavelet transform filter, a Fourier transform filter, a cosine transform filter in the frequency domain filter, or the filter includes a morphological filter that performs noise reduction through morphological operations such as dilation, erosion, or opening and closing operations; In step S1.2, edge detection is performed on the filtered and noise-reduced full-width image B of the fabric through the sobel algorithm, or the Roberts algorithm, or the Prewitt algorithm, or the Laplacian algorithm, or the Canny algorithm to obtain the fabric flower pattern area C. The width of the fabric flower pattern area C is C W , and the height is C H; or performing brightness threshold segmentation on the full-width fabric image B after filtering and noise reduction through a fixed threshold segmentation method, a threshold segmentation method based on a grayscale histogram, an adaptive threshold segmentation method, a maximum entropy threshold segmentation method, or a maximum inter-class variance threshold segmentation method to obtain a fabric flower pattern region C, where the width of the fabric flower pattern region C is C W , and the height is C H .

[0016] As a preferred embodiment of the present invention, in step S2, operations are performed on the fabric flower pattern region to obtain the width, height, and number of the flower pattern repetitions in the fabric flower pattern region. The specific steps are as follows:

[0017] S2.1: The central processing unit performs two-dimensional image Fourier transform on the fabric flower pattern region C to obtain a processed Fourier transform image D;

[0018] S2.2: The translation method is used to place the low-frequency components of the Fourier transform image D at the center of the image and the high-frequency components at the four corners of the image to obtain a processed Fourier transform image E;

[0019] S2.3: According to the Fourier transform image E, calculate the width D of the flower pattern repetition in the width direction of the fabric flower pattern region C W , the height D in the movement direction H and the number D E .

[0020] As a preferred embodiment of the present invention, in step S2.3, according to the Fourier transform image E, the Fourier cross-correlation algorithm is used to calculate the width D of the flower pattern repetition in the width direction of the fabric flower pattern region C W , the height D in the movement direction H and the number D E .

[0021] As a preferred embodiment of the present invention, in step S3, a flower pattern template is set in the fabric flower pattern region, and the position coordinates of the flower pattern in the fabric flower pattern region are detected by using the flower pattern template. The specific steps are as follows:

[0022] S3.1: According to the width D of the flower pattern repetition W , the fabric flower pattern region C is equally divided into F equal parts from left to right in the width direction, where F ≥ 2;

[0023] S3.2: Starting from the left edge of the first equal part of the flower pattern repetition and ending at the left edge of the second equal part of the flower pattern repetition, N flower patterns are selected and set as N flower pattern templates, and the distance W between adjacent two flower pattern templates i(i = 1, 2, …, N - 1) satisfies the condition: W i (i = 1, 2, …, N - 1) ≤ 3D H When i = 1, W 1 is the distance between the first flower pattern template and the second flower pattern template; when i = 2, W 2 is the distance between the second flower pattern template and the third flower pattern template, and so on;

[0024] The height H of each flower pattern template j (j = 1, 2, …, N) satisfies the condition: H j (j = 1, 2, …, N) ≤ D H When j = 1, H 1 is the height of the first flower pattern template, when j = 2, H 2 is the height of the second flower pattern template, and so on; and, the first flower pattern template located at the left edge of the first equal - part flower pattern repeat and the last flower pattern template located at the left edge of the second equal - part flower pattern repeat select the same flower pattern, and the last two flower pattern templates are on the same horizontal line;

[0025] S3.3: Sequentially starting from the left - most edge of the previous equal - part flower pattern repeat of two adjacent equal - part flower pattern repeats and ending at the left - most edge of the next equal - part flower pattern repeat for the fabric flower pattern area C from left to right, create a search area according to the information of N flower pattern templates, and determine the central position coordinates of each flower pattern corresponding to the N flower pattern templates within the search area, and finally obtain the central position coordinates of F×(N - 1) flower patterns within the fabric flower pattern area C.

[0026] As a preferred implementation of the present invention, in step S4, the central processing unit calculates the flower bend deviation amount and the flower skew deviation amount of the F×(N - 1) flower patterns according to the central position coordinates of the F×(N - 1) flower patterns within the fabric flower pattern area C.

[0027] After adopting the above - mentioned whole - flower method, the present invention has the following beneficial effects:

[0028] For fabrics with flower patterns, especially for fabrics such as lace with flower pattern repeats or batik fabrics with background patterns in the case of flower tilt or bend, the present invention can automatically and accurately calculate the flower bend deviation amount and the flower skew deviation amount, and achieve the purpose of automatically and accurately correcting the flower tilt and bend of the fabric through machine vision. It has strong operability, can greatly improve the qualified rate of fabrics, especially lace fabrics, solves the problem of flower pattern distortion during post - finishing processes such as washing, drying, stenter setting or pre - shrinking of existing fabrics, and greatly reduces the operation difficulty.

[0029] The present invention has good overall flower arranging effect and high quality, can accurately correct distortions such as the inclination and bending of the flower shape, and better meets the requirements of fabric overall flower arranging.

[0030] The present invention greatly reduces the skill requirements and labor intensity of operators and improves production efficiency.

[0031] The present invention ensures the qualification rate of printed products and brings greater economic benefits to enterprises. Specific embodiments

[0032] The following further illustrates the present invention in conjunction with embodiments.

[0033] An online automatic fabric overall flower arranging method based on machine vision preferably uses an existing weft straightening machine as the automatic flower arranging equipment, such as the weft straightening machines disclosed in Chinese Utility Model Patents with patent numbers 202121938486.5, 202121174391.0, 202121175396.5, etc., and includes the following steps:

[0034] S1: Collect the full-width image of the running fabric, and obtain the fabric flower shape area through image processing. In the present invention, the full-width image of the fabric is an image including the overall width of the fabric;

[0035] S2: Perform operations on the fabric flower shape area to obtain the width, height, and number of the flower shape repeating patterns in the fabric flower shape area;

[0036] S3: Set a flower shape template in the fabric flower shape area, and use the flower shape template to detect the position of the flower shape pattern in the fabric flower shape area to obtain the position coordinates of the flower shape pattern;

[0037] S4: According to the position coordinates of the flower shape pattern, obtain the flower shape deformation amount of the flower shape pattern;

[0038] S5: According to the flower shape deformation amount of the flower shape pattern, correct the fabric flower shape deformation through the weft straightening machine correction device.

[0039] As a preferred implementation scheme of the present invention, in step S1, an industrial camera collects the full-width image A of the running fabric, and conveys the collected full-width image A of the fabric to a central processor for image processing to obtain the fabric flower shape area C. The width of the fabric flower shape area C is C W , and the height is C H ; the industrial camera includes a line array camera or a area array camera, the central processor includes a digital controller with a human-machine interface such as a DDC digital controller or an embedded control system or an industrial computer, the industrial camera is connected to the central processor, and the central processor can be the central processor of the weft straightening machine or a separately provided central processor.

[0040] As a preferred embodiment of the present invention, the step of transporting the collected full-width fabric image A to the central processor for image processing includes:

[0041] S1.1: Perform filtering and noise reduction processing on the full-width fabric image A to obtain the processed full-width fabric image B;

[0042] S1.2: Perform edge detection or brightness threshold segmentation processing on the filtered and noise-reduced full-width fabric image B to obtain the fabric pattern region C.

[0043] In step S1.1 of the present invention, preferably, the full-width fabric image A is subjected to filtering and noise reduction processing by a filter. The filter includes a mean filter, a median filter, a low-pass filter, a Gaussian filter in the spatial domain filter, or the filter includes a wavelet transform filter, a Fourier transform filter, a cosine transform filter in the frequency domain filter, or the filter includes a morphological filter that performs noise reduction by morphological operations such as dilation, erosion, or opening and closing operations.

[0044] In step S1.2 of the present invention, preferably, the filtered and noise-reduced full-width fabric image B is subjected to edge detection by the sobel algorithm or the Roberts algorithm or the Prewitt algorithm or the Laplacian algorithm or the Canny algorithm to obtain the fabric pattern region C. The width of the fabric pattern region C is C W , and the height is C H ; or perform brightness threshold segmentation processing on the filtered and noise-reduced full-width fabric image B by a fixed threshold segmentation method or a threshold segmentation method based on a gray histogram or an adaptive threshold segmentation method or a maximum entropy threshold segmentation method or a maximum inter-class variance threshold segmentation method to obtain the fabric pattern region C. The width of the fabric pattern region C is C W , and the height is C H .

[0045] As a preferred embodiment of the present invention, in step S2, perform operations on the fabric pattern region to obtain the width, height, and number of the pattern repeat patterns in the fabric pattern region. The specific steps are:

[0046] S2.1: The central processor performs two-dimensional image Fourier transform on the fabric pattern region C to obtain the processed Fourier transform image D;

[0047] S2.2: Use the translation method to place the low-frequency components of the Fourier transform image D at the center of the image and the high-frequency components at the four corners of the image to obtain the processed Fourier transform image E;

[0048] S2.3: Calculate the width D of the flower pattern repeat pattern of the fabric flower pattern area C in the width direction based on the Fourier transform image E W , the height D in the moving direction H and the number D E .

[0049] As a preferred embodiment of the present invention, in step S2.3, based on the Fourier transform image E, the Fourier cross-correlation algorithm is used to calculate the width D of the flower pattern repeat pattern of the fabric flower pattern area C in the width direction W , the height D in the moving direction H and the number D E .

[0050] As a preferred embodiment of the present invention, in step S3, a flower pattern template is set in the fabric flower pattern area, and the position coordinates of the flower pattern in the fabric flower pattern area are detected by using the flower pattern template. The specific steps are as follows:

[0051] S3.1: Divide the fabric flower pattern area C into F equal parts from left to right in the width direction according to the width D W or the number D of flower pattern repeats E , where F≥2; for example, F = 6, that is, divide the fabric flower pattern area C into 6 equal parts according to the width D W or the number D of flower pattern repeats E , the leftmost side of the fabric flower pattern area C is the first equal part, and then they are the second equal part, the third equal part, the fourth equal part, the fifth equal part, and the sixth equal part in sequence, and the sixth equal part is located on the rightmost side;

[0052] S3.2: Starting from the left edge of the first equal part of the flower pattern repeat pattern and ending at the left edge of the second equal part of the flower pattern repeat pattern, select N flower patterns and set them as N flower pattern templates , that is, N is greater than or equal to the quotient of D W divided by 3D H rounded up. The distance W i (i = 1, 2,..., N - 1) between adjacent two flower pattern templates should meet the condition: W i (i = 1, 2,..., N - 1) ≤ 3D H , when i = 1, W 1 is the distance between the first flower pattern template and the second flower pattern template; when i = 2, W 2 is the distance between the second flower pattern template and the third flower pattern template, and so on;

[0053] The height H j (j = 1, 2,..., N) of each flower pattern template should meet the condition: H j(j = 1, 2, …, N) ≤ D H When j = 1, H 1 is the height of the first flower-shaped template. When j = 2, H 2 is the height of the second flower-shaped template, and so on. Moreover, the first flower-shaped template located at the left edge of the first equal-part flower-shaped repeating pattern and the last flower-shaped template located at the left edge of the second equal-part flower-shaped repeating pattern should select the same flower-shaped pattern, and the last two flower-shaped templates should be on the same horizontal line. For example, when N = 7, that is, 7 flower-shaped patterns are selected as 7 flower-shaped templates, and the 7 flower-shaped templates should meet the above adjacent spacing and height requirements. Moreover, the first flower-shaped template located at the left edge of the first equal-part flower-shaped repeating pattern and the seventh flower-shaped template located at the left edge of the second equal-part flower-shaped repeating pattern should select the same flower-shaped pattern, that is, the first and the seventh flower-shaped templates should select the same flower-shaped pattern, and the last two flower-shaped templates, namely the sixth and the seventh flower-shaped templates, should be on the same horizontal line;

[0054] S3.3: Starting from the left edge of the previous equal-part flower-shaped repeating pattern of two adjacent equal-part flower-shaped repeating patterns in sequence from left to right for the fabric flower-shaped area C, and ending at the left edge of the next equal-part flower-shaped repeating pattern, create a search area according to the information of N flower-shaped templates, and automatically calculate and determine the central position coordinates of each flower-shaped pattern corresponding to the N flower-shaped templates, that is, the central position coordinates of N flower-shaped patterns within the search area through the central processing unit. The automatic calculation includes the calculation of the central position coordinates of the flower-shaped pattern based on the template matching algorithm and the image correlation algorithm. Finally, the central position coordinates of F×(N - 1) flower-shaped patterns within the fabric flower-shaped area C are obtained. For example, when F = 6 and N = 7 as above, finally, 6×(7 - 1) = 36 central position coordinates of flower-shaped patterns within the fabric flower-shaped area C can be obtained.

[0055] As a preferred implementation of the present invention, in step S4, the central processing unit calculates the flower bending deviation amount and the flower skew deviation amount of these flower-shaped patterns according to the central position coordinates of F×(N - 1) flower-shaped patterns within the fabric flower-shaped area C.

[0056] During the correction work of the present invention, the central processing unit controls the action of the weft straightening device of the whole weft machine according to the obtained flower bending deviation amount and flower skew deviation amount of the flower-shaped pattern to complete the automatic flower arrangement of the fabric.

[0057] After testing, for the flower-shaped tilting or bending conditions of fabrics such as lace with flower-shaped repeating patterns or batik fabrics with background patterns, the present invention can automatically and accurately calculate the flower bending deviation amount and the flower skew deviation amount, with good flower arrangement effect and high quality. It solves the problem of flower-shaped deformation of existing fabrics in post-treatment processes such as washing, drying, stentering setting, or pre-shrinking, ensures the qualification rate of printed products, and achieves good results.

Claims

1. An online automatic fabric pattern rectification method based on machine vision, characterized in that, it includes the following steps: S1: Collect the full-width image of the running fabric, and obtain the fabric pattern area through image processing; S2: Perform operations on the fabric pattern area to obtain the width, height and number of the pattern repeat patterns in the fabric pattern area; S3: Set a pattern template in the fabric pattern area, and use the pattern template to detect the pattern position of the fabric pattern area to obtain the position coordinates of the pattern; S4: According to the position coordinates of the pattern, obtain the pattern deformation amount of the pattern; S5: According to the pattern deformation amount of the pattern, correct the fabric pattern deformation through the weft straightening device of the weft straightening machine; In step S2, perform operations on the fabric pattern area to obtain the width, height and number of the pattern repeat patterns in the fabric pattern area. The specific steps are: S2.1: The central processing unit performs two-dimensional image Fourier transform on the fabric pattern area C to obtain the processed Fourier transform image D; S2.2: Use the translation method to place the low-frequency components of the Fourier transform image D at the center of the image and the high-frequency components at the four corners of the image to obtain the processed Fourier transform image E; S2.3: Calculate the width D of the flower pattern repeat pattern of the fabric flower pattern region C in the width direction based on the Fourier transform image E W , the height D in the movement direction H and the number D E ; In step S2.3, according to the Fourier transform image E, the Fourier cross-correlation algorithm is used to calculate the width D of the flower pattern repeat pattern of the fabric flower pattern region C in the width direction W , the height D in the moving direction H and the number D E ; In step S3, set a pattern template in the fabric pattern area, and use the pattern template to detect the pattern position of the fabric pattern area to obtain the position coordinates of the pattern. The specific steps are: S3.1: According to the width D of the flower-shaped repeating pattern W , equally divide the fabric flower-shaped area C into F equal parts in the width direction from left to right, where F ≥ 2; S3.2: Starting from the left edge of the first equal - part flower - shaped repeating pattern and ending at the left edge of the second equal - part flower - shaped repeating pattern, select N flower - shaped patterns and set them as N flower - shaped templates. The distance W between two adjacent flower - shaped templates i (i = 1, 2, …, N−1) satisfies the condition: W i (i = 1, 2, …, N−1) ≤ 3D H , when i = 1, W 1 is the distance between the first flower - shaped template and the second flower - shaped template; when i = 2, W 2 is the distance between the second flower - shaped template and the third flower - shaped template, and so on. The height H of each flower-shaped template j (j = 1, 2, …, N) satisfies the condition: H j (j = 1, 2, …, N) ≤ D H , when j = 1, H 1 is the height of the first flower-shaped template, when j = 2, H 2 is the height of the second flower-shaped template, and so on; and, the first flower-shaped template located at the left edge of the first equal-part flower-shaped repeating pattern and the last flower-shaped template located at the left edge of the second equal-part flower-shaped repeating pattern select the same flower-shaped pattern, and the last two flower-shaped templates are on the same horizontal line; S3.3: Starting from the left edge of the previous equal-part pattern repeat pattern of two adjacent equal-part pattern repeat patterns in the fabric pattern area C from left to right, and ending with the left edge of the next equal-part pattern repeat pattern, create a search area according to the information of N pattern templates, and determine the center position coordinates of each pattern corresponding to the N pattern templates in the search area. Finally, obtain the center position coordinates of F×(N - 1) patterns in the fabric pattern area C.

2. The online automatic fabric pattern rectification method based on machine vision according to claim 1, characterized in that: In step S1, an industrial camera acquires a full-width image A of the running fabric, and conveys the acquired full-width image A of the fabric to a central processor for image processing to obtain a fabric pattern area C, where the width of the fabric pattern area C is C W , and the height is C H ; the industrial camera includes a line array camera or a area array camera, and the central processor includes a digital controller with a human-machine interface, or an embedded control system, or an industrial computer.

3. The online automatic fabric pattern rectification method based on machine vision according to claim 2, characterized in that: The transmission of the collected full-width fabric image A to the central processing unit for image processing includes: S1.1: Perform filtering and noise reduction processing on the full-width fabric image A to obtain the processed full-width fabric image B; S1.2: Perform edge detection or brightness threshold segmentation processing on the filtered and noise-reduced full-width fabric image B to obtain the fabric pattern area C.

4. The online automatic fabric pattern rectification method based on machine vision according to claim 3, characterized in that: In step S1.1, the full-width fabric image A is filtered and noise-reduced through a filter. The filter includes a mean filter, a median filter, a low-pass filter, a Gaussian filter in the spatial domain filter, or the filter includes a wavelet transform filter, a Fourier transform filter, a cosine transform filter in the frequency domain filter, or the filter includes a morphological filter that performs noise reduction through morphological operations such as dilation, erosion, or opening and closing operations; In step S1.2, edge detection is performed on the full-width fabric image B after filtering and noise reduction through the Sobel algorithm, Roberts algorithm, Prewitt algorithm, Laplacian algorithm, or Canny algorithm to obtain the fabric pattern area C, where the width of the fabric pattern area C is C W , and the height is C H ; alternatively, brightness threshold segmentation processing is performed on the full-width fabric image B after filtering and noise reduction through the fixed threshold segmentation method, threshold segmentation method based on gray histogram, adaptive threshold segmentation method, maximum entropy threshold segmentation method, or maximum inter-class variance threshold segmentation method to obtain the fabric pattern area C, where the width of the fabric pattern area C is C W , and the height is C H .

5. The on-line automatic fabric pattern rectifying method based on machine vision according to claim 1, characterized in that: in step S4, the central processing unit calculates the flower bend deviation amount and the flower skew deviation amount of the F×(N−1) flower patterns according to the central position coordinates of the F×(N−1) flower patterns in the fabric flower pattern area C.