CN109825944B - Online fabric weaving defect detection method based on line laser - Google Patents
Online fabric weaving defect detection method based on line laser Download PDFInfo
- Publication number
- CN109825944B CN109825944B CN201910034418.7A CN201910034418A CN109825944B CN 109825944 B CN109825944 B CN 109825944B CN 201910034418 A CN201910034418 A CN 201910034418A CN 109825944 B CN109825944 B CN 109825944B
- Authority
- CN
- China
- Prior art keywords
- bright spot
- yarn
- laser
- dimensional
- obtaining
- 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.)
- Active
Links
Images
Abstract
The invention relates to a method for detecting online defects of fabric weaving based on linear laser, which comprises the following steps: irradiating the laser line or the laser surface to the yarn and the knitting needle, and interweaving the yarn and the knitting needle with a light curtain formed by the laser to form a bright spot; the camera collects the formed bright spots; obtaining a bright spot position formed by interweaving yarns, knitting needles and a laser plane by adopting threshold segmentation on each frame of image within a specified sampling time, and extracting the central coordinate of the bright spot; arranging the same bright spot positions in different images in sequence to form a sequence, thereby obtaining the motion track of a single bright spot in the sampling time and obtaining a bright spot one-dimensional motion signal diagram; converting the bright spot one-dimensional motion signal into a two-dimensional spectrogram through wavelet change; deep learning is carried out through a CNN neural network, and a two-dimensional spectrogram corresponding to a normal yarn motion rule and a two-dimensional spectrogram corresponding to an abnormal condition are respectively marked and trained and learned.
Description
Technical Field
The invention belongs to the field of textile quality detection, and relates to a method for detecting defects of woven fabrics on line based on line laser.
Background
China is the largest fabric production and abroad in the world, and defect detection is most important in actual production. At present, the quality control and detection of fabric defects are important, the fabric is generally finished by visual inspection of workers, and the defects of low detection speed, large influence on detection results by subjective factors of the workers, high false detection rate and high omission factor are also caused. Mainly export processing, low added value of textiles and lack of high-level production equipment for autonomous production, and mainly enter machines in Germany, Japan and the like.
The machine networking and automatic control transformation are carried out on the warp knitting equipment, the intelligent manufacturing demonstration is realized on warp knitting enterprises, and the method is an important development direction of the warp knitting industry under the intelligent manufacturing. During the weaving process, the piece goods may have some defects due to mechanical failure or operational errors, etc. Textile enterprises traditionally adopt manual methods to detect fabric flaws, and are not ideal in the aspects of detection precision, speed and detection rate. Therefore, it is necessary to develop an online visual inspection system for detecting defects of cloth of a warp knitting machine, and to realize real-time detection of the defects of the cloth during weaving of the warp knitting machine based on machine vision. At present, the problem of rapid and accurate identification of warp knitting defects is that a successful general warp knitting visual detection system is still in an exploration stage and has a considerable distance from practical online measurement.
Disclosure of Invention
The invention aims to provide a method for detecting defects of fabric weaving on line based on line laser, which can detect typical problems of yarn breakage, abnormal motion law of knitting needles and the like in weaving in real time and can be stopped for reminding, so that the defects in the fabric weaving are fundamentally avoided. The method has the advantages of good adaptability, timely defect judgment, low system cost, stability, reliability and the like, and meets the requirement of online weaving detection of the fabric defects. The technical scheme is as follows:
a fabric weaving online defect detection method based on line laser comprises the following steps:
(1) irradiating the laser line or the laser surface to the yarn and the knitting needle, and interweaving the yarn and the knitting needle with a light curtain formed by the laser to form a bright spot, so that the yarn and the knitting needle are lightened by a laser light source;
(2) the camera collects the formed bright spots and instantaneous images of the yarns and the knitting needles in the weaving process;
(3) obtaining a bright spot position formed by interweaving yarns, knitting needles and a laser plane by adopting threshold segmentation on each frame of image within a specified sampling time, and extracting the central coordinate of the bright spot by a gravity center calculation method;
(4) arranging the same bright spot positions in different images in sequence to form a sequence, thereby obtaining the motion track of a single bright spot in the sampling time and obtaining a bright spot one-dimensional motion signal diagram; converting the bright spot one-dimensional motion signal into a two-dimensional spectrogram through wavelet change;
(5) and deep learning is carried out through a CNN neural network, a two-dimensional spectrogram corresponding to a normal yarn motion rule and a two-dimensional spectrogram corresponding to an abnormal condition are respectively marked and trained, so that whether the abnormal condition including yarn breakage and yarn loosening occurs or not is judged.
The invention provides an online defect detection method for fabric weaving, which is characterized in that yarns and knitting needles are lightened by means of line laser, images of the yarns and the knitting needles in the weaving process are acquired in real time by adopting a high-speed camera, the motion rules of the yarns and the knitting needles in the weaving process are learned through a deep learning algorithm, the typical problems of abnormal motion rules and the like of the yarns during weaving and the knitting needles are judged by judging the damaged motion rules, real-time detection and halt reminding are carried out, and the defects in the fabric weaving are fundamentally avoided. The method has the advantages of good adaptability, timely defect judgment, low system cost, stability, reliability and the like, and meets the requirement of online weaving detection of the fabric defects.
Drawings
FIG. 1 is a schematic structural diagram of a fabric online defect detection system. 1, a camera; a 2-wire laser; 3, knitting needles; 4, light spots; 5, yarns; 6, knitting machine tool; 7 weaving.
FIG. 2 is a flow chart of the fabric on-line defect detection of the present invention.
FIG. 3 is a flow chart of the detection algorithm.
FIG. 4 is a schematic view of a split zone process.
Detailed Description
The invention is described in detail below with reference to the figures and examples.
The system structure of the invention is shown in figure 2, a linear structure laser is arranged at the front end of a knitting machine tool, a laser line directly irradiates yarn from the oblique upper part of the front end of a knitting machine tool, the other laser transmits the laser line to irradiate a knitting needle, a camera is also arranged at the front end of the machine tool and deviates a certain distance from the laser, and bright spot images of the yarn and the knitting needle are synchronously acquired.
The detection flow chart of the invention is shown in fig. 1. Irradiating a laser line or a laser surface on a yarn 5 and a knitting needle 3 on a knitting machine bed 6, and interweaving the yarn and the knitting needle with a light curtain formed by laser to form a bright spot 4, so that the yarn and the knitting needle are lightened by a laser light source; the camera 1 collects the bright spots formed and can collect the instantaneous images of the yarn and the knitting needles during the weaving process. Real-time images of yarns and knitting needles in the weaving process are acquired at a high speed, so that the motion rules of the yarns and the knitting needles in the field of view can be captured, and the motion rules can be learned through a deep learning algorithm. The motion law is collected and judged in real time, when the motion law is damaged, the problems of yarn breakage or abnormal motion of the knitting needle and the like at the corresponding position of the yarn or the knitting needle can be judged, and the pre-form of the defect can also be judged, so that the machine can be stopped in time to avoid waste in the weaving process.
The specific detection algorithm is as shown in fig. 3, firstly, for each frame of picture within a specified sampling time, a bright spot position formed by interweaving yarns, knitting needles and a laser plane is obtained by adopting threshold segmentation, and the center coordinates of the bright spot are extracted by a gravity center calculation method. Then, the same bright spot positions in different pictures are arranged in sequence to form a sequence, so that the motion track of a single bright spot in the sampling time is obtained. The buffeting of the machine tool in the yarn weaving process causes the yarn vibration direction to be single, so the bright spots can be considered to do reciprocating motion in a one-dimensional direction, the bright spot motion condition can be expressed through one-dimensional signals, and a bright spot motion signal diagram is obtained. And then, converting the one-dimensional motion signal of the bright spot into a two-dimensional spectrogram through wavelet change. And finally, deep learning is carried out through the CNN neural network. And respectively marking the two-dimensional spectrogram corresponding to the normal yarn motion law and the two-dimensional spectrogram corresponding to the abnormal conditions, and training and learning to judge whether the abnormal conditions are yarn breakage, yarn loosening and the like.
In order to reduce the system cost, a knitting machine tool needs to adopt a camera and a laser as few as possible, but if a set of measuring device shoots an overlarge view field, position judgment is difficult when yarn breakage or abnormal movement of a knitting needle occurs, so that the judging method can only judge the existence but can not position the knitting machine tool to a specific position. Therefore, a regional learning and judging mode is adopted in the processing process, namely, the image acquired by the camera is divided into small regions, the small regions are respectively subjected to learning and real-time judgment, once abnormality occurs, which small region or small regions occur can be accurately judged, and the abnormal position can be quickly positioned, as shown in fig. 3.
The laser can adopt visible light sources or near infrared light sources with any colors, and the near infrared light sources have the advantage that human eyes cannot see the visible light sources, so that the visual field of field operation workers is prevented from being interfered, but a camera with good near infrared spectrum response characteristics is required to be used in cooperation with the visible light sources or the near infrared light sources.
The specific implementation flow of the fabric defect detection method provided by the invention is as follows:
(1) the method comprises the following steps of building a fabric online defect detection system, installing a linear structure laser at the front end of a knitting machine tool, directly irradiating yarns by a laser line from the oblique upper side of the front end of a knitting machine tool, irradiating knitting needles by the other laser device through the laser line, and installing a camera at the front end of the machine tool and deviating a certain distance from the laser device;
(2) the camera collects the formed bright spots and can collect instantaneous images of the yarns and the knitting needles in the weaving process;
(3) the knitting machine starts a weaving process, real-time collection of real collection images of yarns and knitting needles in the weaving process is carried out for a period of time, appropriate region division is carried out on the real collection images, deep learning is carried out according to regions respectively, and the motion rules of the yarns and the knitting needles in each region are obtained;
(4) circularly and real-timely acquiring motion images of yarns and knitting needles in the actual weaving process, and analyzing whether the motion rules in each subarea are normal or not by using a deep learning result;
(5) and (4) finding abnormal motion rules in any subarea, determining the abnormal motion place of the yarn or the knitting needle, stopping the machine, and performing fault treatment.
Claims (1)
1. A fabric weaving online defect detection method based on line laser comprises the following steps:
(1) irradiating line laser onto the yarn and the knitting needle, and interweaving the yarn and the knitting needle with a plane light curtain formed by the laser to form a bright spot, so that the yarn and the knitting needle are lightened by a laser light source;
(2) the camera collects the formed bright spots and instantaneous images of the yarns and the knitting needles in the weaving process;
(3) obtaining a bright spot position formed by interweaving yarns, knitting needles and a laser plane by adopting threshold segmentation on each frame of image within a specified sampling time, and extracting the central coordinate of the bright spot by a gravity center calculation method;
(4) arranging the same bright spot positions in different images in sequence to form a sequence, thereby obtaining the motion track of a single bright spot in the sampling time and obtaining a bright spot one-dimensional motion signal diagram; converting the bright spot one-dimensional motion signal diagram into a two-dimensional spectrogram through wavelet change;
(5) and deep learning is carried out through a CNN neural network, a two-dimensional spectrogram corresponding to a normal yarn motion rule and a two-dimensional spectrogram corresponding to an abnormal condition are respectively marked and trained, so that whether the abnormal condition including yarn breakage and yarn loosening occurs or not is judged.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910034418.7A CN109825944B (en) | 2019-01-15 | 2019-01-15 | Online fabric weaving defect detection method based on line laser |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910034418.7A CN109825944B (en) | 2019-01-15 | 2019-01-15 | Online fabric weaving defect detection method based on line laser |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109825944A CN109825944A (en) | 2019-05-31 |
CN109825944B true CN109825944B (en) | 2020-04-28 |
Family
ID=66861098
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910034418.7A Active CN109825944B (en) | 2019-01-15 | 2019-01-15 | Online fabric weaving defect detection method based on line laser |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109825944B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111429422A (en) * | 2020-03-19 | 2020-07-17 | 中国工程物理研究院激光聚变研究中心 | Laser near-field state analysis method and device based on deep learning |
Family Cites Families (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US3760184A (en) * | 1972-03-10 | 1973-09-18 | Sick Erwin Fa | Photoelectric monitoring device for pluralities of threads |
US6522402B1 (en) * | 1998-04-30 | 2003-02-18 | California Institute Of Technology | Apparatus and method for analyzing microscopic samples based on optical parametric oscillation |
CN102661724B (en) * | 2012-04-10 | 2014-10-22 | 天津工业大学 | RGBPSP (red green blue phase shift profilometry) three-dimensional color reconstruction method applied to online detection for fabric defects |
CN103116351A (en) * | 2012-12-28 | 2013-05-22 | 福州科迪电子技术有限公司 | Spinning defective cloth detection camera and detection system thereof |
CN104849275A (en) * | 2015-04-22 | 2015-08-19 | 上海工程技术大学 | High speed digital imaging device and method for yarn appearance |
CN105155126A (en) * | 2015-08-25 | 2015-12-16 | 哈尔滨展达机器人自动化有限责任公司 | Broken line detection device based on laser vision for warp knitting |
-
2019
- 2019-01-15 CN CN201910034418.7A patent/CN109825944B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN109825944A (en) | 2019-05-31 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN202175829U (en) | On-line real-time detection system for gray fabric flaw based on machine vision | |
CN104764750B (en) | Elevator balanced compensated chain automatic detection device for quality and method based on machine vision | |
CN107886500A (en) | A kind of production monitoring method and system based on machine vision and machine learning | |
KR102445162B1 (en) | Fabric inspection system | |
CN203688459U (en) | On-line fabric quality detection system based on image recognition | |
CN202770781U (en) | Fabric defect online detection device based on machine vision | |
CN113166987A (en) | Machine and method for controlling textile quality | |
CN104949990A (en) | Online detecting method suitable for defects of woven textiles | |
CN106596568A (en) | Real-time non-contact yarn breakage detection method based on line laser | |
CN102175692A (en) | System and method for detecting defects of fabric gray cloth quickly | |
CN109632817B (en) | Fabric weaving on-line defect detection method based on collimated laser beam | |
US20220170189A1 (en) | A device and a method for real-time identification of defects in fabrics, during weaving | |
CN207866722U (en) | Mobile Automatic Detection of Fabric Defects system | |
CN103116351A (en) | Spinning defective cloth detection camera and detection system thereof | |
CN110318245A (en) | A kind of machine vision fabric inspecting system being integrated on loom | |
CN109825944B (en) | Online fabric weaving defect detection method based on line laser | |
CN113592852A (en) | Carbon fiber carbon wire quality online detection system and method | |
CN204807470U (en) | Two -sided perching device | |
CN111337509A (en) | Textile fabric surface flaw detection device, detection system and detection method thereof | |
CN107144570A (en) | A kind of tufting machine row's yarn error-detecting method based on machine vision | |
CN211122582U (en) | Visual detection device for defects of embryonic cloth | |
CN115508282B (en) | Online intelligent cloth inspection detection system | |
CN208023291U (en) | A kind of perching machine detecting device of clothes processing | |
CN108221339A (en) | A kind of perching machine detecting device of clothes processing | |
CN205334553U (en) | Spinning frame detection device that breaks end on line based on spun yarn image |
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 |