CN115496751A - Winding detection method of fiber winding machine - Google Patents
Winding detection method of fiber winding machine Download PDFInfo
- Publication number
- CN115496751A CN115496751A CN202211429710.7A CN202211429710A CN115496751A CN 115496751 A CN115496751 A CN 115496751A CN 202211429710 A CN202211429710 A CN 202211429710A CN 115496751 A CN115496751 A CN 115496751A
- Authority
- CN
- China
- Prior art keywords
- winding
- gray
- cluster
- value
- target image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
- G06V10/763—Non-hierarchical techniques, e.g. based on statistics of modelling distributions
Abstract
The invention relates to the technical field of image processing, in particular to a winding detection method of a fiber winding machine. The method comprises the steps of collecting winding images and corresponding target images of fiber rolls wound by a fiber winding machine; determining diamond edge points according to the gray level change and the density of pixel points in the target image; clustering the diamond-shaped edge points according to the gray value and the density to obtain an initial cluster and a small cluster; horizontally extending the central point of a small cluster in the initial cluster with the minimum gray value to two sides to obtain the gray value change value of a pixel point on an extension line; when the gray level change value is larger than a preset change threshold value, calculating the surface flatness according to the entropy of the target image, the distance from the pixel points in the small clusters to the cluster centers of the small clusters and the number of the initial clusters; and determining the winding quality grade of the filament winding machine according to the surface flatness. The invention realizes the winding quality detection of the filament winding machine under the condition of complex filament winding by calculating the surface flatness of the target image.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a winding detection method of a fiber winding machine.
Background
The fiber winding formation is one of the main production processes of resin-base composite material, and is one of the processes of impregnating continuous fiber roving or cloth belt with resin glue solution, winding the impregnated fiber roving or cloth belt onto the core mold or inner lining of the inner cavity size corresponding to the product and curing at room temperature or heating condition to form the product in certain shape. Whatever form of entanglement falls into three categories: hoop winding, planar winding, and spiral winding. The spiral winding is characterized in that each bundle of fibers corresponds to a tangent point on the circumference of the pole hole; adjacent yarn sheets in the same direction are connected but not intersected, and fibers in different directions are intersected. Thus, when the fibers are uniformly wound on the surface of the core mold, a double-layer fiber layer is formed.
At present, the common method for detecting the fiber winding is to adjust the gray value of the detected image, perform integral noise reduction processing on the image, obtain a static digital image after integral noise reduction, perform depth positioning on the defect part in the static digital image, and evaluate the defect property according to the difference between the gray value of the defect part and the gray value of the normal part. However, since the winding of the spirally wound fiber roll is complicated, and there are cases of multi-layer winding, it is difficult to locate the defect portion in the image based on the gray threshold value, so that the evaluation result of the winding of the fiber winding machine based on the difference between the gray values of the defect portion and the normal portion is affected.
Disclosure of Invention
In order to solve the technical problems, the invention aims to provide a winding detection method of a filament winding machine, which adopts the following technical scheme:
acquiring a winding image of a fiber roll wound by a fiber winding machine, and preprocessing the winding image to obtain a target image;
determining diamond-shaped edge points to be selected formed during winding according to the gray level change of the target image; determining diamond edge points according to the density of the edge points to be selected;
clustering the rhombic edge points according to the gray values to obtain at least two initial clusters; clustering the diamond-shaped edge points of each initial cluster based on the density to obtain at least two small clusters; selecting each small cluster in the initial cluster with the minimum gray value as an edge cluster, extending horizontally to two sides by using the central point of the edge cluster, and acquiring a gray change value according to the gray value of a pixel point on an extension line; when the gray level change value is larger than a preset change threshold value, calculating the surface flatness of the target image according to the entropy of the target image, the distance from the pixel points in the small clusters to the cluster centers of the small clusters and the number of the initial clusters;
and determining the winding quality grade of the filament winding machine according to the surface flatness.
Preferably, the calculation formula of the surface flatness is as follows:
wherein, the first and the second end of the pipe are connected with each other,
is the surface flatness;
is a natural constant;
the entropy of the target image;
the maximum gray value of the pixel points in the target image is obtained;
the minimum gray value of a pixel point in the target image is obtained;
the number of small clusters corresponding to the initial cluster with the gray value I;
the Euclidean distance from the ith pixel point in the small cluster with the gray value of I to the cluster center point of the small cluster; r is the number of pixels in the initial cluster with the gray value I; and I is the gray value of the pixel points in the initial cluster.
Preferably, the determining of the diamond-shaped edge point to be selected formed during winding according to the gray scale change of the target image includes:
and calculating the gradient amplitude of each pixel point in the target image, and taking the pixel point with the gradient amplitude larger than a preset gradient threshold value as an edge point to be selected.
Preferably, the determining the diamond-shaped edge points according to the density of the edge points to be selected includes:
calculating the number proportion of pixel points with the same gradient amplitude value as the edge point to be selected in the eight neighborhoods of the edge point to be selected; and taking the pixel points with the number ratio larger than a preset ratio threshold value as diamond edge points.
Preferably, the extending of the central point of the edge cluster to both sides horizontally, and obtaining the gray variation value according to the gray value of the pixel point on the extension line, include:
and acquiring the gray value difference value of each pixel point on the extension line and the central point of the edge cluster, wherein the sum of the gray value difference values corresponding to all the pixel points on the extension line is the gray change value.
Preferably, the determining the winding quality grade of the filament winding machine according to the surface flatness comprises:
when the surface flatness is larger than a preset first threshold value, the winding quality grade of the filament winding machine is excellent; when the surface flatness is smaller than or equal to a preset first threshold value and larger than a preset second threshold value, the winding quality grade of the fiber winding machine is qualified; and when the surface flatness is less than or equal to a preset second threshold value, the winding quality grade of the filament winding machine is unqualified.
Preferably, the preprocessing the winding image to obtain a target image includes:
graying the winding image to obtain a gray image, and performing semantic segmentation on the gray image to obtain a target image only containing fiber rolls.
The embodiment of the invention at least has the following beneficial effects:
firstly, determining diamond-shaped edge points to be selected formed during winding according to the gray level change of a target image; and then determining the diamond-shaped edge points according to the density degree of the edge points to be selected, and firstly determining the diamond-shaped edge and the corresponding diamond-shaped edge points formed during winding according to the gray value and the density degree of the pixel points in the target image so as to be convenient for subsequent analysis of the diamond-shaped edge and obtain the surface flatness corresponding to the target image.
Clustering the diamond edge points twice according to the gray value and the density to obtain an initial cluster and a small cluster; selecting each small cluster in the initial cluster with the minimum gray value as an edge cluster, extending horizontally to two sides by using the central point of the edge cluster, obtaining a gray change value according to the gray value of the pixel point on the extension line, clustering according to the gray value and the density of the diamond-shaped edge point, calculating the gray change value on the extension line of the clustered small clusters, and calculating the surface evenness of the target image by combining the gray change value.
And when the gray level change value is larger than a preset change threshold value, calculating the surface flatness of the target image according to the entropy of the target image, the distance from the pixel points in the small clusters to the cluster centers of the small clusters and the number of the initial clusters. And determining the winding quality grade of the fiber winding machine according to the surface flatness. When the gray level change value is larger than the preset change threshold value, the surface flatness of the target image is calculated because the gray level change value is larger than the preset change threshold value when the condition of multi-layer rhombus superposition occurs, and the condition of the multi-layer rhombus superposition is complex and difficult to analyze, so the surface flatness of the target image is analyzed by calculating the surface flatness of the target image. The entropy of the target image reflects the complexity of the target image, the distance from the pixel points in the small clusters to the clustering centers of the small clusters reflects the density of diamond-shaped edge points, the surface evenness of the target image is reflected by combining the number of the edge clusters, and the winding quality of the corresponding filament winding machine can be reflected according to the surface evenness. According to the invention, the surface flatness of the target image is calculated, so that the condition that the winding quality is not in accordance with the standard due to the overlapping of the fiber yarns or the overlarge gap between the fiber yarns in the winding process of the fiber yarns is reflected, and the winding quality detection of the fiber winding machine under the condition of complicated fiber winding is realized.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions and advantages of the prior art, the drawings used in the embodiments or the description of the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
FIG. 1 is a flow chart of a method of a filament winding machine winding detection method according to an embodiment of the present invention;
FIG. 2 is a schematic illustration of a wound image of a fiber roll provided by one embodiment of the present invention;
FIG. 3 is a schematic diagram of a multilayer diamond stack according to an embodiment of the present invention.
Detailed Description
To further illustrate the technical means and effects of the present invention adopted to achieve the predetermined objects, the following detailed description of the winding detection method of the filament winding machine according to the present invention, the specific implementation manner, structure, features and effects thereof will be provided in conjunction with the accompanying drawings and the preferred embodiments. In the following description, different "one embodiment" or "another embodiment" refers to not necessarily the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The embodiment of the invention provides a specific implementation method of a winding detection method of a fiber winding machine, which is suitable for a winding detection scene of the fiber winding machine. And acquiring a winding image of the fiber roll wound by the fiber winding machine through the camera under the scene, wherein the winding form of the fiber winding machine is spiral winding under the scene. The method aims to solve the problems that the winding condition of a spirally wound fiber roll is complex, the defect part in an image is difficult to position according to a gray threshold value, and the subsequent evaluation result of the winding of a fiber winding machine according to the difference of the gray values of the defect part and the normal part is influenced. The method analyzes the gray value and the dense condition of the pixel points in the target image corresponding to the winding image of the fiber roll wound by the fiber winding machine, calculates the surface flatness of the target image, and can indicate the condition that the winding quality does not meet the standard due to the overlapping of the fiber yarns or the overlarge gap between the fiber yarns in the winding process of the fiber yarns. The smaller the surface flatness, the more drastic the gradation change of the target image, and the less standard the quality of the corresponding fiber roll. The surface flatness was used as a quality evaluation parameter for evaluating filament winding.
The specific scheme of the winding detection method of the filament winding machine provided by the invention is specifically described below by combining the attached drawings.
Referring to fig. 1, a flow chart of steps of a winding detection method of a filament winding machine according to an embodiment of the present invention is shown, the method includes the following steps:
and S100, acquiring a winding image of the fiber roll wound by the fiber winding machine, and preprocessing the winding image to obtain a target image.
The invention mainly aims to detect the quality of a fiber winding roll of a fiber winding machine, so that a surface image, namely a winding image, of the fiber winding roll wound by the fiber winding machine needs to be acquired. Specifically, the CCD camera is used for collecting RGB images on the surface of the fiber roll through a overlooking visual angle, and in the process of collecting the winding images, the required ambient light is uniform without other influence factors. Since the fiber roll is cylindrical, the winding image of the fiber roll is acquired from four directions for one fiber roll, and the winding image of the fiber roll acquired from any one direction is analyzed in the present invention as an example.
Further, preprocessing the winding image to obtain a target image, specifically: performing graying processing on the acquired winding image to obtain a corresponding grayscale image; and performing semantic segmentation on the gray level image to obtain a target image only containing the fiber roll. The embodiment of the invention adopts a DNN semantic segmentation mode to identify and segment the fiber volume in the gray level image. The semantic segmentation method is well known to those skilled in the art and will not be described herein.
Step S200, determining diamond-shaped edge points to be selected formed in the winding process according to the gray level change of the target image; and determining the diamond edge points according to the density of the edge points to be selected.
In the process of winding the fibers, the fiber winding machine can wind the fiber yarns according to a certain rule in order to prevent the loose of the wound fibers, and the winding times of each circle are the same, so that the surface of the fiber roll after the fibers are wound is smooth and flat, and the texture is regular. Therefore, whether the fiber roll has problems in the winding process is judged by the regular degree of the surface texture of the fiber roll.
And in the process of winding the fibers, the fixed die can be rotated according to a certain rule, so that the fibers have a certain error after being wound, the surfaces of the fibers present rhombic patterns, and the fibers are uniformly arranged. Therefore, the degree of regularity of the surface texture of the fiber roll is obtained by detecting the degree of regularity of the entire surface of the fiber roll.
The surface of the fiber yarn can form diamond patterns in the winding process, and the regular degree of the patterns can show whether the winding speed is uniform in the winding process of the winding machine, so that the regular degree of the surface pattern of the fiber roll can be obtained according to the change of the surface texture. Referring to fig. 2, fig. 2 is a schematic view of a winding image of a fiber roll.
And winding the fiber by a fiber winding machine to form rhombic grains. Because the direction of the fiber yarn is changed when the fiber yarn is wound on the edge of the diamond pattern, the fiber yarn in different directions is overlapped to form the gray scale change. For ease of description, the texture is described directly below as diamond-shaped texture. The gray value of the pixel point at the edge of the diamond is smaller, and the gray value of the pixel point in the diamond area is larger, so that the edge of the diamond area and the corresponding diamond edge point are obtained according to the gray change, and then the rule degree of the diamond area is judged.
Firstly, determining diamond-shaped edge points to be selected formed in winding according to the gray level change of a target image, specifically: calculating the gradient amplitude of each pixel point in the target image, and taking the pixel points with the gradient amplitudes larger than a preset gradient threshold value as edge points to be selected, namely, passing through
Calculating gradient amplitude of each pixel point in image by operator
And taking the pixel points with the gradient amplitude values larger than the preset gradient threshold value as edge points to be selected. In the embodiment of the present invention, the value of the preset gradient threshold is 5, and in other embodiments, an implementer may adjust the value according to an actual situation. Because the gray level change is large on the edge of the rhombus, the gradient of the pixel point is large, and the gray level change of the pixel point in the inner area of the rhombus is small, and the gradient of the pixel point is small. Therefore, the texture change of the image is obtained according to the change of the gradient of the pixel points. Firstly, dividing pixel points of the image according to the size of the gradient, taking the size of each gradient as a gradient level, and classifying the pixel points of the same gradient level into one class. Because the distribution of the pixel points with large gradients on the diamond edges is more, the aggregation degree of the pixel points with different gradients is calculated to obtain the edge of each diamond. And calculating the aggregations of the pixel points with the gradient amplitude larger than 5, so that the pixel points in the diamond can be removed.
Further, determining the diamond edge points according to the density of the edge points to be selected, specifically: calculating the number ratio of pixel points with the same gradient amplitude value as the edge point to be selected in the eight neighborhoods of the edge point to be selected; and taking the pixel points with the number ratio larger than a preset ratio threshold value as diamond edge points. Namely, a 3 × 3 sliding window is set, each pixel point in the image is traversed, and the concentration degree of the pixel points with the gradient larger than 5 is calculated, wherein the concentration degree is the ratio of the number of the pixel points with the same gradient amplitude value as the central point of the sliding window in the sliding window. If the number of the pixel points with the same gradient amplitude value in the sliding window is large, the pixel points may be located at the edge of the diamond grains. And marking the pixel points with the number ratio larger than a preset ratio threshold value to obtain all marked pixel points in the target image. Because the pixel points on the diamond grains are all positioned on the straight line of the edge, the marked pixel points are the diamond edge points. In the embodiment of the present invention, the value of the preset duty ratio threshold is 0.65, and in other embodiments, an implementer may adjust the value according to actual conditions.
Step S300, clustering the diamond-shaped edge points according to the gray values to obtain at least two initial clusters; clustering the diamond-shaped edge points of each initial cluster based on the density to obtain at least two small clusters; selecting each small cluster in the initial cluster with the minimum gray value as an edge cluster, extending horizontally to two sides by using the central point of the edge cluster, and acquiring a gray change value according to the gray value of a pixel point on an extension line; and when the gray level change value is larger than a preset change threshold value, calculating the surface flatness of the target image according to the entropy of the target image, the distance from the pixel points in the small clusters to the cluster centers of the small clusters and the number of the initial clusters.
When winding the fiber wire, the winding machine winds the fiber wire into a fixed pattern shape according to a certain rule, and whether the fiber wire is uniformly wound is the key for evaluating the winding quality. Whether the surface of the wound texture is flat or not is also a standard for evaluating winding quality, and the more uniform the pattern is, the greater the surface flatness is, and the flat surface and uniform and regular pattern indicate that the winding quality of the fiber winding machine is qualified.
The diamond-shaped edge points are obtained, namely the diamond grains are determined. However, when the edges are divided according to the gradient, not only uniform rhomboid lines can appear, but also multilayer gaps can appear between rhomboids due to the superposition of multiple layers of rhomboids. While multiple layers of voids may form regions of lesser gray scale, resulting in greater variation in edge gradients. Therefore, after the diamond edge points of the diamond-shaped area are determined, the multi-layer gap area needs to be identified. Referring to fig. 3, fig. 3 is a schematic diagram of a multi-layer diamond stack. The diamond in the middle of fig. 3 is the diamond in the lowest layer of the stack, and a void appears in the non-stacked area, and the gray value of the void is smaller than that of the normal edge, so that whether the void appears due to the stacking of multiple layers of diamonds is judged according to the determined diamond edge.
Clustering the diamond-shaped edge points according to the gray values to obtain at least two initial clusters; and clustering the diamond-shaped edge points of each initial cluster based on the density to obtain at least two small clusters. Specifically, the method comprises the following steps: the rhombic edge points are clustered by a gray level clustering method, the rhombic edge points with the same gray level are clustered into a class, and at least two clustering clusters are called initial clusters. Because only the gray scale difference between the pixel points of the fiber yarns and the pixel points of the gaps is formed in the winding image, fewer categories can be formed during clustering. Therefore, density clustering is carried out on each initial cluster to obtain at least two small clusters with different densities, and the clustering center of each small cluster is obtained. It should be noted that, since the gray values of the diamond-shaped edge points in the initial cluster are the same, the gray values of the diamond-shaped edge points in each small cluster obtained by performing density clustering on the diamond-shaped edge points in the initial cluster are also the same.
Because the obtained small clusters are also cluster clusters of different areas under the same gray level, when the cluster is used for determining the area, a multilayer cavity area is obtained according to the gray level change of pixel points in the clusters and the gray level change between the clusters. When the hollow areas are arranged in multiple layers, the gray scale of the hollow areas is gradually increased from inside to outside because the rhombic areas are overlapped, the gray scale change of the edges of the rhombuses is large, and the gray scale change of the interiors of the rhombuses is small, so that the hollow areas are obtained according to the gray scale change of the cluster clusters. The clustering cluster with the large gray value is an internal clustering cluster in the diamond, the clustering cluster with the small gray value is an edge clustering cluster on the edge of the diamond, and the method specifically comprises the following steps: and selecting each small cluster in the initial cluster with the minimum gray value as an edge cluster, wherein each small cluster corresponding to other initial clusters is an internal cluster.
Because the diamond-shaped edge is obtained in the calculation process, the pixel points at the diamond-shaped edge are converged into an edge cluster of one type, and other areas form a plurality of types. For the edge cluster corresponding to the diamond-shaped edge, the central point of the edge cluster horizontally extends towards two sides, and a gray level change value is obtained according to the gray level value of the pixel points on the extension line, namely when the pixel points on the extension line have continuous change of gray level and the middle has mutation of gray level, the changed area is considered as the cavity area of multilayer diamond superposition.
The method for acquiring the gray scale change value comprises the following steps: and acquiring gray value difference values of each pixel point on the extension line and the central point of the edge cluster, wherein the sum of the gray value difference values corresponding to all the pixel points on the extension line is a gray variation value. The gray level change value is also the gray level change condition of the cluster.
When the multilayer rhombus superposition occurs, the process that the gray value of the clustering cluster extends outwards is gradually reduced, and the gray value of the area without the superposed rhombus is changed alternately, so whether the multilayer rhombus superposition occurs is judged according to whether the gray value in the area is continuously changed.
After the gray level change value is obtained, when the condition of multilayer rhombus superposition occurs, the value of the gray level change value is larger than a preset change threshold value; when the condition of no multilayer rhombus superposition occurs, the gray scale change value is less than or equal to the preset change threshold value. In the embodiment of the present invention, the value of the preset change threshold is 0, and in other embodiments, an implementer may adjust the value according to an actual situation. When multilayer rhombus superposition does not occur, the defect part in the current target image can be determined through the prior art, the defect property is judged according to the difference between the gray value of the defect part and the gray value of the normal part, and the corresponding defect property is also the defect property when the fiber winding machine winds, and the description is not repeated. The embodiment of the invention mainly analyzes the quality evaluation of the winding of the fiber winding machine when a multilayer overlapping condition occurs.
Therefore, after the condition that the multilayer rhombus is overlapped is determined, when the surface flatness is judged, the size of the overlapped cavity and the number of the overlapped layers need to be considered, because the number of the overlapped layers is large, the gray value of the middle hole is smaller, and therefore when the judgment is carried out according to the gray change of the surface, the more drastic the gray change is, the larger the possible degree of the abnormal condition of the surface of the fiber roll is. Therefore, when the gray level change value is larger than the preset change threshold value, the entropy of the target image is calculated according to the clustering result, and the larger the entropy is, the more drastic the surface gray level change is, and the more uneven the surface is. The calculation formula of the entropy of the target image is as follows:
wherein, the first and the second end of the pipe are connected with each other,
the entropy of the target image is represented and,
representing an initial cluster
The ratio of the number of the middle pixel points in the whole image,
represents the maximum gray value of the pixel points in the target image,
and representing the minimum gray value of the pixel points in the target image. The reason why the gray levels of the images are used as the upper and lower limits is that the clustering is performed according to the gray levels of the images, and how many initial clusters correspond to how the gray levels of the images are distributed. It should be noted that the calculation formula of the entropy is well known to those skilled in the art, and is not described herein.
Further, the surface flatness of the surface texture is obtained according to the change of the image gray scale, specifically: and calculating the surface flatness of the target image according to the entropy of the target image, the distance from the pixel points in the small clusters to the cluster centers of the small clusters and the number of the initial clusters. The surface flatness reflects the flatness of the surface texture. And calculating the flatness of the surface texture according to the entropy value of the surface texture obtained by calculation. Because in the winding process of the fiber yarns, the winding number of turns of each turn is the same as the winding rule, when the fiber yarns wound by a certain turn are changed, the flatness of the surface texture can be changed, and the gray scale of the image can be changed. The more irregular the edge, the less planar the surface will be. When the degree of the rule is calculated according to the edge, if the fiber filaments are overlapped up and down in the winding process, a convex covered wire can be formed on the surface of the coil, the gray scale of the convex covered wire can be changed, the calculated gradient can be larger, when the diamond edge points are obtained according to the gradient change of the pixel points, the diamond edge points can be marked, and therefore the calculated degree of the irregularity can be increased.
And judging the surface flatness of the surface according to the gray scale of the surface of the target image. Since the more uneven the surface, the more drastic the grey scale changes in the image. In the process of winding the fiber yarns, a small gap exists between every two thin fiber yarns, so that the overlapping phenomenon cannot occur, the gray level change is uniform, and only small gray level change can be formed between the fiber yarns. When the overlapping of the fiber yarns or the direction of the fiber texture is changed in the winding process, the gray value between the fiber yarns is reduced because the gaps between the fiber yarns are increased, and the larger the gap is, the smaller the gray value is, a gray belt with darker color is formed, and the gray belt is different from the normal texture. Therefore, the surface flatness of the surface texture is obtained according to the gray scale change of the image texture.
Because the pixel points with smaller gray levels are mainly the gaps of the fibers or the pixel points when the fibers are overlapped, after clustering, the smaller the gray level value of the clustering cluster is, the more the number of the pixel points is, the larger the gaps between the fibers are, or the fibers are overlapped. Therefore, the surface flatness is obtained according to a fitting formula of the result after the aggregation.
The calculation formula of the surface evenness is as follows:
wherein the content of the first and second substances,
surface flatness;
is a natural constant;
is the entropy of the target image;
the maximum gray value of the pixel points in the target image is obtained;
the minimum gray value of a pixel point in the target image is obtained;
the number of small clusters corresponding to the initial cluster with the gray value I;
the Euclidean distance from the ith pixel point in the small cluster with the gray value of I to the cluster center point of the small cluster; r is the number of pixels in the initial cluster with the gray value I; and I is the gray value of the pixel points in the initial cluster.
Wherein, the first and the second end of the pipe are connected with each other,
the gray scale difference of the whole image is represented, and the larger the difference is, the lower the surface flatness is. The calculation formula of the Euclidean distance is as follows:
in the formula (I), wherein,
the coordinates of the cluster center point of the small cluster are represented,
indicating the coordinates of the ith pixel point.
Expressing the average distance from each pixel point in the initial cluster to the clustering center of the small cluster, wherein the smaller the average distance is, the larger the density of the pixel points is; conversely, the larger the average distance is, the smaller the density of the pixel points is.
Indicating that the surface flatness gradually increases with decreasing average distance and vice versa. The larger the average distance, the smaller the surface flatness, the purpose of which is to change the monotonicity of the function.
Is expressed in a gray value of
In the initial clusters, under the condition that the gray value difference of the target image is larger, the larger the number of the small clusters is, the more uneven the distribution of the pixel points is, and meanwhile, the smaller the average distance is, the larger the number of the pixel points in the same small cluster is, the larger the gray value change of the pixel points in the image is, the larger the possibility of image gaps or fiber yarn overlapping is, and therefore, the smaller the flatness degree of the image is.
The smaller the gray value representing the initial cluster is, the smaller the influence of the gray value on the surface flatness is; conversely, the larger the initial cluster gray value is, the greater its effect on surface flatness is.
The average feature quantity of all the initial clusters is shown, and it should be noted that the division is not performed here to show the complete logic. By taking the entropy of the target image as a parameter reflecting the flatness of the surfaceThe reason is that in the winding process of the fiber yarns, the number of winding turns of each turn is the same as the winding rule, when the fiber yarns wound by a certain turn are changed, the flatness of the surface texture can be changed, and the gray scale of the image can be changed, so that the larger the entropy value of the target image is, the lower the corresponding surface flatness of the target image is.
It should be noted that the inference of the surface flatness formula needs to use a control variable method, and the controlled quantity is a gray value, that is, the number of initial clusters and the density of small clusters are calculated under the adjustment of the gray value.
The logic of the formula for surface flatness is: since the smaller the gray scale change during filament winding, the smaller the overlap of the filaments or the uneven spacing between the filaments. Therefore, the small clusters with different densities under different gray values obtained by the clustering algorithm are obtained by calculating the density of each small cluster to obtain the gray value of
Then the variation of the pixel points of the entire image is calculated.
By calculating the surface flatness of the target image, it is possible to indicate a case where the winding quality of the filament is not satisfactory due to overlapping of the filaments or excessive gaps between the filaments during the winding process. The smaller the surface flatness, the more drastic the gradation change of the target image, and the less standard the quality of the corresponding fiber roll. The surface flatness was used as a quality evaluation parameter for evaluating filament winding.
And S400, determining the winding quality grade of the filament winding machine according to the surface flatness.
And obtaining the surface flatness of the target image of the fiber roll wound by the fiber winding machine, wherein the larger the surface flatness is, the better the winding quality of the fiber winding machine is. According to the actual situation of the invention, a first threshold value and a second threshold value are preset, when the surface evenness is larger than the preset first threshold value, the quality grade of the fiber roll is excellent, namely the winding quality grade of the fiber winding machine is excellent; when the surface flatness is less than or equal to a preset first threshold value and greater than a preset second threshold value, the quality grade of the fiber roll is qualified, and the quality grade of the fiber winding machine is qualified; and when the surface flatness is less than or equal to the preset second threshold value, the quality grade of the fiber roll is unqualified, namely the winding quality grade of the fiber winding machine is not qualified. In the embodiment of the present invention, the value of the preset first threshold is 2.54, and the value of the preset second threshold is 1.85, and in other embodiments, the implementer may adjust the value according to the actual situation.
In summary, the present invention relates to the field of image processing technology. Firstly, acquiring a winding image of a fiber roll wound by a fiber winding machine, and preprocessing the winding image to obtain a target image; determining diamond-shaped edge points to be selected formed in the winding process according to the gray level change of the target image; determining diamond edge points according to the density of the edge points to be selected; clustering the diamond-shaped edge points according to the gray values to obtain at least two initial clusters; clustering the diamond-shaped edge points of each initial cluster based on the density to obtain at least two small clusters; selecting each small cluster in the initial cluster with the minimum gray value as an edge cluster, extending the central point of the edge cluster to two sides horizontally, and acquiring a gray value change value according to the gray value of a pixel point on an extension line; when the gray level change value is larger than a preset change threshold value, calculating the surface flatness of the target image according to the entropy of the target image, the distance from the pixel points in the small clusters to the cluster centers of the small clusters and the number of the initial clusters; and determining the winding quality grade of the filament winding machine according to the surface flatness. According to the invention, the surface flatness of the target image is calculated, so that the condition that the winding quality is not in accordance with the standard due to the overlapping of the fiber yarns or the overlarge gap between the fiber yarns in the winding process of the fiber yarns is reflected, and the winding quality detection of the fiber winding machine under the condition of complicated fiber winding is realized.
It should be noted that: the precedence order of the above embodiments of the present invention is only for description, and does not represent the merits of the embodiments. The processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.
All the embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from other embodiments.
Claims (7)
1. A winding detection method of a filament winding machine is characterized by comprising the following steps:
acquiring a winding image of a fiber roll wound by a fiber winding machine, and preprocessing the winding image to obtain a target image;
determining diamond-shaped edge points to be selected formed during winding according to the gray level change of the target image; determining diamond edge points according to the density of the edge points to be selected;
clustering the rhombic edge points according to the gray values to obtain at least two initial clusters; clustering the diamond-shaped edge points of each initial cluster based on the density to obtain at least two small clusters; selecting each small cluster in the initial cluster with the minimum gray value as an edge cluster, extending horizontally to two sides by using the central point of the edge cluster, and acquiring a gray change value according to the gray value of a pixel point on an extension line; when the gray level change value is larger than a preset change threshold value, calculating the surface flatness of the target image according to the entropy of the target image, the distance from the pixel points in the small clusters to the cluster centers of the small clusters and the number of initial clusters;
and determining the winding quality grade of the fiber winding machine according to the surface flatness.
2. The filament winding machine winding detection method according to claim 1, wherein the calculation formula of the surface flatness is as follows:
wherein, the first and the second end of the pipe are connected with each other,
is the surface flatness;
is a natural constant;
entropy of the target image; the maximum gray value of a pixel point in the target image is obtained;
the minimum gray value of a pixel point in the target image is obtained;
the number of small clusters corresponding to the initial cluster with the gray value I;
the Euclidean distance from the ith pixel point in the small cluster with the gray value of I to the cluster center point of the small cluster; r is the number of pixels in the initial cluster with the gray value I; and I is the gray value of the pixel points in the initial cluster.
3. The method for detecting the winding of the filament winding machine according to claim 1, wherein the determining of the diamond-shaped edge points to be selected during the winding according to the gray scale change of the target image comprises:
and calculating the gradient amplitude of each pixel point in the target image, and taking the pixel point with the gradient amplitude larger than a preset gradient threshold value as a to-be-selected edge point.
4. The method for detecting the winding of the filament winding machine according to claim 1, wherein the step of determining the diamond-shaped edge points according to the density of the edge points to be selected comprises the following steps:
calculating the number ratio of pixel points with the same gradient amplitude value as the edge point to be selected in the eight neighborhoods of the edge point to be selected; and taking the pixel points with the number ratio larger than a preset ratio threshold value as diamond edge points.
5. The method for detecting the winding of the filament winding machine according to claim 1, wherein the step of extending the center point of the edge cluster horizontally to both sides and obtaining the gray variation value according to the gray value of the pixel point on the extension line comprises:
and acquiring the gray value difference value of each pixel point on the extension line and the central point of the edge cluster, wherein the sum of the gray value difference values corresponding to all the pixel points on the extension line is the gray change value.
6. The filament winding machine winding detection method according to claim 1, wherein the determining of the quality grade of the filament winding machine winding according to the surface flatness comprises:
when the surface flatness is larger than a preset first threshold value, the winding quality grade of the filament winding machine is excellent; when the surface flatness is smaller than or equal to a preset first threshold value and larger than a preset second threshold value, the winding quality grade of the fiber winding machine is qualified; and when the surface flatness is less than or equal to a preset second threshold value, the winding quality grade of the filament winding machine is unqualified.
7. The method for detecting the winding of the fiber winding machine according to claim 1, wherein the preprocessing the winding image to obtain a target image comprises:
graying the winding image to obtain a gray image, and performing semantic segmentation on the gray image to obtain a target image only containing the fiber roll.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211429710.7A CN115496751B (en) | 2022-11-16 | 2022-11-16 | Winding detection method of fiber winding machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211429710.7A CN115496751B (en) | 2022-11-16 | 2022-11-16 | Winding detection method of fiber winding machine |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115496751A true CN115496751A (en) | 2022-12-20 |
| CN115496751B CN115496751B (en) | 2023-03-24 |
Family
ID=85115734
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211429710.7A Active CN115496751B (en) | 2022-11-16 | 2022-11-16 | Winding detection method of fiber winding machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115496751B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018197579A1 (en) * | 2017-04-25 | 2018-11-01 | Muhr Und Bender Kg | Method and device for ascertaining the flatness of strip material, and machining system comprising such a device |
| WO2021254381A1 (en) * | 2020-06-17 | 2021-12-23 | 京东方科技集团股份有限公司 | Image processing method and apparatus, electronic device, and computer-readable storage medium |
| CN115082445A (en) * | 2022-07-25 | 2022-09-20 | 山东鲁泰防水科技有限公司 | Method for detecting surface defects of building waterproof roll |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113724253A (en) * | 2021-10-13 | 2021-11-30 | 南通保利金纺织科技有限公司 | Spinning carding process quality monitoring method based on image processing |
| CN115147416B (en) * | 2022-09-02 | 2022-11-15 | 山东大山不锈钢制品有限公司 | Rope disorder detection method and device for rope rewinder and computer equipment |
-
2022
- 2022-11-16 CN CN202211429710.7A patent/CN115496751B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018197579A1 (en) * | 2017-04-25 | 2018-11-01 | Muhr Und Bender Kg | Method and device for ascertaining the flatness of strip material, and machining system comprising such a device |
| WO2021254381A1 (en) * | 2020-06-17 | 2021-12-23 | 京东方科技集团股份有限公司 | Image processing method and apparatus, electronic device, and computer-readable storage medium |
| CN115082445A (en) * | 2022-07-25 | 2022-09-20 | 山东鲁泰防水科技有限公司 | Method for detecting surface defects of building waterproof roll |
Non-Patent Citations (1)
| Title |
|---|
| 陈鹿民等: "图像技术在包芯线自动卷绕机械中的应用", 《微计算机信息》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115496751B (en) | 2023-03-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114897894B (en) | Method for detecting defects of cheese chrysanthemum core | |
| CN114862862B (en) | Pump body cold shut defect identification method and system based on image processing | |
| CN114723701A (en) | Gear defect detection method and system based on computer vision | |
| CN113989280A (en) | Steel structure welding crack defect detection method based on image processing technology | |
| CN116310360A (en) | Reactor surface defect detection method | |
| CN114049489A (en) | Early warning method for breakage of optical fiber cable sheath | |
| CN114926436A (en) | Defect detection method for periodic pattern fabric | |
| CN113638104B (en) | Intelligent yarn cleaning control method and system for bobbin winder | |
| CN113724241B (en) | Broken filament detection method and device for carbon fiber warp-knitted fabric and storage medium | |
| CN117095009B (en) | PVC decorative plate defect detection method based on image processing | |
| CN115496751B (en) | Winding detection method of fiber winding machine | |
| CN115719344A (en) | Fabric defect analysis and identification method for textile fabric quality inspection | |
| CN114842016A (en) | Fire hose defect detection method and system based on computer vision | |
| CN115100210B (en) | Anti-counterfeiting identification method based on automobile parts | |
| CN117422712B (en) | Plastic master batch visual detection method and system based on image filtering processing | |
| CN114998321A (en) | Textile material surface hairiness degree identification method based on optical means | |
| CN116843627A (en) | Broken yarn detection method, device, equipment and storage medium for glass fiber cloth cover binding yarns | |
| CN114782425A (en) | Spooling process parameter control method and artificial intelligence system in textile production process | |
| CN117036357B (en) | Profile steel surface defect detection method based on visual characteristics | |
| CN114789927B (en) | Artificial intelligent control method and system for textile fabric gray cloth winding machine | |
| CN115082453A (en) | Intelligent control method of edge covering machine for bedding production based on neural network | |
| CN116385416B (en) | Intelligent control method and system for high-speed braiding machine | |
| CN116740056B (en) | Defect detection method for coating layer of whole-core high-pattern conveyer belt | |
| CN115187601B (en) | Deep sea fishing net production process defect detection method based on image understanding | |
| CN114926411A (en) | Ribbon wave defect detection method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |