CN110813795A - Device and method for detecting breakage of chemical fiber spindle paper tube on line - Google Patents
Device and method for detecting breakage of chemical fiber spindle paper tube on line Download PDFInfo
- Publication number
- CN110813795A CN110813795A CN201910625029.1A CN201910625029A CN110813795A CN 110813795 A CN110813795 A CN 110813795A CN 201910625029 A CN201910625029 A CN 201910625029A CN 110813795 A CN110813795 A CN 110813795A
- Authority
- CN
- China
- Prior art keywords
- unit
- paper tube
- image
- rejection
- camera
- 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
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/34—Sorting according to other particular properties
- B07C5/342—Sorting according to other particular properties according to optical properties, e.g. colour
- B07C5/3422—Sorting according to other particular properties according to optical properties, e.g. colour using video scanning devices, e.g. TV-cameras
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/89—Investigating the presence of flaws or contamination in moving material, e.g. running paper or textiles
- G01N21/892—Investigating the presence of flaws or contamination in moving material, e.g. running paper or textiles characterised by the flaw, defect or object feature examined
Abstract
The invention discloses an online detection device and method for breakage of a chemical fiber spindle paper tube. The automatic paper tube sorting machine comprises a conveying unit, an image acquisition unit, a control end, a rejection unit and a sorting unit, wherein the conveying unit is used for conveying paper tubes, a detection area of the rejection unit on a transmission unit sorts the paper tubes, the image acquisition unit carries out image acquisition on the paper tubes, the control end is provided with a classifier for detecting the breakage of the paper tubes, the control end receives image information of the image acquisition unit, the image information is detected and output by the classifier to obtain a detection result, the detection result is compared with a set grade index, a control instruction is sent to the rejection unit according to the comparison result, and the paper tubes are sorted. The invention does not affect the normal production of the paper tube and the silk ingot, and realizes the online automatic detection of the paper tube damage. Through the unified coordination of the transmission unit, the image acquisition unit and the rejection unit, the damaged paper tube is timely, accurately and quickly rejected.
Description
Technical Field
The invention relates to the technical field of paper tube breakage defect detection, in particular to a device and a method for detecting breakage of a chemical fiber spindle paper tube on line.
Background
The production process of the chemical fiber filament is discontinuous, so that each production process of the chemical fiber filament needs to apply a winding mechanism to wind the filament into a package with certain shape and capacity so as to realize intermediate storage and process circulation. As a carrier of the filament, the quality of the paper tube plays a crucial role in stabilizing and improving the spinning quality.
During the winding process, if the paper tube is damaged, broken filaments or abnormal filaments are generated during the winding process, and the abnormal filaments are contained in a package, so that poor winding is caused; if the paper tube has deformation, tube bursting can be caused by the tension action in the winding process, and the normal production process is influenced. In the process of unwinding the silk spindle, if the paper tube is deformed or damaged, light people can hang silk, the long silk is difficult to retreat from the tube, and heavy people can cause rejection of spinning cakes or equipment accidents, so that the normal production flow is influenced, and inestimable loss is caused to enterprises.
Firstly, detection results are influenced by subjective factors such as quality and experience of detection personnel, different detection personnel can judge the defects of the unified degree differently due to different experiences of the detection personnel, so that the defect types and the defect levels are inconsistent, the detection accuracy and the detection standardization are lacked, and the detection result reliability is lower; secondly, the labor intensity of manual detection is high, the manual detection is easy to fatigue, fine defects are difficult to detect, misjudgment and missed detection are easy to cause, the effective monitoring of the detection quality is influenced, and most importantly, for large-scale detection projects, a corresponding number of workers need to be equipped, so that the operation cost of enterprises is increased; thirdly, the manual detection speed is low, the working efficiency is low, the continuous working characteristics are not achieved, the requirements of modern mass production are not met, and the mass and high-quality automatic flow line production process is difficult to realize. Therefore, the traditional manual detection is far from meeting the requirements of high-efficiency and low-consumption production, so that the online detection system becomes a key technology for mutual competition of various large manufacturers. The machine vision system is used as an information source of an appearance detection system, and how to simulate human vision even surpass the limit of the human vision is always a problem that the machine vision is committed to break through. Machine vision is a technology for converting an image into a digital signal to be analyzed and processed by combining an image processing technology, non-contact perception is carried out on an object through an optical device, an image of a large-range real scene is automatically acquired and interpreted, and finally information acquisition and machine control are achieved. The machine vision has the greatest advantages that the machine vision is not in contact with the detected object, so that the machine vision does not cause any damage to the observed and observed people, is safe and reliable, and does not influence the production process; moreover, machine vision in a range which can not be observed by human eyes can be observed through certain mathematical calculation, such as measurement of the area and the length of an object; in addition, the human eyes are not sensitive to the illumination conditions such as infrared rays, microwaves and ultrasonic waves, so that detailed information under the illumination conditions can be ignored, and machine vision can form specific images such as infrared rays, microwaves and ultrasonic waves by using a specific sensor to assist the human eyes to observe; in addition, human eyes cannot observe objects for a long time due to the influences of visual fatigue, subjective feeling and the like, and machine vision has no time limit and has consistent high precision and resolution; finally, the machine vision can collect and store data in production and manufacturing, and the data is analyzed and processed by a software algorithm to make a decision and a judgment, so that an intelligent and data industrial chain is required to be made, the machine is made to make a 'smart' decision, and the assistance of the machine vision is indispensable.
In recent years, with the popularization of informatization degrees of various industries, data accumulated in various industries is more and more, when massive data are faced, a traditional machine learning method is unconscious, and an efficient and accurate analysis, identification and mining method for image content is lacked, at this time, deep learning relying on large-scale data for training appears, and becomes a favorite of a machine vision algorithm field once, the appearance of the deep learning greatly shortens the research and development period while ensuring the precision, and saves the time cost for the whole production chain. Deep learning has its own advantages, but not all the more, it is sensitive to noise and influence caused by light source, focusing, shooting angle, etc., and thus there is a great instability to slightly inconsistent images, and conventional machine vision algorithms can deal well with these problems. Detection tasks can be better accomplished only by combining traditional machine vision algorithms with deep learning, such as extracting easily quantifiable features with traditional machine vision algorithms: the characteristics such as color, area, roundness, length, angle and the like are detected by deep learning, and the characteristics which are difficult to extract are as follows: flaws, defects, etc.
Disclosure of Invention
The invention mainly solves the problems of low accuracy, false detection, missed detection and low speed of manual detection of a paper tube in the prior art and the problems of lack of accurate analysis, identification and excavation of image content in a traditional machine learning method, and provides the device and the method for detecting the damage of the paper tube of the chemical fiber spindle on line.
The technical problem of the invention is mainly solved by the following technical scheme: an on-line detection device for the breakage of a chemical fiber spindle paper tube comprises,
the conveying unit is used for conveying the paper tubes, a rejection unit for sorting the paper tubes and an image acquisition unit for acquiring images of the paper tubes are arranged in a detection area on the transmission unit and are positioned around the rejection unit; the paper tube has two different states before winding and after winding on the automatic production line, for the state before winding, the paper tube has no filaments, the paper tube enters the conveying unit in a horizontal rolling mode, for the state after winding, the paper tube has filaments already thereon, and the filament ingots are loaded into the conveying unit from the tray in a governing mode.
The control end is provided with a classifier for detecting the damage of the paper tube, receives the image information of the image acquisition unit, detects the image information through the classifier and outputs a detection result, compares the detection result with a set grade index, and sends a control instruction to the rejection unit according to the comparison result to sort the paper tube.
The detection device does not influence the normal production of the paper tube and the silk ingots, and realizes the online automatic detection of the damage of the paper tube. The invention realizes timely, accurate and high-speed removal of the damaged paper tube by the unified coordination of the transmission unit, the image acquisition unit and the removal unit.
As a preferable scheme, the image acquisition unit comprises a mechanical claw, a camera group, a light source group and a sensor group,
the mechanical claw is arranged at the rejecting unit, grabs the paper tube and rotates; the gripper is rotated by motor drive, and the gripper can penetrate the fiber container cavity and snatch the fiber container. For the paper tube before winding, the mechanical claw is arranged on one side of the removing unit through a support rod, and for the paper tube after winding, the mechanical claw is arranged on the top of the removing unit through the support rod.
The camera set comprises a first camera, a second camera and a third camera, and is respectively arranged at the top, the front end and the rear end of the detection area of the conveying unit through supporting rods; aiming at two different states of the paper tube before and after winding on the automatic production line, the first camera, the second camera and the third camera are respectively used for acquiring images of the top, the front end and the rear end of the paper tube before winding.
Or comprises a first camera and a fourth camera which are respectively arranged at the top and the bottom of the detection area of the conveying unit through a support rod; the first camera and the fourth camera are respectively used for collecting images of the top and the bottom of the wound paper tube. The camera adopts a linear array CCD camera to take pictures of the motion of each end face of the paper tube to finish image acquisition.
The light source group comprises a plurality of linear light sources, and each camera is correspondingly provided with one linear light source; the light source group adopts linear LED light sources, and each linear light source is installed corresponding to one camera to provide a light source for the work of the camera.
And the sensor group is arranged in the detection area of the conveying unit and used for detecting whether the paper tube reaches the detection area. The sensor group adopts an infrared sensor, is arranged in a detection area, namely the position of the rejecting unit, and sends a signal to the control end when the paper tube reaches the position of the rejecting unit.
Preferably, the control terminal comprises an image receiving unit, a processing unit and a control unit,
the image receiving unit is connected with the camera group and used for acquiring images shot by the camera group; the image receiving unit adopts an image acquisition card, the image acquisition card is installed in the control end in an insertion card mode, and the image receiving unit acquires images shot by the camera and sends the images to the processing unit.
The processing unit is used for detecting the image by the classifier and outputting a detection result, generating a control instruction and sending the control instruction to the control unit; and detecting the images of the paper tube including the top, front and rear images or the top and bottom images by using a classifier, and outputting the damage level of the paper tube.
And the control unit is used for controlling the camera set and the light source set to work after the sensor set detects that the paper tube reaches the detection area, and controlling the sorting unit to sort the paper tube after receiving a control instruction. When the paper tube reaches the detection area of the conveyor belt, namely the rejection unit, the sensor group sends information to the processing unit, the processing unit sends the information to the control unit, the control unit controls the camera group and the light source group to work, and simultaneously sends an instruction to the mechanical claw, the paper tube is grabbed and rotated, and the camera group collects images of the paper tube.
The control unit respectively sends out starting pulses to the camera group and the light source group according to a set program and time delay, the exposure mechanism is opened before the camera group starts scanning a new frame, after the camera is exposed, scanning and outputting of a frame are formally started, the image acquisition unit receives an analog video signal and digitalizes a charge signal through an A/D conversion circuit, then the digitalized signal of the image is stored in a memory, the processing unit acquires the image from the memory, processes, analyzes and identifies the image and obtains a measurement result, the control unit converts the measurement result into a logic control value and controls a relevant mechanical claw, a rejection valve and the like to make corresponding actions.
Preferably, the sorting unit comprises, in combination,
the rejection valve is arranged in the conveying unit, the conveying unit comprises a front section and a rear section, the rejection valve is arranged between the front section and the rear section of the conveying unit, and the rejection valve comprises an opening and closing sorting door; this scheme is used for handling the paper section of thick bamboo before coiling, and the conveying unit includes two sections around input and output, connects between two sections and rejects the valve, rejects the letter sorting door of valve and opens and shuts, detects the paper section of thick bamboo damage, rejects the valve and opens letter sorting door, is provided with the recovery basket in rejecting the valve lower part, and damaged paper section of thick bamboo falls into in the recovery basket.
Or a rejection valve arranged on the transmission unit, wherein the rejection valve comprises a telescopic baffle; and a baffle extends out of the paper tube reaching the detection area. This scheme is used for handling the back fiber container of convoluteing, rejects the valve and is located on the transfer unit, reachs the detection zone at the fiber container, and the control unit control is rejected the valve and is stretched out the baffle, blocks the fiber container, and control mechanical arm snatchs the fiber container simultaneously and carries out image shooting, and when judging that the fiber container does not have the damage, rejects the valve and withdraws the baffle, and the fiber container gets back to the production line. When the paper tube is judged to be damaged, the paper tube is grabbed by the mechanical claw and put into the backflow frame on the side edge.
An online detection method for breakage of a chemical fiber spindle paper tube comprises the following steps:
s1, constructing a classifier for detecting the damage of a paper tube, and presetting a paper tube damage grade index;
s2, collecting an image of the paper tube to be detected, and detecting the image through a classifier to obtain a damage detection result;
s3, comparing the detection result with the grade index to generate a sorting control instruction; and comparing the detection result in which grade index range, and obtaining the grade index corresponding to the detection result after comparison. The control unit is used for presetting a paper tube, removing one part of grade indexes, and obtaining good products for other part of grade indexes, and the processing unit generates a control instruction for removing or obtaining good products according to the obtained grade indexes.
And S4, sorting the paper tubes according to the control instruction.
The invention does not affect the normal production of the paper tube and the silk ingot, and realizes the online automatic detection of the paper tube damage. The invention realizes the detection of the damaged paper tube through the classifier trained by big data and can realize the timely, accurate and high-speed rejection of the paper tube.
As a preferable scheme, the process of constructing the classifier for detecting the breakage of the paper tube in the step S1 includes:
s11, carrying out image acquisition on a plurality of paper tubes;
s12, preprocessing the acquired image;
s13, extracting and classifying pre-selected blocks from the preprocessed image;
s14, carrying out damage grade division and marking on the classified pre-selected blocks; as an extension
And S15, inputting the classification preselection block as a training sample set into a VGG network for training to obtain a classifier for detecting the damage of the paper tube. In addition, non-damaged images are collected, and round blocks are extracted and classified from the images and are put into a training sample set as negative samples.
The preprocessing process of the image comprises graying processing, histogram correction and Gaussian filtering. The graying process carries out weighted average on RGB three components according to a formula: f is 0.3R +0.59G +0.11B, where F is the converted grayscale image. The obtained grey image histogram is corrected to make the grey interval of the image be separated or the grey distribution be uniform, so as to increase the contrast, and to make the image detail be cleaned, and can attain the goal of image enhancement, and its correspondent mapping method is:
wherein S is the statistics of the mapped gray values, n is the sum of the number of pixels in the original image, and n isjThe number of pixels corresponding to the gray level j, and L is the total number of possible gray levels in the original image. And the corrected image is subjected to Gaussian filtering, so that random noise in the image can be well suppressed, loss of edge information in the filtering process can be avoided, and the contour is well reserved.
Marking the classified pre-selected blocks, firstly finely grading the damage degree of the paper tube in advance, and respectively grading and marking the classified pre-selected blocks according to grades. And the grades of the marked classified preselected blocks are counted, the ratio of each grade classified preselected block is calculated, the number of grade samples with less ratio is supplemented, and the difference between each grade classified preselected block is not more than 0.5 times at most.
And (3) scaling each image in the collected training sample set to a fixed size, turning left and right, turning up and down, amplifying the training sample set by PCA operation, wherein the number of the obtained training sample set is at least 10 ten thousand, using the training sample set as the input of a VGG network, and setting a proper hyper-parameter to train so as to obtain a stable classifier for detecting the damage of the paper tube. And in addition, the classifier can be iterated, wherein the iteration process comprises the steps of collecting a large number of samples, testing the samples according to the obtained classifier, screening the samples with wrong classification in the test result, marking and correcting errors, putting the samples into a training sample set, and repeating iteration until the classifier is within a tolerable error detection range and reaches a required positive detection target.
As a preferable scheme, the process of extracting the classified pre-selected block in step S13 includes:
s131, setting a priori threshold L of the length of a connected domain and a priori threshold A of the area of the connected domain of each image acquired in the step S11 respectively; for the paper tube before winding, the camera group completes one shooting for one paper tube through the step S11 to obtain 3 rectangular graphs, which respectively correspond to the two circular end surface expansion graphs and the cylindrical surface expansion graph of the paper tube. For the wound paper tube, the camera group completes one shooting of one spindle to obtain 2 rectangular pictures through the step S11, and the rectangular pictures respectively correspond to the two circular end face development pictures of the paper tube. And respectively setting a priori threshold of the length and area of the connected domain of each expansion map.
S132, carrying out binarization processing on the preprocessed image;
the specific content of the binarization processing comprises the following steps:
calculating a global threshold T of the image by counting the histogram characteristics of the whole image through a large law method OSTU; and according to the calculated threshold value T, setting the pixel value f (x, y) corresponding to each pixel point on the image to be 255 if f (x, y) > T and 0 if f (x, y) < T, thereby realizing the binarization processing of the image and obtaining a binary image.
S133, hole filling and small area removal are achieved on the image after binarization processing through a seed area growing method;
s134, performing independent connected domain calculation on the image processed in the step S133, calculating the length and the area of each connected domain, filling the connected domains with the lengths smaller than the prior threshold value L with black, and filling the connected domains with the areas smaller than the prior threshold value A with black;
s135, calculating the minimum circumscribed rectangle of each connected domain in the image processed in the step S134, corresponding each obtained minimum circumscribed rectangle to the original image area, and intercepting the minimum circumscribed rectangle as a classified pre-selected block.
As a preferable scheme, the specific process of sorting in step S4 includes:
for paper tube before winding
If the control unit sends a rejection instruction to the rejection unit, the mechanical claw is controlled to place the paper tube with the collected image back onto the rejection valve, the mechanical claw contracts back to the initial position, the rejection unit opens the sorting door when receiving the rejection instruction, so that the paper tube flows back to fall below, the sorting door is closed, and the rejection unit waits for the next paper tube to arrive; if the control unit sends a good product instruction to the rejecting unit, the mechanical claw is controlled to return the paper tube to the conveying unit, the mechanical claw contracts back to the initial position, and the rejecting unit waits for the next paper tube to arrive;
for the wound paper tube
The rejecting unit extends out of the baffle to block the tray from moving, if the control unit sends a rejecting instruction to the rejecting unit, the mechanical claw is controlled to place the filament ingots with the collected images into the side edge backflow frame, the rejecting unit retracts the baffle, the mechanical claw retracts to the initial position, and the rejecting unit waits for the next paper tube to arrive; if the control unit sends a good product instruction to the removing unit, the removing unit withdraws the baffle when receiving the instruction, the filament ingot passes through the baffle, and the mechanical claw retracts to the initial position. The wound paper tube is a yarn spindle which is vertically arranged on a tray and is conveyed on a conveying unit.
Therefore, the invention has the advantages that:
1. the detection device does not influence the normal production of the paper tube and the silk ingots, and realizes the online automatic detection of the damage of the paper tube.
2. The invention realizes timely, accurate and high-speed removal of the damaged paper tube by the unified coordination of the transmission unit, the image acquisition unit and the removal unit.
3. The detection device can be adapted to two different states of the paper tube on the production line before and after winding simultaneously through different assembly modes, and realizes one-line detection on the paper tube.
4. Through the combined use of the camera set and the mechanical claw, the three end faces of the paper tube can be well unfolded, the camera vision blind area is avoided, and the obtained image is convenient for post processing.
Drawings
FIG. 1 is a block diagram of one configuration of the present invention;
FIG. 2 is a schematic diagram of an embodiment of the present invention;
FIG. 3 is a schematic flow diagram of the present invention;
FIG. 4 is a schematic flow chart of the steps of constructing the classifier according to the present invention.
The method comprises the following steps of 1, a control end 2, an image acquisition unit 3, a camera set 4, a light source set 5, a mechanical claw 6, a sensor set 7, an image receiving unit 8, a processing unit 9, a control unit 10, a rejection unit 11, a second camera 12, a third camera 13, a first camera 14, a rejection valve 15 and a support rod.
Detailed Description
The technical scheme of the invention is further specifically described by the following embodiments and the accompanying drawings.
Example (b):
the device for detecting the breakage of a chemical fiber spindle paper tube in the embodiment, as shown in fig. 1, comprises a transmission unit and a control end 1, wherein
The conveying unit is used for conveying the paper tubes, a rejection unit 10 for sorting the paper tubes and an image acquisition unit 2 for acquiring images of the paper tubes are arranged in a detection area on the transmission unit and are positioned around the rejection unit; the paper tube has two different states before winding and after winding on the automatic production line, the paper tube has no wires on the paper tube and enters the conveying unit in a horizontal rolling mode in the state before winding, and the paper tube has wires on the paper tube and the wire ingots are loaded into the conveying unit from the tray in an upright mode in the state after winding.
And the control end is provided with a classifier for detecting the damage of the paper tube, receives the image information of the image acquisition unit 2, detects the image information by the classifier, outputs a detection result, compares the detection result with a set grade index, and sends a control instruction to the rejection unit 10 according to the comparison result to sort the paper tube.
The image acquisition unit comprises a mechanical claw 5, a camera group 3, a light source group 4 and a sensor group 6.
The mechanical claw is arranged at the rejecting unit, grabs the paper tube and rotates; the mechanical claw is a mechanical claw capable of being selectively extended and retracted. The gripper is rotated by motor drive, and the gripper can penetrate the fiber container cavity and snatch the fiber container inner wall. For the paper tube before winding, the mechanical claw is arranged on one side of the removing unit through a support rod, and for the paper tube after winding, the mechanical claw is arranged on the top of the removing unit through the support rod. The mechanical claw controls the paper cylinder to a set height to be aligned with the light source combination camera set.
The camera set is used for the state before the paper tube is wound, as shown in fig. 2, the structure of the camera set comprises a first camera 13, a second camera 11 and a third camera 12 which are respectively arranged at the top, the front end and the rear end of the detection area of the conveying unit through supporting rods 15; for the state after the paper tube is wound, the first camera 13 and the fourth camera are respectively arranged at the top and the bottom of the detection area of the conveying unit through supporting rods; the camera group can be adapted to two different states of the paper tube on the production line before and after winding through different assembling modes.
The light source group comprises a plurality of linear light sources, and each camera is correspondingly provided with one linear light source;
and the sensor group is arranged in the detection area of the conveying unit and used for detecting whether the paper tube reaches the detection area. The sensor group adopts an infrared sensor, is arranged in a detection area, namely the position of the rejecting unit, and sends a signal to the control end when the paper tube reaches the position of the rejecting unit.
The control terminal comprises in particular an image receiving unit 7, a processing unit 8 and a control unit 9,
the image receiving unit is connected with the camera group and used for acquiring images shot by the camera group; the image acquisition unit adopts an image acquisition card, the image acquisition card is installed in the control end in an insertion card mode, and the image acquisition unit acquires images shot by the camera and sends the images to the processing unit.
The processing unit is used for detecting the image by the classifier and outputting a detection result, generating a control instruction and sending the control instruction to the control unit; the images are shot images of the paper tube, including top, front end and rear end images or top and bottom images, and the images are detected by a classifier to output the damage level of the paper tube.
The control unit controls the camera set and the light source set to work after the paper tube reaches the detection area after the sensor set detects that the paper tube reaches the detection area, specifically, the paper tube reaches the conveyor belt detection area and is rejected at the unit, the sensor set sends information to the processing unit, the processing unit sends the information to the control unit, the control unit controls the camera set and the light source set to work, meanwhile, an instruction is sent to the mechanical claw, the paper tube is grabbed and rotated, and the camera set collects images of the paper tube. And in addition, after receiving the control instruction, the sorting unit is controlled to sort the paper tubes.
The sorting unit, for the paper tube before winding, as shown in fig. 2, comprises a reject valve 14 arranged in the transfer unit. The conveying unit comprises an input section, an output section and a front section, the rejecting valve is installed between the front section and the rear section of the conveying unit, the rejecting valve comprises an opening and closing sorting door, when the damage of the paper tube is detected, the rejecting valve opens the sorting door, the lower portion of the rejecting valve is provided with a recovery basket, and the damaged paper tube falls into the recovery basket. For the wound paper tube, the paper tube comprises a rejection valve arranged on a transmission unit, wherein the transmission unit is not segmented as a whole. The rejection valve comprises a telescopic baffle; and a baffle extends out when the paper tube reaches the detection area to block the paper tube. When judging that there is the damage, snatch the fiber container by the gripper and put into the backward flow frame of side, when judging that the fiber container is not damaged, reject the valve and withdraw the baffle, the fiber container gets back to the production line.
The transmission unit is set to move at a constant speed, the speed is v, paper tubes or silk ingots are placed on the transmission unit at equal intervals, the distance between the paper tubes or the silk ingots is d, the paper tubes are pulled by mechanical claws and are grabbed, the paper tubes are taken to shoot, the paper tubes are returned to the eliminating valve, then the paper tubes are contracted to the original position again at t1, the opening and closing time of the eliminating valve is t2, d > v (t1+ t2) needs to be ensured, the range of t1+ t2 is basically less than 3s under the general condition, the condition that congestion does not occur on the transmission unit is ensured, and the automatic production line is ensured to be smoothly carried out.
An online detection method for breakage of a chemical fiber spindle paper tube is shown in fig. 3, and comprises the following steps:
s1, constructing a classifier for detecting the damage of a paper tube, and presetting a paper tube damage grade index;
s2, collecting an image of the paper tube to be detected, and detecting the image through a classifier to obtain a damage detection result;
s3, comparing the detection result with the grade index to generate a sorting control instruction;
and S4, sorting the paper tubes according to the control instruction.
The specific process of constructing the classifier in step S1, as shown in fig. 4, includes,
s11, carrying out image acquisition on a plurality of paper tubes;
s12, preprocessing the acquired image;
the preprocessing process comprises graying, histogram correction and Gaussian filtering.
Graying processing performs weighted averaging on the RGB three components according to the formula: f is 0.3R +0.59G +0.11B, where F is the converted grayscale image.
The obtained grey image histogram is corrected to make the grey interval of the image be separated or the grey distribution be uniform, so as to increase the contrast, and to make the image detail be cleaned, and can attain the goal of image enhancement, and its correspondent mapping method is:
wherein S is the statistics of the mapped gray values, n is the sum of the number of pixels in the original image, and n isjThe number of pixels corresponding to the gray level j, and L is the total number of possible gray levels in the original image.
And the corrected image is subjected to Gaussian filtering, so that random noise in the image can be well suppressed, loss of edge information in the filtering process can be avoided, and the contour is well reserved.
S13, extracting and classifying pre-selected blocks from the preprocessed image; the process comprises the following steps:
s131, setting a priori threshold L of the length of a connected domain and a priori threshold A of the area of the connected domain of each image acquired in the step S11 respectively; for the paper tube before winding, the camera group completes one shooting for one paper tube through the step S11 to obtain 3 rectangular graphs, which respectively correspond to the two circular end surface expansion graphs and the cylindrical surface expansion graph of the paper tube. For the wound paper tube, the camera group completes one shooting of one spindle to obtain 2 rectangular pictures through the step S11, and the rectangular pictures respectively correspond to the two circular end face development pictures of the paper tube. And respectively setting a priori threshold of the length and area of the connected domain of each expansion map.
S132, carrying out binarization processing on the preprocessed image;
the specific content of the binarization processing comprises the following steps: calculating a global threshold T of the image by counting the histogram characteristics of the whole image through a large law method OSTU; and according to the calculated threshold value T, setting the pixel value f (x, y) corresponding to each pixel point on the image to be 255 if f (x, y) > T and 0 if f (x, y) < T, thereby realizing the binarization processing of the image and obtaining a binary image.
S133, hole filling and small area removal are achieved on the image after binarization processing through a seed area growing method;
s134, calculating independent connected domains of the image processed in the step S133, calculating the length and the area of each connected domain, judging whether the length of each connected domain is smaller than a prior threshold L, if the connected domains are filled with black, selecting the next connected domain to judge the length and the area of each connected domain, if the connected domains are not smaller than the prior threshold A, if the connected domains are filled with black, selecting the next connected domain to judge the length and the area of each connected domain, if the connected domains are not filled with black, judging whether all the connected domains are traversed, if the connected domains are not selected to judge the length and the area of each connected domain, and if the connected domains are not selected to judge the length and the area of each connected domain, entering the next step.
S135, calculating the minimum circumscribed rectangle of each connected domain in the image processed in the step S134, corresponding each obtained minimum circumscribed rectangle to the original image area, and intercepting the minimum circumscribed rectangle as a classified pre-selected block.
S14, carrying out damage grade division and marking on the classified pre-selected blocks; and finely grading the damage degree of the paper tube in advance, and grading and marking the classified pre-selected blocks according to grades respectively. And the grades of the marked classified preselected blocks are counted, the ratio of each grade classified preselected block is calculated, the number of grade samples with less ratio is supplemented, and the difference between each grade classified preselected block is not more than 0.5 times at most.
And S15, inputting the classification preselection block as a training sample set into a VGG network for training to obtain a classifier for detecting the damage of the paper tube. Except that the classification preselected blocks obtained through the processing are used as a training sample set, each classification preselected block in the training sample set is zoomed to a fixed size and is turned left and right, and turned up and down, and the PCA operation is used for amplifying the training sample set. In addition, a non-damaged image is obtained and processed to obtain a classified preselected block, and the classified preselected block is used as a negative sample and put into a training sample set. The number of the training sample sets obtained finally is at least 10 ten thousand.
The specific process of sorting in step S4 includes:
for paper tube before winding
If the control unit sends a rejection instruction to the rejection unit, the mechanical claw is controlled to place the paper tube with the collected image back onto the rejection valve, the mechanical claw contracts back to the initial position, the rejection unit opens the sorting door when receiving the rejection instruction, so that the paper tube flows back to fall below, the sorting door is closed, and the rejection unit waits for the next paper tube to arrive; if the control unit sends a good product instruction to the rejecting unit, the mechanical claw is controlled to return the paper tube to the conveying unit, the mechanical claw contracts back to the initial position, and the rejecting unit waits for the next paper tube to arrive;
for the wound paper tube
The rejecting unit extends out of the baffle to block the tray from moving, if the control unit sends a rejecting instruction to the rejecting unit, the mechanical claw is controlled to place the filament ingots with the collected images into the side edge backflow frame, the rejecting unit retracts the baffle, the mechanical claw retracts to the initial position, and the rejecting unit waits for the next paper tube to arrive; if the control unit sends a good product instruction to the removing unit, the removing unit withdraws the baffle when receiving the instruction, the filament ingot passes through the baffle, and the mechanical claw retracts to the initial position.
The specific embodiments described herein are merely illustrative of the spirit of the invention. Various modifications or additions may be made to the described embodiments or alternatives may be employed by those skilled in the art without departing from the spirit or ambit of the invention as defined in the appended claims.
Although the terms control terminal, image acquisition unit, camera group, light source group, gripper, etc. are used more often herein, the possibility of using other terms is not excluded. These terms are used merely to more conveniently describe and explain the nature of the present invention; they are to be construed as being without limitation to any additional limitations that may be imposed by the spirit of the present invention.
Claims (8)
1. The utility model provides a chemical fibre silk spindle fiber container breakage on-line measuring device which characterized in that: comprises the steps of (a) preparing a mixture of a plurality of raw materials,
the conveying unit is used for conveying the paper tubes, a rejection unit for sorting the paper tubes and an image acquisition unit for acquiring images of the paper tubes are arranged in a detection area on the transmission unit and are positioned around the rejection unit;
the control end is provided with a classifier for detecting the damage of the paper tube, receives the image information of the image acquisition unit, detects the image information through the classifier and outputs a detection result, compares the detection result with a set grade index, and sends a control instruction to the rejection unit according to the comparison result to sort the paper tube.
2. The device of claim 1, wherein the image capturing unit comprises a gripper, a camera set, a light source set and a sensor set,
the mechanical claw is arranged at the rejecting unit, grabs the paper tube and rotates;
the camera set comprises a first camera, a second camera and a third camera, and is respectively arranged at the top, the front end and the rear end of the detection area of the conveying unit through supporting rods;
or comprises a first camera and a fourth camera which are respectively arranged at the top and the bottom of the detection area of the conveying unit through a support rod;
the light source group comprises a plurality of linear light sources, and each camera is correspondingly provided with one linear light source;
and the sensor group is arranged in the detection area of the conveying unit and used for detecting whether the paper tube reaches the detection area.
3. The device of claim 2, wherein the control end comprises an image receiving unit, a processing unit and a control unit,
the image receiving unit is connected with the camera group and used for acquiring images shot by the camera group;
the processing unit is used for detecting the image by the classifier and outputting a detection result, generating a control instruction and sending the control instruction to the control unit;
and the control unit is used for controlling the camera set and the light source set to work after the sensor set detects that the paper tube reaches the detection area, and controlling the sorting unit to sort the paper tube after receiving a control instruction.
4. The device of claim 1, wherein the sorting unit comprises,
the rejection valve is arranged in the conveying unit, the conveying unit comprises a front section and a rear section, the rejection valve is arranged between the front section and the rear section of the conveying unit, and the rejection valve comprises an opening and closing sorting door;
or a rejection valve arranged on the transmission unit, wherein the rejection valve comprises a telescopic baffle; and a baffle extends out of the paper tube reaching the detection area.
5. An on-line detection method for the breakage of a chemical fiber spindle paper tube, which adopts the device in any one of claims 1 to 4, and is characterized by comprising the following steps:
s1, constructing a classifier for detecting the damage of a paper tube, and presetting a paper tube damage grade index;
s2, collecting an image of the paper tube to be detected, and detecting the image through a classifier to obtain a damage detection result;
s3, comparing the detection result with the grade index to generate a sorting control instruction;
and S4, sorting the paper tubes according to the control instruction.
6. The method of claim 1, wherein the step of constructing a classifier for detecting fiber bobbin breakage in step S1 comprises:
s11, carrying out image acquisition on a plurality of paper tubes;
s12, preprocessing the acquired image;
s13, extracting and classifying pre-selected blocks from the preprocessed image;
s14, carrying out damage grade division and marking on the classified pre-selected blocks;
and S15, inputting the classification preselection block as a training sample set into a VGG network for training to obtain a classifier for detecting the damage of the paper tube.
7. The method of claim 6, wherein the step of extracting the sorted pre-selected blocks in step S13 comprises:
s131, setting a priori threshold L of the length of a connected domain and a priori threshold A of the area of the connected domain of each image acquired in the step S11 respectively;
s132, carrying out binarization processing on the preprocessed image;
s133, hole filling and small area removal are achieved on the image after binarization processing through a seed area growing method;
s134, performing independent connected domain calculation on the image processed in the step S133, calculating the length and the area of each connected domain, filling the connected domains with the lengths smaller than the prior threshold value L with black, and filling the connected domains with the areas smaller than the prior threshold value A with black;
s135, calculating the minimum circumscribed rectangle of each connected domain in the image processed in the step S134, corresponding each obtained minimum circumscribed rectangle to the original image area, and intercepting the minimum circumscribed rectangle as a classified pre-selected block.
8. The method for detecting the breakage of the chemical fiber spindle paper tube in the claim 1, wherein the step S4 comprises the following steps:
for paper tube before winding
If the control unit sends a rejection instruction to the rejection unit, the mechanical claw is controlled to place the paper tube with the collected image back onto the rejection valve, the mechanical claw contracts back to the initial position, the rejection unit opens the sorting door when receiving the rejection instruction, so that the paper tube flows back to fall below, the sorting door is closed, and the rejection unit waits for the next paper tube to arrive; if the control unit sends a good product instruction to the rejecting unit, the mechanical claw is controlled to return the paper tube to the conveying unit, the mechanical claw contracts back to the initial position, and the rejecting unit waits for the next paper tube to arrive;
for the wound paper tube
The rejecting unit extends out of the baffle to block the tray from moving, if the control unit sends a rejecting instruction to the rejecting unit, the mechanical claw is controlled to place the filament ingots with the collected images into the side edge backflow frame, the rejecting unit retracts the baffle, the mechanical claw retracts to the initial position, and the rejecting unit waits for the next paper tube to arrive; if the control unit sends a good product instruction to the removing unit, the removing unit withdraws the baffle when receiving the instruction, the filament ingot passes through the baffle, and the mechanical claw retracts to the initial position.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910625029.1A CN110813795B (en) | 2019-07-11 | 2019-07-11 | Device and method for detecting breakage of chemical fiber spindle paper tube on line |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910625029.1A CN110813795B (en) | 2019-07-11 | 2019-07-11 | Device and method for detecting breakage of chemical fiber spindle paper tube on line |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110813795A true CN110813795A (en) | 2020-02-21 |
| CN110813795B CN110813795B (en) | 2022-02-22 |
Family
ID=69547624
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910625029.1A Active CN110813795B (en) | 2019-07-11 | 2019-07-11 | Device and method for detecting breakage of chemical fiber spindle paper tube on line |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110813795B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111337506A (en) * | 2020-03-30 | 2020-06-26 | 河南科技学院 | Intelligent device for clothes quality inspection |
| CN111659633A (en) * | 2020-06-05 | 2020-09-15 | 珠海格力智能装备有限公司 | Tooling product classification system and method |
| CN112418276A (en) * | 2020-11-03 | 2021-02-26 | 北京五八信息技术有限公司 | Processing method and device of classifier |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101614528A (en) * | 2009-07-31 | 2009-12-30 | 上海微曦自动控制技术有限公司 | Pipe is to tube vision inspection device |
| CN103041992A (en) * | 2012-12-18 | 2013-04-17 | 深圳市神拓机电设备有限公司 | On-line detection device for carrier roller semifinished product |
| CN106124520A (en) * | 2016-08-26 | 2016-11-16 | 武汉捷普瑞科技有限公司 | A kind of full-automatic loading and unloading and the device of vision-based detection |
| CN206689060U (en) * | 2017-04-11 | 2017-12-01 | 佛山海悦智达科技有限公司 | A kind of pipe fitting quality detection device |
| CN107486417A (en) * | 2017-10-12 | 2017-12-19 | 重庆远创光电科技有限公司 | A kind of detection machine for being easy to detect rubber tube open defect |
| JP2018017639A (en) * | 2016-07-29 | 2018-02-01 | 株式会社 深見製作所 | Surface defect inspection method and surface defect inspection device |
| CN108891843A (en) * | 2018-07-17 | 2018-11-27 | 浙江盛达机器人科技有限公司 | A kind of dacron thread sleeve rotating device |
-
2019
- 2019-07-11 CN CN201910625029.1A patent/CN110813795B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101614528A (en) * | 2009-07-31 | 2009-12-30 | 上海微曦自动控制技术有限公司 | Pipe is to tube vision inspection device |
| CN103041992A (en) * | 2012-12-18 | 2013-04-17 | 深圳市神拓机电设备有限公司 | On-line detection device for carrier roller semifinished product |
| JP2018017639A (en) * | 2016-07-29 | 2018-02-01 | 株式会社 深見製作所 | Surface defect inspection method and surface defect inspection device |
| CN106124520A (en) * | 2016-08-26 | 2016-11-16 | 武汉捷普瑞科技有限公司 | A kind of full-automatic loading and unloading and the device of vision-based detection |
| CN206689060U (en) * | 2017-04-11 | 2017-12-01 | 佛山海悦智达科技有限公司 | A kind of pipe fitting quality detection device |
| CN107486417A (en) * | 2017-10-12 | 2017-12-19 | 重庆远创光电科技有限公司 | A kind of detection machine for being easy to detect rubber tube open defect |
| CN108891843A (en) * | 2018-07-17 | 2018-11-27 | 浙江盛达机器人科技有限公司 | A kind of dacron thread sleeve rotating device |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111337506A (en) * | 2020-03-30 | 2020-06-26 | 河南科技学院 | Intelligent device for clothes quality inspection |
| CN111659633A (en) * | 2020-06-05 | 2020-09-15 | 珠海格力智能装备有限公司 | Tooling product classification system and method |
| CN112418276A (en) * | 2020-11-03 | 2021-02-26 | 北京五八信息技术有限公司 | Processing method and device of classifier |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110813795B (en) | 2022-02-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109454006B (en) | Detection and classification method based on device for online detection and classification of chemical fiber spindle tripping defects | |
| CN110813795B (en) | Device and method for detecting breakage of chemical fiber spindle paper tube on line | |
| Sahu et al. | Defect identification and maturity detection of mango fruits using image analysis | |
| CN102221559B (en) | Online automatic detection method of fabric defects based on machine vision and device thereof | |
| JP2583146B2 (en) | Top cleanliness inspection method | |
| CN111402226A (en) | Surface defect detection method based on cascade convolution neural network | |
| Sahu et al. | Identification and classification of mango fruits using image processing | |
| CN114757900B (en) | Artificial intelligence-based textile defect type identification method | |
| CN109550712A (en) | A kind of chemical fiber wire tailfiber open defect detection system and method | |
| CN110403232B (en) | Cigarette quality detection method based on secondary algorithm | |
| CN108038838A (en) | A kind of cotton fibriia species automatic testing method and system | |
| CN104792789A (en) | Chemical fiber paper tube appearance detection device and method | |
| CN104048966B (en) | The detection of a kind of fabric defect based on big law and sorting technique | |
| CN106780464A (en) | A kind of fabric defect detection method based on improvement Threshold segmentation | |
| CN104809725A (en) | Cloth defect visual identify detecting device and method | |
| CN113538424A (en) | Wood board joint classification data identification method based on artificial intelligence | |
| CN106093051A (en) | Paper roll tangent plane burr detection method based on machine vision and device | |
| CN113916127A (en) | Visual inspection system and method for appearance of valve guide pipe finished product | |
| CN107121063A (en) | The method for detecting workpiece | |
| CN110694940A (en) | Control method and system for adjusting blowing of spray valve in real time based on dead pixel and size | |
| CN115672778A (en) | Intelligent visual recognition system | |
| CN107833222A (en) | A kind of non-metal workpiece fifth wheel detection means and method | |
| CN114764802A (en) | Equipment defect detection repeated image eliminating method | |
| CN114820453A (en) | Method for detecting surface flaws of packaged filament based on deep learning | |
| Lei | A machine vision system for inspecting bearing-diameter |
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 |