阅读说明：本技术 一种卷装长丝的成型不良缺陷检测方法 (Method for detecting poor forming defect of coiled filament ) 是由 冯培 侯曦 杨崇倡 肖凌云 张俊平 于 2018-08-24 设计创作，主要内容包括：本发明公开了一种卷装长丝的成型不良缺陷检测方法,包括：获取包含卷装长丝的第一锥面的多张采集图像；对采集图像进行去噪处理,并过滤去噪后的采集图像中的背景图案,获得连续长曲线；确定连续长曲线上的多个曲率突变点以及每两个相邻的曲率突变点之间的曲线极值点；计算每两个相邻的曲率突变点之间的突变点连线的连线长度和曲线极值点到突变点连线之间的极值点距离；根据连线长度和极值点距离,确定卷装长丝是否存在成型不良缺陷。本发明实施例的卷装长丝的成型不良缺陷检测方法,能够降低人工目测的错误率,提高检测效率,且能够节约生产成本。(The invention discloses a method for detecting poor forming defects of a coiled filament, which comprises the following steps: acquiring a plurality of collected images of a first conical surface containing package filaments; denoising the acquired image, and filtering a background pattern in the denoised acquired image to obtain a continuous long curve; determining a plurality of curvature abrupt change points on the continuous long curve and curve extreme points between every two adjacent curvature abrupt change points; calculating the connecting line length of a mutation point connecting line between every two adjacent curvature mutation points and the distance from a curve extreme point to the extreme point between the mutation point connecting lines; and determining whether the packaged filament has poor forming defects according to the connecting line length and the extreme point distance. The method for detecting the defect of poor forming of the packaged filament can reduce the error rate of manual visual inspection, improve the detection efficiency and save the production cost.)

1. A method for detecting defective molding of a wound filament yarn, comprising:

acquiring a plurality of collected images of a first cone surface containing the package of filaments;

denoising the acquired image, and filtering a background pattern in the denoised acquired image to obtain a continuous long curve;

determining a plurality of curvature abrupt points on the continuous long curve and curve extreme points between every two adjacent curvature abrupt points;

calculating the connecting line length of a catastrophe point connecting line between every two adjacent curvature catastrophe points and the distance from the curve extreme point to the extreme point between the catastrophe point connecting lines;

and determining whether the packaged filament has poor forming defects or not according to the connecting line length and the extreme point distance.

2. A method of detecting poor formation of a wound filament yarn as in claim 1 wherein said capturing an image for de-noising comprises:

intercepting a target image corresponding to the first conical surface in the collected image;

and denoising the target image by adopting a median filtering method, and obtaining the denoised acquisition image.

3. A method of detecting faulty molding of a wound filament according to claim 2, wherein said intercepting a target image corresponding to said first cone in said collected image comprises:

acquiring a copy image of the acquired image;

denoising the copied image by a Gaussian filtering method;

acquiring a plurality of edge lines of the first conical surface in the denoised copy image by using an image segmentation algorithm;

and acquiring coordinate values of all pixel points of the plurality of edge lines, and intercepting the target image in the acquired image according to the coordinate values of all pixel points of the plurality of edge lines.

4. A method of detecting poor formation of a wound filament according to claim 3, wherein obtaining the plurality of edge lines of the first cone in the de-noised copy image using an image segmentation algorithm comprises:

acquiring a plurality of edges in the denoised copy image by using an image segmentation algorithm;

determining a first edge line, a second edge line and a third edge line of the first conical surface according to the plurality of edge lines, and determining a first intersection point of the first edge line and the second edge line and a second intersection point of the first edge line and the third edge line;

determining a reference line according to the first intersection point and the second intersection point, and rotating the copied image to enable the reference line to be horizontally arranged;

and determining coordinate values of all pixel points of the rotated first intersection point, the rotated second intersection point and the rotated first edge line, and determining a plurality of edge lines of the first conical surface.

5. A method of detecting poor formation of a wound filament according to claim 1 wherein filtering the background pattern in the de-noised captured image to obtain a continuous length curve comprises:

filtering the background pattern of the de-noised collected image by using a binarization method;

and filling a region which is smaller than a preset area between every two adjacent first patterns in the acquired image after the background patterns are filtered by using an expansion processing method to obtain the continuous long curve.

6. A method of detecting defective molding of a wound filament according to claim 1, wherein determining whether or not a defective molding of the wound filament exists based on the connection length and the extreme point distance comprises:

determining whether each two adjacent curvature mutation points are mutation areas or not according to the length of the connecting line and the distance between the extreme points;

determining the mutation condition of the corresponding acquired image according to the mutation area condition in each acquired image;

and determining whether the packaged filament has poor forming defects according to the mutation conditions of the plurality of collected images.

7. A method of detecting poor formation of a wound filament according to claim 6 wherein determining whether each two adjacent curvature discontinuities are discontinuities based on said link length and said extreme distance comprises:

judging whether two adjacent curvature mutation points of the continuous long curve meet the following conditions: the length of the connecting line is more than or equal to 5mm, the distance between the extreme points is more than or equal to 1mm, and if the distance between the extreme points is more than or equal to 1mm, the two adjacent curvature mutation points are determined to be A-level mutation areas; or

Judging whether two adjacent curvature mutation points of the continuous long curve meet the following conditions: the length of the connecting line is more than or equal to 2mm, the distance between the extreme points is more than or equal to 3mm, and if the distance between the extreme points is more than or equal to 3mm, the two adjacent curvature mutation points are determined to be A-level mutation areas; or

Judging whether two adjacent curvature mutation points of the continuous long curve meet the following conditions: the length of the connecting line is less than 5mm in a range of 2mm, the distance of the extreme point is less than 3mm in a range of 1mm, and if the distance of the extreme point and the length of the connecting line meet the requirement, whether the product of the distance of the extreme point and the length of the connecting line meets the requirement is judged: and the product is more than or equal to 6, and if the product is satisfied, the two adjacent curvature mutation points are determined to be A-level mutation areas.

8. A method of detecting poor formation of a wound filament according to claim 7, wherein determining whether each two adjacent curvature discontinuities are abrupt changes according to the connection length and the extreme point distance after determining that two adjacent curvature discontinuities of the continuous length curve are class a abrupt changes comprises:

judging whether two adjacent curvature mutation points of the continuous long curve meet the following conditions: the length of the connecting line is more than or equal to 8mm, the distance between the extreme points is more than or equal to 1mm, and if the distance between the extreme points is more than or equal to 1mm, the two adjacent curvature mutation points are determined to be B-level mutation areas; or

Judging whether two adjacent curvature mutation points of the continuous long curve meet the following conditions: the length of the connecting line is more than or equal to 2mm, the distance between the extreme points is more than or equal to 4mm, and if the distance between the extreme points is more than or equal to 4mm, the two adjacent curvature mutation points are determined to be B-level mutation areas; or

Judging whether two adjacent curvature mutation points of the continuous long curve meet the following conditions: the length of the connecting line is less than 8mm in a range of 2mm, the distance of the extreme point is less than 4mm in a range of 1mm, and if the distance of the extreme point and the length of the connecting line meet the requirement, whether the product of the distance of the extreme point and the length of the connecting line meets the requirement is judged: and the product is more than or equal to 9, and if the product is satisfied, the two adjacent curvature mutation points are determined to be B-level mutation areas.

9. A method of detecting faulty molding of a wound filament according to claim 8, wherein determining a sudden change in each of said collected images based on the condition of a sudden change region in said collected image comprises:

determining the mutation condition corresponding to the acquired image according to the conditions of the A-level mutation region and the B-level mutation region in any acquired image;

the abrupt change condition is that the collected image is an A-level degraded picture, the collected image is a B-level degraded picture or the collected image has no abrupt change.

10. A method of detecting defective molding of a wound filament according to claim 9, wherein determining whether or not the wound filament has defective molding based on a sudden change in the plurality of collected images comprises:

determining whether the packaged filament has poor forming defects according to the number of the collected pictures corresponding to different mutation conditions;

and if the packaged filament has poor forming defects, determining the poor forming grade of the packaged filament.

Technical Field

The invention relates to the technical field of surface detection of packaged filaments, in particular to a method for detecting poor forming defects of the packaged filaments.

Background

The coiled filament is a coiled product with a certain shape and capacity which is made by a winding mechanism in the production process of the filament. The defects of the wound filament are mainly divided into two parts of physical and chemical property defects of the filament and appearance defects of the wound filament. It has been found through practice that appearance defects in the wound filaments have a significant impact on the quality of the fabric, resulting in a reduction in fabric yield. Therefore, in the production process, the detection of appearance defects of the wound filaments is intensified.

The defective molding of the wound filament refers to a defect in which the shape of the filament is deformed abnormally after the filament is wound into a wound product. Poor formation of the wound filament is caused by a failure of the winding equipment or an improper tension at the time of winding, and generally occurs symmetrically at a first tapered surface at the upper end and a second tapered surface at the lower end of the wound filament. The poor formation of the wound filament has a great influence on the beauty of the wound filament, and the subsequent steps of unwinding, dyeing and the like are mainly influenced in the production process. Therefore, the factory has a high demand for detecting defects of poor molding.

Because the coiled filament has the characteristics of deformability, multiple curved surfaces and large detected surface, the appearance defect characteristics of the coiled filament are difficult to extract unified standards, the appearance detection of the coiled filament always influences the realization of the intelligent production of a long production flow, and the defect detection of poor forming can only be carried out by a manual visual inspection method.

Although some defects can be detected by detecting the molding defects through manual visual inspection, the manual visual inspection method increases the labor cost of production, and does not have a uniform and strict quality standard, so that the accuracy of the detection result is difficult to ensure.

Disclosure of Invention

In order to solve the above problems, embodiments of the present invention provide a method for detecting a defective molded wound filament, which can reduce an error rate of manual visual inspection, improve detection efficiency, and save production cost.

In order to achieve the above object, an embodiment of the present invention provides a method for detecting a defective formation of a wound filament, including:

acquiring a plurality of collected images of a first conical surface containing package filaments;

denoising the acquired image, and filtering a background pattern in the denoised acquired image to obtain a continuous long curve;

determining a plurality of curvature abrupt change points on the continuous long curve and curve extreme points between every two adjacent curvature abrupt change points;

calculating the connecting line length of a mutation point connecting line between every two adjacent curvature mutation points and the distance from a curve extreme point to the extreme point between the mutation point connecting lines;

and determining whether the packaged filament has poor forming defects according to the connecting line length and the extreme point distance.

Further, the acquiring the image for denoising includes:

intercepting a target image corresponding to the first conical surface in the collected image;

and denoising the target image by adopting a median filtering method, and obtaining a denoised acquired image.

Further, intercepting the target image corresponding to the first cone in the acquired image comprises:

acquiring a copy image of the acquired image;

denoising the copied image by a Gaussian filtering method;

acquiring a plurality of edge lines of a first conical surface in the denoised copy image by using an image segmentation algorithm;

and acquiring coordinate values of all pixel points of the plurality of edge lines, and intercepting the target image in the acquired image according to the coordinate values of all pixel points of the plurality of edge lines.

Further, the obtaining the plurality of edge lines of the first cone in the denoised copy image by using an image segmentation algorithm includes:

acquiring a plurality of edges in the denoised copy image by using an image segmentation algorithm;

determining a first edge line, a second edge line and a third edge line of the first conical surface according to the plurality of edge lines, and determining a first intersection point of the first edge line and the second edge line and a second intersection point of the first edge line and the third edge line;

determining a reference line according to the first intersection point and the second intersection point, and rotating the copied image to enable the reference line to be horizontally arranged;

and determining coordinate values of all pixel points of the rotated first intersection point, the rotated second intersection point and the rotated first edge line, and determining a plurality of edge lines of the first conical surface.

Further, filtering the background pattern in the de-noised acquired image to obtain a continuous long curve includes:

filtering the background pattern of the de-noised acquired image by using a binarization method;

and filling a region smaller than a preset area between every two adjacent first patterns in the acquired image after the background patterns are filtered by using an expansion processing method to obtain a continuous long curve.

Further, determining whether the packaged filament has the poor forming defect according to the connecting line length and the extreme point distance comprises the following steps:

determining whether each two adjacent curvature mutation points are mutation areas or not according to the length of the connecting line and the distance between the extreme points;

determining the mutation condition of the corresponding acquired image according to the mutation area condition in each acquired image;

and determining whether the packaged filament has poor forming defects according to the mutation conditions of the plurality of collected images.

Further, determining whether each two adjacent curvature mutation points are mutation regions according to the connecting line length and the extreme point distance comprises:

judging whether two adjacent curvature mutation points of the continuous long curve meet the following conditions: the length of the connecting line is more than or equal to 5mm, the distance between the extreme points is more than or equal to 1mm, and if the distance between the extreme points is met, two adjacent curvature mutation points are determined to be A-level mutation areas; or

Judging whether two adjacent curvature mutation points of the continuous long curve meet the following conditions: the length of the connecting line is more than or equal to 2mm, the distance between the extreme points is more than or equal to 3mm, and if the distance between the extreme points is more than or equal to 3mm, two adjacent curvature mutation points are determined to be A-level mutation areas; or

Judging whether two adjacent curvature mutation points of the continuous long curve meet the following conditions: the length of the connecting line is more than 2mm and less than 5mm, the distance of the extreme point is more than 1mm and less than 3mm, if the distance satisfies, whether the product of the distance of the extreme point and the length of the connecting line satisfies: and if the product is more than or equal to 6, determining two adjacent curvature mutation points as A-level mutation areas.

Further, after determining that two adjacent curvature mutation points of the continuous long curve are a class a mutation region, determining whether each two adjacent curvature mutation points are mutation regions according to the connection line length and the extreme point distance includes:

judging whether two adjacent curvature mutation points of the continuous long curve meet the following conditions: the length of the connecting line is more than or equal to 8mm, the distance between the extreme points is more than or equal to 1mm, and if the distance between the extreme points is met, two adjacent curvature mutation points are determined to be B-level mutation areas; or

Judging whether two adjacent curvature mutation points of the continuous long curve meet the following conditions: the length of the connecting line is more than or equal to 2mm, the distance between the extreme points is more than or equal to 4mm, and if the distance between the extreme points is more than or equal to 4mm, two adjacent curvature mutation points are determined to be B-level mutation areas; or

Judging whether two adjacent curvature mutation points of the continuous long curve meet the following conditions: the length of the connecting line is more than 2mm and less than 8mm, the distance of the extreme point is more than 1mm and less than 4mm, if the distance of the extreme point and the length of the connecting line meet, whether the product of the distance of the extreme point and the length of the connecting line meets the following requirements is judged: and the product is more than or equal to 9, and if the product meets the requirement, two adjacent curvature mutation points are determined to be B-level mutation areas.

Further, determining the abrupt change condition of the corresponding acquired image according to the condition of the abrupt change region in each acquired image comprises:

determining the mutation condition corresponding to the acquired image according to the conditions of the A-level mutation region and the B-level mutation region in any acquired image;

the abrupt change condition is that the collected image is an A-level degraded image, the collected image is a B-level degraded image or the collected image has no abrupt change.

Further, the step of determining whether the packaged filament has poor forming defects according to the mutation conditions of the plurality of collected images comprises the following steps:

determining whether the packaged filament has the defect of poor forming according to the number of the collected pictures corresponding to different mutation conditions;

if the wound filament has a defective molding, the defective molding grade of the wound filament is determined.

The method for detecting the poor forming defects of the coiled filaments can be applied to a detection device on a production line, the coiled filaments passing through each detection station are subjected to image acquisition to obtain corresponding acquired images, then the images are processed to extract defect characteristics (namely the connection length of a mutation point connection line between every two adjacent curvature mutation points on a continuous long curve and the distance from a curve extreme point to an extreme point connection line) for judging the poor forming defects, and whether the poor forming defects exist in the coiled filaments is determined according to the properties of the defect characteristics of different poor forming defects.

The method for detecting the poor forming defects of the coiled filaments is suitable for detecting the poor forming defects of the coiled filaments in the chemical fiber field, can quickly analyze the appearance defects of the coiled filaments, reduces errors of manual visual inspection, reduces false detection rate, improves detection efficiency and reduces production cost through high-precision analysis and processing of images.

Drawings

FIG. 1 is a flowchart of a method for detecting defective formation of a wound filament according to an embodiment of the present invention;

FIG. 2 is a schematic structural view of a wound filament according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of an acquired image according to an embodiment of the invention;

FIG. 4 is a flowchart illustrating a specific method of step S120 in one embodiment of FIG. 1;

FIG. 5 is a flowchart illustrating a specific method of step S121 in FIG. 4;

FIG. 6 is a flowchart illustrating a specific method of step S230 in FIG. 5;

FIG. 7 is a schematic diagram of a target image according to an embodiment of the invention;

FIG. 8 is a flowchart illustrating a specific method of step S120 according to another embodiment of FIG. 1;

fig. 9 is a flowchart illustrating a specific method of step S150 in fig. 1.

Detailed Description

The structure, operation principle, and the like of the embodiments of the present invention will be further described with reference to the accompanying drawings.

As shown in fig. 1, a method for detecting a defective formation of a wound filament according to an embodiment of the present invention includes:

s110, acquiring a plurality of collected images of the first conical surface 1 containing the coiled filaments;

s120, denoising the acquired image, and filtering a background pattern in the denoised acquired image to obtain a continuous long curve;

s130, determining a plurality of curvature mutation points on the continuous long curve and curve extreme points between every two adjacent curvature mutation points;

s140, calculating the connecting line length of a mutation point connecting line between every two adjacent curvature mutation points and the distance from a curve extreme point to the extreme point between the mutation point connecting lines;

s150, determining whether the packaged filament has poor forming defects according to the connecting line length and the extreme point distance.

In the embodiment of the present invention in which the wound filament is to be inspected, as shown in fig. 2, the molding defect may occur simultaneously on the first tapered surface 1 at the upper end and the second tapered surface 2 at the lower end of the wound filament. At this time, any one of the first tapered surface 1 and the second tapered surface 2 may be selected for detection to reduce the workload. In the embodiment of the present invention, the detection of the first tapered surface 1 will be described as an example.

Therefore, an area-array camera may be disposed facing the first tapered surface 11 and may be able to acquire a captured image as shown in fig. 3. The area-array camera can reduce the shooting times and save time. And according to the characteristics of fixed position and limited shooting angle of the area-array camera, the coiled filament can be rotated, and the collected image containing the first conical surface 1 is collected while the coiled filament rotates for one circle, so that the image collection of the whole circle of the coiled filament is completed. In one embodiment, the photographing time may be 2.2s, and 20 acquired images of the region to be detected are acquired at equal time intervals within the photographing time.

Because the defect characteristic of the defects can be highlighted by using low-angle illumination, which is higher than the defect of poor molding of the surface of the detected object, in the embodiment of the invention, the low-angle illumination light source is adopted for illuminating the area to be detected, and the acquisition directions of the light source light ray area-array cameras are the same.

In the embodiment of the present invention, as shown in fig. 4, the acquiring an image for denoising includes:

s121, intercepting a target image corresponding to the first conical surface 1 in the collected image;

and S122, denoising the target image by adopting a median filtering method, and obtaining a denoised acquired image.

In the embodiment of the present invention, the noise point in the target image may be analyzed first, and then the target image may be filtered according to the existing noise point. Since the noise points of the target image are concentrated in the high-frequency region, the filtering process may employ a median filtering method. The median filtering has a stable amplitude characteristic, and can obtain an image for eliminating isolated noise points, so that the deterioration of the imaging quality of the image caused by various interference factors in the process of acquiring an original signal, transmitting the original signal to signal conversion and signal transmission of the area-array camera is improved.

In the embodiment of the present invention, as shown in fig. 5, intercepting the target image corresponding to the first cone 1 in the captured image includes:

s210, acquiring a copy image of the acquired image;

s220, denoising the copied image by using a Gaussian filtering method;

s230, acquiring a plurality of edge lines of the first conical surface 1 in the denoised copy image by using an image segmentation algorithm;

s240, obtaining coordinate values of all pixel points of the plurality of edge lines, and intercepting the target image in the collected image according to the coordinate values of all pixel points of the plurality of edge lines.

In the embodiment of the invention, the interference of a non-processing object in the image can be eliminated by intercepting the target image in the acquired image. By adopting the steps, the situation that the position of the first conical surface 1 is deviated or the area range of the first conical surface 1 is different can be processed, and the image capturing precision is improved.

In the embodiment of the present invention, as shown in fig. 6, the obtaining, by using an image segmentation algorithm, a plurality of edge lines of the first cone 1 in the denoised copy image includes:

s231, acquiring a plurality of edges in the denoised copy image by using an image segmentation algorithm;

s232, determining a first edge line, a second edge line and a third edge line of the first conical surface 1 according to the plurality of edge lines, and determining a first intersection point of the first edge line and the second edge line and a second intersection point of the first edge line and the third edge line;

s233, determining a reference line according to the first intersection point and the second intersection point, and rotating the copied image to enable the reference line to be horizontally arranged;

and S234, determining coordinate values of all pixel points of the rotated first intersection point, the rotated second intersection point and the rotated first edge line, and determining a plurality of edge lines of the first conical surface 1.

In the embodiment of the invention, the acquired image can be copied firstly, the copied image can be obtained, and the edge line can be obtained by processing the copied image, so that the influence on the subsequent processing of the original acquired image caused by processing the original acquired image can be prevented.

Then, denoising processing may be performed on the copied image through a filtering method, and in this embodiment, the filtering method may preferably be gaussian filtering. The denoised copy image can obtain a plurality of edge lines through an image segmentation algorithm. The image segmentation algorithm can be an edge detection algorithm or a binarization method, and the sideline can be a sudden change position of gray level brightness change of the edge of the first conical surface 1 of the coiled filament.

Next, an edge closest to the top of the image in the copied image may be selected as a first edge 13, an edge closest to the left of the image in the copied image may be selected as a second edge 11, an edge closest to the right of the image in the copied image may be selected as a third edge 12, and a first intersection 14 of the first edge 13 and the second edge 11 and a second intersection 15 of the first edge 13 and the third edge 12 are determined, as shown in fig. 3, and at the same time, coordinates (x1, y1) of the first intersection 14 and coordinates (x2, y2) of the second intersection 12 are obtained.

Again, reference line 16 is obtained by connecting first intersection 14 and second intersection 15. An angle phi between the reference line 16 and the horizontal reference line is measured, and the copied image is rotated about the first intersection 14 as a rotation axis based on the reference line angle phi, so that the reference line 16 is horizontally disposed. At this time, the coordinates of the second intersection point 15 become (x3, y 1).

Then, an average y coordinate value of all the pixels of the rotated first edge 13 is obtained, an edge line coordinate y3 at the top end of the first edge 13 is obtained by adding the average y coordinate value to a preset coordinate difference value, and an edge line coordinate y4 at the bottom end of the first edge 13 is obtained by subtracting the preset coordinate difference value from the average y coordinate value. Therefore, the left edge line x is x1, the right edge line x is x3, the upper edge line y is y3, and the lower edge line y is the coordinate values of all the pixels in the copied image in y 4.

Finally, the coordinate values of all the pixel points of the edge line in the copied image can be brought into the acquired image, the same pixel point coordinate values of the acquired image are extracted, a plurality of segmentation edge lines of the acquired image can be obtained by using the extracted coordinate values, wherein the segmentation edge lines comprise a left edge line, a right edge line, an upper edge line and a lower edge line, and then the image in the segmentation edge lines is captured to obtain the target image shown in fig. 7.

In the embodiment of the present invention, as shown in fig. 8, filtering the background pattern in the de-noised captured image to obtain the continuous long curve includes:

s123, filtering the background pattern of the denoised acquired image by using a binarization method;

and S124, filling a region with an area smaller than a preset area between every two adjacent first patterns in the acquired image after the background patterns are filtered by using an expansion processing method to obtain a continuous long curve.

Specifically, the edge detection can be performed by using a Prewitt operator + logarithmic form calculation or a Roberts operator + logarithmic form, so as to obtain a plurality of edges. And then, according to the gray difference of upper, lower, left and right adjacent points of the pixel point, an extreme value detection edge is achieved at the edge, part of pseudo edges are removed, the image enhancement effect is obvious, the original details of the acquired image after noise are not lost, and the image enhancement effect is optimal.

In the embodiment of the invention, the target image after the edge detection processing can be subjected to binarization processing according to a preset optimal threshold value, so that an image with only red and black colors is obtained, and the image after the binarization processing is filtered according to preset length and area parameters so as to remove background interference. Wherein, the optimal threshold value can be set to 79-83, and the area parameter can be filtered out to be less than or equal to 0.05mm²And a preset red area with the length less than or equal to 0.5 mm.

In the embodiment of the invention, every two adjacent first patterns are connected by using an expansion processing method to obtain a continuous long curve and a plurality of short lines, so that unnecessary interference on the acquired image after the background pattern is filtered can be avoided. Specifically, the image to be analyzed may be divided into a plurality of blocks, and then the red color image may be connected according to a red image of a predetermined area to obtain a continuous long curve and a plurality of short lines. Because the red image with the preset area can effectively fill the area which is smaller than the preset area between every two adjacent first patterns, the boundary of the line can be smoothed under the condition that the line is obviously not changed, and the subsequent extraction of the geometric parameters of the line is convenient.

Since only the characteristics of the continuous long curve need to be studied in the embodiment of the present invention, the short line can be removed. Specifically, short lines with area parameter less than or equal to 0.5mm2 and preset length less than or equal to 5mm can be filtered to remove the interference of all background patterns, and only continuous long curves are reserved.

In the embodiment of the present invention, as shown in fig. 9, the determining whether the packaged filament has the poor forming defect according to the connection line length and the extreme point distance includes:

s151, determining whether each two adjacent curvature mutation points are mutation areas or not according to the connecting line length and the extreme point distance;

s152, determining the mutation condition of the corresponding acquired image according to the mutation area condition in each acquired image;

and S153, determining whether the packaged filament has poor forming defects according to the mutation conditions of the plurality of collected images.

As shown in fig. 7, the two adjacent curvature discontinuities are curvature discontinuities 17 and 18, respectively, and the curve extreme point between the curvature discontinuities 17 and 18 is the curve highest point between the two adjacent curvature discontinuities.

In the embodiment of the present invention, determining whether each two adjacent curvature mutation points are mutation regions according to the connection line length and the extreme point distance may include:

judging whether two adjacent curvature mutation points of the continuous long curve meet the following conditions: the length of the connecting line is more than or equal to 5mm, the distance between the extreme points is more than or equal to 1mm, and if the distance between the extreme points is met, two adjacent curvature mutation points are determined to be A-level mutation areas; or

Judging whether two adjacent curvature mutation points of the continuous long curve meet the following conditions: the length of the connecting line is more than or equal to 2mm, the distance between the extreme points is more than or equal to 3mm, and if the distance between the extreme points is more than or equal to 3mm, two adjacent curvature mutation points are determined to be A-level mutation areas; or

Judging whether two adjacent curvature mutation points of the continuous long curve meet the following conditions: the length of the connecting line is more than 2mm and less than 5mm, the distance of the extreme point is more than 1mm and less than 3mm, if the distance satisfies, whether the product of the distance of the extreme point and the length of the connecting line satisfies: the product is more than or equal to 6, if the product meets the requirement, two adjacent curvature mutation points are determined to be A-level mutation areas;

if two adjacent curvature mutation points of the continuous long curve meet the condition, the following judgment is continued.

In the embodiment of the present invention, after determining that two adjacent curvature mutation points of the continuous long curve are a class a mutation region, determining whether each two adjacent curvature mutation points are a mutation region according to the connection line length and the extreme point distance may further include:

judging whether two adjacent curvature mutation points of the continuous long curve meet the following conditions: the length of the connecting line is more than or equal to 8mm, the distance between the extreme points is more than or equal to 1mm, and if the distance between the extreme points is met, two adjacent curvature mutation points are determined to be B-level mutation areas; or

Judging whether two adjacent curvature mutation points of the continuous long curve meet the following conditions: the length of the connecting line is more than or equal to 2mm, the distance between the extreme points is more than or equal to 4mm, and if the distance between the extreme points is more than or equal to 4mm, two adjacent curvature mutation points are determined to be B-level mutation areas; or

Judging whether two adjacent curvature mutation points of the continuous long curve meet the following conditions: the length of the connecting line is more than 2mm and less than 8mm, the distance of the extreme point is more than 1mm and less than 4mm, if the distance of the extreme point and the length of the connecting line meet, whether the product of the distance of the extreme point and the length of the connecting line meets the following requirements is judged: the product is more than or equal to 9, if the product meets the requirement, two adjacent curvature mutation points are determined to be B-level mutation areas;

and if the two adjacent curvature mutation points of the continuous long curve do not meet all the conditions, the two adjacent curvature mutation points of the continuous long curve are normal, and the two adjacent curvature mutation points of the continuous long curve are not considered when the subsequent processing is carried out.

In the embodiment of the present invention, determining the abrupt change condition of the corresponding acquired image according to the abrupt change region condition in each acquired image includes:

determining the mutation condition corresponding to the acquired image according to the conditions of the A-level mutation region and the B-level mutation region in any acquired image;

the abrupt change condition is that the collected image is an A-level degraded image, the collected image is a B-level degraded image or the collected image has no abrupt change.

In one embodiment of the invention, if a single acquired image contains a B-level mutation region, the acquired image is a B-level degraded picture regardless of whether the acquired image contains an A-level mutation region or not; if a single collected image does not have a B-level mutation region but has an A-level mutation region, the collected image is an A-level degradation picture; and if the single collected image does not contain the B-level mutation region and the A-level mutation region, the collected image is a normal image.

In the embodiment of the invention, the step of determining whether the packaged filament has poor forming defects according to the mutation conditions of the plurality of collected images comprises the following steps:

determining whether the packaged filament has the defect of poor forming according to the number of the collected pictures corresponding to different mutation conditions;

if the wound filament has a defective molding, the defective molding grade of the wound filament is determined. In one embodiment of the present invention, taking 20 collected images as an example, it is determined whether the wound filament has a molding defect and a molding defect grade of the molding defect. If the number of the B-grade degraded pictures is less than 3 pictures and the number of the A-grade degraded pictures +1.5 + the number of the B-grade degraded pictures is less than 6 pictures in 20 collected images, the winding filament can be determined to have no poor forming defects, otherwise, the winding filament has poor forming defects. When the packaged filament is determined to have poor molding, the poor molding grade can be continuously judged. If the number of the B-level degradation pictures is more than or equal to 3, the package of the filament is a B-level poor forming defect; if the number of B-grade degraded pictures is less than 3, and the number of A-grade degraded pictures +1.5 × the number of B-grade degraded pictures is more than or equal to 6, the winding filament is the poor forming defect of the A-grade.

According to the method for detecting the poor forming defect of the packaged filament, disclosed by the embodiment of the invention, the rate of the poor forming defect can be correctly detected to reach 98.2%, the efficiency is 4 seconds per ingot, the method is superior to a manual visual inspection method, the online automatic detection of the appearance quality of the packaged filament is realized, and the manual workload is reduced by 95%.

In summary, the method for detecting defective molding of a wound filament yarn of the present invention can be widely applied to online detection of defective molding of a wound filament yarn in the chemical fiber field, and is easy to analyze an appearance defect. In addition, the molding defect is detected by a method for extracting and judging the molding defect characteristics, the defect grade can be automatically identified, the detection precision can be improved, and errors caused by manual visual inspection can be reduced. Meanwhile, in the image processing process, interference can be eliminated, the defect of poor forming can be rapidly judged, and the false detection rate is reduced.

The foregoing is merely illustrative of the present invention, and it will be appreciated by those skilled in the art that various modifications may be made without departing from the principles of the invention, and the scope of the invention is to be determined accordingly.