WO2020103324A1

WO2020103324A1 - On-line mask detection system and method

Info

Publication number: WO2020103324A1
Application number: PCT/CN2019/071404
Authority: WO
Inventors: 曾庆好; 马亮; 马博文; 赵东宁; 张勇; 汤奇; 马少勇; 袁宏焱
Original assignee: 深圳市维图视技术有限公司
Priority date: 2018-11-20
Filing date: 2019-01-11
Publication date: 2020-05-28
Also published as: JP2021513057A; JP7055212B2

Abstract

An on-line mask detection system, comprising: a cabinet (5); a conveying mechanism (1) for driving a mask to be detected to move; a turnover mechanism (3) for turning over said mask by 180 degrees; an image acquisition mechanism (2) for acquiring a front-face image and a reverse-face image of said mask; image processing software (6) for comparing the front-face image and the reverse-face image with a pre-set mask standard to determine whether said mask is a good product or a defective product; a sorting mechanism (4) for distinguishing defective products from good products and gathering same separately; and a PLC control system (7) electrically connected to the conveying mechanism (1), the turnover mechanism (3), the image acquisition mechanism (2), the image processing software (6) and the sorting mechanism (4), respectively. The on-line detection system has the advantages of a high degree of automation, high detection efficiency, guaranteed mask quality and low costs.

Description

Mask online detection system and method

Technical field

The invention belongs to the technical field of mask production, and more specifically relates to a mask online detection system and method.

Background technique

With the process of industrialization and urbanization, air pollution is becoming more and more serious, and masks have become an indispensable daily necessity. In the mass production process of masks, due to various reasons such as raw materials, production machinery and personnel operations, masks will be produced. There are many kinds of defective mask products, such as wrong mask size, different ear strap lengths, nose strips not installed, dirty masks, etc. If such defective products flow to the market, it will cause a lot of inconvenience to users. Will cause economic losses to mask manufacturers. Under normal circumstances, mask manufacturers only rely on the human eye to distinguish one by one, and the efficiency is extremely low. As a result, the current mask quality inspection process is time-consuming and labor-intensive, and it is easy to cause visual fatigue to the inspectors, which causes hidden quality problems in the product and causes a large number of product rework and Raw material waste.

technical problem

The purpose of the embodiments of the present invention is as follows: In the first aspect, an online mask detection system is provided to solve the technical problems of low efficiency, high cost, and unqualified quality of masks in the prior art.

In a second aspect, an online mask detection method is provided to solve the technical problems of low detection efficiency, high cost, and unqualified quality of masks in the prior art through manual detection.

Technical solution

To solve the above technical problems, the technical solutions adopted by the embodiments of the present invention are:

In the first aspect, an online mask detection system is provided, which includes: a cabinet; a conveying mechanism provided on the cabinet to drive the mask to be tested to move; a flip mechanism provided on the cabinet to flip the mask to be detected 180 ° ; The image collection mechanism installed on the cabinet is used to collect the front and back images of the mask to be detected; the image processing software installed on the cabinet is used to collect the front and back images collected by the image collection mechanism and the preset mask The standard is compared to judge the mask to be detected as good or bad; the sorting mechanism provided at the discharge end of the conveying mechanism is used to separate the bad and good products; the PLC control system installed on the cabinet is separate from the conveying mechanism , Flip mechanism, image acquisition mechanism, image processing software and sorting mechanism are electrically connected.

In a second aspect, a method for online detection of masks is provided, which includes: S1. Collecting the background image of No. 1 without the mask to be detected at the detection position No. 1 and collecting the detection of No. 2 by an image acquisition mechanism electrically connected to the image processing software There is no background image of the mask to be detected at the location, and the first background image and the second background image are converted into the first gray background image and the second gray background image respectively; S2, the conveying mechanism conveys the mask to be detected to At the No. 1 detection location, the No. 1 detection image containing the mask to be detected is collected by the image acquisition mechanism at the No. 1 detection location. The image processing software identifies the No. 1 detection image and compares it with the preset mask standard. If it is a defective product, the mask to be detected is collected as a defective product through the sorting mechanism. If it is judged as a good product, the next step is performed; S3. The good product is flipped 180 ° through the flip mechanism, and the conveying mechanism will flip the good product after 180 ° Convey to the No. 2 inspection location; S4. The image collection agency collects the No. 2 inspection image containing good products at the No. 2 inspection location. The image processing software recognizes the No. 2 inspection image and compares it with the preset mask standard. If the product is judged to be defective, it will be collected. If it is judged to be good, it will be collected.

Beneficial effect

Compared with the prior art, the beneficial effect of an on-line mask detection system provided by embodiments of the present invention is that the on-line mask detection system automatically transports the mask to be detected under the image acquisition mechanism through the transport mechanism, and the image acquisition mechanism will acquire The image is transferred to the image processing software for comparison processing, and the image processing software feeds back the information that the processing result is good or bad product to the PLC control system. The PLC control system controls the sorting mechanism to distinguish and collect the good or bad product, so that the present invention provides The on-line inspection system has the advantages of high automation, high inspection efficiency, guaranteed mask quality and low cost.

The beneficial effect of an on-line mask detection method provided by an embodiment of the present invention is that the mask detection method is implemented based on the above-mentioned mask online detection system, which makes the online detection method provided by the present invention highly automated, high detection efficiency, and mask quality It has the advantages of protection and labor cost saving.

BRIEF DESCRIPTION

In order to more clearly explain the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required in the embodiments or the description of the prior art. Obviously, the drawings in the following description are only for the invention. For some embodiments, those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts.

1 is a schematic diagram of an online detection system provided by an embodiment of the present invention;

2 is a schematic top plan view of an online detection system in an embodiment of the present invention;

3 is a schematic diagram of a background diagram of No. 1 in an embodiment of the present invention;

4 is a schematic diagram of No. 1 detection diagram in an embodiment of the present invention;

5 is a schematic diagram of the grayscale image of the front of the mask in the embodiment of the present invention;

FIG. 6 is a schematic diagram of the grayscale shift diagram of No. 1 detection in the embodiment of the present invention;

7 is a schematic diagram of a grayscale rotation diagram of No. 1 detection in an embodiment of the present invention;

FIG. 8 is a schematic diagram of rule configuration of an ear band detection rule frame, a dirt detection rule frame and a nose strip detection rule frame in an embodiment of the present invention.

Among them, the reference numbers in the drawings are as follows:

1. Conveying mechanism; 11. Conveying motor; 12. Conveying belt; 13. First optical fiber sensor; 14. Blower belt mechanism; 141, upper clamping belt; 142, lower clamping belt; 143, solenoid valve; 144, Blowpipe; 2. Image acquisition mechanism; 21, image acquisition module; 211, camera; 212, lens; 213, light source; 22, bracket; 23, second fiber sensor; 24, third fiber sensor; 25, linear motor ; 26, linear guide; 3. Flip mechanism; 31, flip motor; 32, flip rod; 321, main rod; 322, support rod; 4, sorting mechanism; 41, defective product processing mechanism; 411, fourth fiber sensor 412, first cylinder; 42, defective product collection box; 43, good product processing mechanism; 431, fifth optical fiber sensor; 432, second cylinder; 44, good product collection mechanism; 441, good product collection belt; 442, drive motor; 5. Cabinet; 6. Image processing software; 7. PLC control system; E1, ear band detection rule frame; D1, dirt detection rule frame; N1, nose strip detection rule frame.

Embodiments of the invention

In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

It should be noted that when a component is referred to as being “fixed” or “disposed on” another component, it can be directly on another component or indirectly on the other component. When a component is said to be "connected to" another component, it can be directly connected to the other component or indirectly connected to the other component.

It should also be noted that the same or similar reference numerals in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if there are terms “upper”, “lower”, and “ The orientation or positional relationship indicated by "left", "right", etc. is based on the orientation or positional relationship shown in the drawings, only to facilitate the description of the present invention and simplify the description, and does not indicate or imply that the device or element referred to must have a specific The orientation and construction and operation in a specific orientation are used. Therefore, the terms used to describe the positional relationship in the drawings are for illustrative purposes only, and cannot be construed as a limitation of this patent. For those of ordinary skill in the art, they can Understand the specific meaning of the above terms.

In addition, the terms “first” and “second” are used for description purposes only, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise specifically limited.

Please refer to FIG. 1-2, an embodiment of the present invention provides an online mask detection system, including: cabinet 5, conveying mechanism 1, turnover mechanism 3, image acquisition mechanism 2, image processing software 6, sorting mechanism 4 and PLC Control system 7.

Among them, the cabinet 5 is used as a main body for installing the conveying mechanism 1, the turnover mechanism 3, the image acquisition mechanism 2, the image processing software 6, the sorting mechanism 4 and the PLC control system 7.

Among them, the conveying mechanism 1 is used to drive the detection mask to move on the cabinet 5 to sequentially convey the mask to be detected to the image collecting mechanism 2, the turning mechanism 3, and the sorting mechanism 4 in sequence.

Wherein, the flip mechanism 3 is used to flip the mask to be detected 180 °, that is, to flip the face-up of the mask to be detected upside-down, so that the image acquisition mechanism 2 collects the front and back images of the mask to be detected respectively.

Among them, the image processing software 6 is used to compare the front image and the back image collected by the image collection mechanism 2 with the preset mask standard. If the front image is within the preset mask standard range, the mask to be detected is good. If the image and / or the reverse image are not within the preset mask standard range, the mask to be detected is a defective product. In this embodiment, the mask standard is set according to the quality requirements of each manufacturer, which is not limited here.

Among them, the sorting mechanism 4 is used to collect defective products and good products separately in order to ensure the quality of the good products and the recycling of the bad products.

Among them, the PLC control system 7 is electrically connected to the conveying mechanism 1, the turnover mechanism 3, the image acquisition mechanism 2, the image processing software 6 and the sorting mechanism 4, respectively, and the PLC control system 7 is used to realize the automatic detection of the mask online detection system. Therefore, the detection efficiency and quality of the mask to be detected are improved, and labor cost is saved.

Specifically, in this embodiment, the conveying mechanism 1 includes: a conveying belt 12 provided on the cabinet 5, a conveying motor 11, a first optical fiber sensor 13, and an ear-blowing belt mechanism 14. In this embodiment, the interior of the cabinet 5 is hollow, and an opening is provided on the upper surface of the cabinet 5. The opening is used to install the conveying mechanism 1 and facilitate the turning movement of the turning mechanism 3.

Among them, two conveyor belts 12 are provided, and two conveyor belts 12 are arranged on the cabinet 5 at intervals. The two ends of the mask to be detected are placed horizontally on the two conveyor belts 12, and the driving motor 442 is used to drive the conveyor belt 12 to run. The ear strap mechanism 14 is used to blow the ear strap to expand the ear strap for easy detection. The first optical fiber sensor 13 is used to sense the mask to be detected. After the mask to be detected moves to the ear strap mechanism 14, the first The optical fiber sensor 13 feeds back the induction signal sensed to the mask to be detected to the PLC control system 7, and the PLC control system 7 controls the ear-blowing belt mechanism 14 to perform the air blowing action on the ear belt. In other embodiments, the conveyor belt 12 may be provided with three, four, or five belts.

More specifically, in this embodiment, the ear-blowing belt mechanism 14 includes: rotating an upper clamping belt 141 and a lower clamping belt 142 provided on the cabinet 5 for pulling the mask to be detected, for performing An air blowing pipe 144 for blowing air, and a solenoid valve 143 for controlling the blowing of the air blowing pipe 144. The solenoid valve 143 is electrically connected to the PLC control system 7. In this embodiment, the solenoid valve 143 is a blow-type solenoid valve 143 , Used to send compressed air into the air blowing tube 144 to expand the earbands at both ends of the mask to be tested. In other embodiments, the air blowing tube 144 is connected to an air blowing pump.

Among them, the upper clamping belt 141 and the lower clamping belt 142 are used to clamp the mask to be detected, to prevent the mask to be detected from being blown away from the conveying mechanism 1 by the air blowing pipe 144, and also to move the mask to be detected to move. In this implementation, the upper clamping belt 141 and the lower clamping belt 142 are installed on the cabinet 5 through two sets of roller shafts, and the two sets of roller shafts are rotatably installed on the cabinet on the cabinet 5, and the cabinet is driven by two sets of roller shafts Rotating stepper motor.

In this embodiment, the turning mechanism 3 includes: turning the turning lever 32 provided on the cabinet 5 and a turning motor 31 for driving the turning lever 32 to turn. Wherein, the turning bar 32 includes: a main bar 321 whose extension direction is perpendicular to the conveying direction of the mask to be detected, and at least two supporting bars 322 arranged vertically on the main bar 321, and the supporting bar 322 is between the two conveying belts 12 The turning rod 322 is used to turn the mask to be detected 180 ° during turning. In other embodiments, the flip lever 32 may be replaced with a flip board.

In this embodiment, the image acquisition mechanism 2 includes: a bracket 22 provided on the cabinet 5, a linear guide 26 provided on the support 22, and an image acquisition module 21 movably provided on the linear guide 26, provided on the support 22 The linear motor 25 for driving the image acquisition module 21 to reciprocate on the linear guide 26 and the second optical fiber sensor 23 and the third optical fiber sensor 24 spaced on the cabinet 5; wherein, the turning mechanism 3 is located Between the second optical fiber sensor 23 and the third optical fiber sensor 24.

The sensing position of the second optical fiber sensor 23 is set to the first detection position, and the sensing position of the third optical fiber sensor 24 is set to the second detection position. The first detection position is used by the image acquisition module 21 to collect the front image of the mask to be detected The second detection position is used by the image collection module 21 to collect the reverse image of the mask to be detected.

Specifically, the image acquisition module 21 includes: a camera 211 disposed on the linear guide 26, electrically connected to the PLC control system 7 and the image processing software 6, a lens 212 disposed on the camera 211, and a camera 211 The light source 213 on the top; wherein, the light of the light source 213 is irradiated to the mask to be detected.

In this embodiment, the linear guide 26 and the linear motor 25 adopt a screw structure to realize the reciprocating movement of the image acquisition module 21 between the first detection position and the second detection position.

When the second optical fiber sensor 23 senses the mask to be detected, the camera 211 takes a picture to collect a frontal image and passes it to the image processing software 6. If the image processing software 6 determines that it is a defective product, it waits for the next mask to be detected. If the image processing software 6 If it is judged as a good product, the linear motor 25 drives the camera 211 to move to the second detection position. At this time, the mask to be detected moves to the turning mechanism 3 to be turned 180 °. When the third fiber sensor 24 senses the turned mask to be detected , The camera 211 takes a picture and collects the reverse image and passes it to the image processing software 6 for judgment.

Specifically, in this embodiment, the light source 213 is a central aperture type surface light source 213, and the lens 212 is placed in the center of the light source 213, so that the setting is convenient for the light source 213 to irradiate the mask to be detected without shadows, effectively ensuring that the camera 211 takes pictures Clarity.

In this embodiment, the sorting mechanism 4 includes: a defective product processing mechanism 41 and a good product processing mechanism 43 arranged side by side above the conveying mechanism 1, and a defective product collection box 42 and a good product collection mechanism 44 arranged side by side below the conveying mechanism 1 ; Among them, the defective product processing mechanism 41 is used to transfer the defective products to the defective product collection box 42, and the good product processing mechanism 43 is used to transfer the good products to the good product collection mechanism 44.

In this embodiment, the defective product processing mechanism 41 includes: a fourth optical fiber sensor 411 electrically connected to the PLC control system 7 and a first air cylinder 412. The first air cylinder 412 is used to store defective products located on the two conveyor belts 12 Push it into the defective product collection box 42.

In this embodiment, the good product processing mechanism 43 includes: a fifth optical fiber sensor 431 and a second cylinder 432, and the good product collection mechanism 44 includes: a good product collection belt 441 disposed below the conveying belt 12, and a device for driving the good product collection belt 441 to operate The driving motor 442, the second cylinder 432 are used to push the good products onto the good product collection belt 441, the fifth optical fiber sensor 431, the second cylinder 432 and the driving motor 442 are electrically connected to the PLC control system 7. When the number of good products collected at the same position on the good product collection belt 441 reaches a predetermined number, the drive motor 442 drives the good product collection belt 441 to run one to two mask widths to be detected, thereby continuing to collect good products.

In this embodiment, two conveyor belts 12 extend out of one end of the cabinet 5 through the support frame, and the first cylinder 412, the second cylinder 432, the fourth fiber sensor 411, and the fifth fiber sensor 431 are disposed on the conveyor belt 12 through the support frame Above and between two conveyor belts 12. The extending direction of the piston rods of the first cylinder 412 and the second cylinder 432 is set downward so as to push the mask to be detected downward into the defective product collection box 42 or the good product collection mechanism 44. In this embodiment, the conveying motor 11 is a stepping motor or a servo motor, the turning motor 31 is a stepping motor, and the driving motor 442 is a speed regulating motor.

As shown in FIGS. 1-8, this embodiment also provides a mask online detection method, including:

S1. Through the image acquisition mechanism 2 electrically connected to the image processing software 6, collect the first background image without the mask to be detected at the first detection position and the second background image without the mask to be detected at the second detection position, and The background image No. 1 and the background image No. 2 are converted into the gray image No. 1 background and the gray image No. 2 background, respectively;

S2. The conveying mechanism 1 conveys the mask to be detected to the No. 1 detection position. The image acquisition mechanism 2 collects the No. 1 detection map containing the mask to be detected at the No. 1 detection position. The image processing software 6 recognizes the No. 1 detection map And compare with the preset mask standard. If it is judged as a defective product, the mask to be detected is collected as a defective product through the sorting mechanism 4. If it is judged as a good product, the next step is performed;

S3. The good product is turned 180 ° through the turning mechanism 3, and the conveying mechanism 1 conveys the good product turned 180 ° to the second detection position;

S4. The image collection mechanism 2 collects the second inspection map containing good products at the second inspection location. The image processing software 6 recognizes the second inspection map and compares it with the preset mask standard. If it is judged as a defective product, Collect defective products. If judged as good products, collect good products.

In this embodiment, step S2 specifically includes the following processing steps: S21. Mark the four vertices of the No. 1 detection map as P1 (0, 0), P2 (w, 0), P3 (0, h), P4 ( w, h) and the central point Pc0 (w / 2, h / 2), and through the image processing software 6, the No. 1 detection map is converted into the No. 1 detection gray map;

S22. Perform image foreground processing on the grayscale image of No. 1 detection and the grayscale image of the background of No. 1 to obtain a grayscale image on the front of the mask, and binarize the grayscale image on the front of the mask to obtain a binary image on the front of the mask; The center point of the grayscale image on the front of the mask is Pc1 (cx, cy);

S23. Perform a fixed number of image dilation and image erosion on the frontal binarized image of the mask, and use the contour extraction algorithm to extract the frontal contour of the mask, select the frontal contour of the mask with the largest area, and use the smallest external rectangle algorithm for the largest frontal wheel of the mask Extract the first smallest circumscribed rectangle on the front of the mask;

S24. If the extraction of the first minimum circumscribed rectangle fails, it is judged that there is no mask processing; if the extraction of the first minimum circumscribed rectangle is successful, the length and width values of the first minimum circumscribed rectangle are calculated, and the length and width values are compared with the pre-respectively. The set mask standard is subtracted to obtain the length difference and height difference. If the length difference and / or height difference exceeds the predetermined mask standard range, it is judged as a defective product, otherwise, the next step is performed.

S25. Output the four vertex coordinates P11, P12, P13, P14 of the first smallest rectangle, and the center point coordinates Pc1 (cx, cy), calculate the line L1 (P13, P14) and the line L2 (P3, P4) Between the included angle A1, the horizontal offset Hx1 and the vertical offset Vy1 between the center point Pc1 (cx, cy) and the center point Pc0 (w / 2, h / 2) are calculated;

S26. The grayscale image of No. 1 is translated by Hx1 in the horizontal direction and Vy1 in the vertical direction, so that the central point Pc1 (cx, cy) and the central point Pc0 (w / 2, h / 2) coincide, and a No. 1 detection gray shift map; among them, the four vertices of No. 1 detection gray shift map are P21, P22, P23, P24, and the center point coordinates Pc2 (cx, cy), at this time cx is equal to w / 2 value, The cy and h / 2 values are equal;

S27. Rotate the No.1 detection grayscale translation image with the center point Pc0 (w / 2, h / 2) as the origin, and the rotation angle is the degree of the included angle A1, to obtain the No.1 detection grayscale rotation image; The four vertices of the grayscale rotation map are P31, P32, P33, P34, and the center point is Pc2 (cx, cy);

S28. Compare and judge the grayscale rotation image of No. 1 detection with the preset mask standard. If it is judged to be a defective product, the mask to be detected is collected by the sorting mechanism 4 as a defective product. If it is judged as a good product, it is collected as a good product. .

In this embodiment, step S28 includes the following comparison judgment step:

S281. Use two preset earband detection rule frames E1 to detect the earband defects on the front of the mask: use the contour extraction algorithm to calculate and judge the outline attributes of the earband image in the earband detection area. If it is not within the predetermined mask standard range, it is determined that the ear strap is bad;

S282. Use the pre-set dirt detection rule frame D1 to detect the dirt defects on the front of the mask: use a fixed threshold binarization algorithm to process the dirt detection area, and count the number of zeros in the image after binarization. If the number of zero points is not within the predetermined mask standard range, it is judged to be dirty;

S283. Use the preset nose strip detection rule frame N1 to detect the nose strip defects on the front of the mask: use the contour extraction algorithm to calculate and judge the contour attributes of the nose strip image in the nose strip detection area. Within the standard range of the mask, it is determined that the nose strip is bad. If the attribute of the contour of the nose strip image is within the preset standard range of the mask, the minimum circumscribed rectangle algorithm is used to calculate the minimum circumscribed rectangle of the nose strip, and the minimum output The coordinates of the four vertices and the center point of the circumscribed rectangle determine whether the coordinates of the four vertices and the center point of the smallest circumscribed rectangle of the nose bar are within the preset standard range of the mask, if they are not within the preset standard range of the mask, It is judged that the nose strip is bad; when none of the masks to be tested at the detection position No. 1 have bad ear straps, dirty dirt and bad nose strips, the mask to be tested is good. The detection order of steps S281, S282, and S283 is in no particular order.

In this embodiment, the processing procedures and methods of step S4 and step S2 are the same, and the drawings are also the same, which will not be provided here. Among them, step S4 includes the following processing steps: S41, the four vertices of the second detection map are also marked as P1 (0, 0), P2 (w, 0), P3 (0, h), P4 (w, h ) And the central point Pc0 (w / 2, h / 2), and through the image processing software 6 to convert the No. 2 detection map into the No. 2 detection grayscale map;

S42: Perform image foreground processing on the grayscale image of the second detection and the grayscale image of the second background to obtain the grayscale image of the reverse side of the mask, and perform binarization of the grayscale image of the reverse side of the mask to obtain a binary image of the reverse side of the mask;

S43. Perform a fixed number of image dilation and image erosion processing on the reverse image of the mask, and use the contour extraction algorithm to extract the contour of the mask, select the contour of the mask with the largest area, and use the smallest external rectangle algorithm for the largest mask. Extract the second smallest circumscribed rectangle on the reverse side of the mask;

S44. If the extraction of the second smallest circumscribed rectangle fails, it is judged that there is no mask processing; if the extraction of the second smallest circumscribed rectangle is successful, the length value and width value of the second smallest circumscribed rectangle are calculated, and the length value and the width value The set mask standard is subtracted to obtain the length difference and height difference. If the length difference and / or height difference exceeds the predetermined mask standard range, it will be judged as a defective product, otherwise, proceed to the next step;

S45, output the four vertex coordinates P11, P12, P13, P14 of the second smallest rectangle, and the center point coordinates Pc1 (cx, cy), calculate the connection line L3 (P13, P14) and the connection line L4 (P3, P4) Between the included angle A2, the horizontal offset Hx2 and the vertical offset Vy2 between the center point Pc1 (cx, cy) and the center point Pc0 (w / 2, h / 2) are calculated;

S46. The grayscale image of No. 1 is translated by Hx2 in the horizontal direction and Vy2 in the vertical direction, so that the center point Pc1 (cx, cy) and the center point Pc0 (w / 2, h / 2) coincide, and two No. detection grayscale shift map;

S47. Rotate the No. 2 detection grayscale translation map with the center point Pc0 (w / 2, h / 2) as the origin, and the rotation angle is the degree of the included angle A2, to obtain the No.2 detection grayscale rotation map;

S48. Compare and judge the grayscale rotation map of No. 2 detection with the preset mask standard as follows:

Use the preset two earband detection rule frames E1 to detect the earband defects on the reverse side of the mask: use the contour extraction algorithm to calculate and judge the earband image outline attributes of the earband detection area, if the earband image outline attributes are not predetermined Within the standard range of the mask, it is judged that the ear strap is bad;

Use the pre-set dirt detection rule frame D1 to detect the dirt defects on the reverse side of the mask: use a fixed threshold binarization algorithm to process the dirt detection area, count the number of zeros in the image after the binarization, if the number of zeros If it is not within the predetermined mask standard, it is judged to be dirty;

The nose strip detection rule frame N1 set in advance is used to detect the nose strip defects on the reverse side of the mask: the contour extraction algorithm is used to calculate and judge the contour attributes of the nose strip image in the nose strip detection area. If the contour attribute of the nose strip image is not within the preset mask standard range, it is determined that the nose strip is bad. If the attribute of the contour image of the nose strip is within the preset mask standard range, the minimum external rectangle algorithm is used to calculate the minimum external connection of the nose strip Rectangle, and output the coordinates of the four vertices and center of the smallest circumscribed rectangle to determine whether the coordinates of the four vertices and the coordinates of the center of the smallest circumscribed rectangle are within the preset standard range of the mask, if they are not within the preset standard range of the mask It is judged that the nose strip is bad; when the mask to be tested at the second detection position does not show bad ear straps, dirty dirt and bad nose strips, the mask to be tested is good.

In this embodiment, the image foreground processing includes: subtracting the pixel values at the same position of the grayscale image No. 1 and the grayscale background image No. 1 to obtain the first absolute value, and detecting the grayscale image No. 2 and No. 2 The pixel values at the same position of the background grayscale image are subtracted to obtain the second absolute value. If the first absolute value is within the preset mask standard range, the pixel value at that position is set to 255; otherwise, if the first absolute value is greater than the preset mask standard range, the pixel value at that position is The pixel value of the degree graph. If the second absolute value is within the preset mask standard range, the pixel value at that position is set to 255; otherwise, if the second absolute value is greater than the preset mask standard range, then the pixel value at that position uses the No. 2 detection gray The pixel value of the degree graph.

The on-line mask detection device and method provided by the embodiments of the present invention automatically transport the mask to be detected, and detect the front and back sides of the mask to be detected, thereby achieving the advantages of high specific detection efficiency and high degree of automation.

Claims

An on-line mask detection system, characterized in that it includes: a cabinet (5); a conveying mechanism (1) provided on the cabinet (5) for driving the mask to be detected to move; it is provided on the cabinet (5) The flipping mechanism (3) on the top is used to flip the mask to be detected 180 °; the image acquisition mechanism (2) provided on the cabinet (5) is used to collect the front image and the back side of the mask to be tested An image; an image processing software (6) provided on the cabinet (5), for comparing the front image and the back image collected by the image collection mechanism (2) with a preset mask standard, To judge the mask to be detected as a good product or a defective product; a sorting mechanism (4) provided at the discharge end of the conveying mechanism (1) for distinguishing and collecting the defective product and the good product; The PLC control system (7) on the cabinet (5) is respectively connected with the conveying mechanism (1), the turning mechanism (3), the image acquisition mechanism (2), the image processing software (6) and The sorting mechanism (4) is electrically connected.
The on-line mask detection system according to claim 1, characterized in that: the conveying mechanism (1) includes: at least two conveying belts (12) arranged on the cabinet (5) at intervals, the A detection mask is placed horizontally on at least two of the conveyor belts (12); a conveyor motor (11) for driving the conveyor belt (12) to operate; and a sensor provided on the cabinet (5) for sensing the A first optical fiber sensor (13) of the mask to be detected; and an ear-blowing band mechanism (14) provided on the cabinet (5) for blowing the ear band.
The on-line mask detection system according to claim 2, characterized in that the ear-blowing belt mechanism (14) includes: a rotary set on the cabinet (5) for pulling the mask to be tested The upper clamping belt (141) and the lower clamping belt (142) are used for blowing the ear tube (144) for blowing the ear band, and a solenoid valve (143) for controlling the blowing of the blowing tube (144).
The on-line mask detection system according to claim 3, characterized in that the turning mechanism (3) includes: turning the turning lever (32) provided on the cabinet (5), and A reversing motor (31) rotating by the reversing rod (32); the reversing rod (32) includes: a main rod (321) whose extension direction is perpendicular to the conveying direction of the mask to be detected, and at least two vertical A supporting rod (322) on the main rod (321), the supporting rod (322) is turned between the two conveying belts (12).
21 according to claim 4, characterized in that an on-line mask detection system according to claim 1, characterized in that: the image acquisition mechanism (2) includes: disposed on the cabinet (5) Support (22); a linear guide (26) provided on the support (22); an image acquisition module (21) movably provided on the linear guide (26); provided on the support (22) ), A linear motor (25) for driving the image acquisition module (21) to reciprocate on the linear guide rail (26); and a second optical fiber sensor (20) spaced on the cabinet (5) 23) and a third optical fiber sensor (24); the turning mechanism (3) is located between the second optical fiber sensor (23) and the third optical fiber sensor (24).
An on-line mask detection system according to claim 5, characterized in that the image acquisition module (21) includes: the PLC control system (7) and the PLC control system (7) provided on the linear guide (26) A camera (211) electrically connected to the image processing software (6); a lens (212) provided on the camera (211); and a light source (213) provided on the camera (211); The light from the light source (213) irradiates the mask to be detected.
The on-line mask detection system according to claim 6, characterized in that the light source (213) is a central aperture type surface light source (213), and the lens (212) is passed through the light source (213) center of.
The online mask detection system according to claim 2, characterized in that the sorting mechanism (4) comprises: a defective product processing mechanism (41) and a good product processing arranged side by side above the conveying mechanism (1) Mechanism (43); and, a defective product collection box (42) and a good product collection mechanism (44) disposed side by side under the conveying mechanism (1); the defective product processing mechanism (41) is used to transfer defective products to In the defective product collection box (42), the good product processing mechanism (43) is used to transfer the good product to the good product collection mechanism (44).
An on-line mask detection system according to claim 8, wherein the defective product processing mechanism (41) includes: a fourth optical fiber sensor (411) electrically connected to the PLC control system (7) and A first cylinder (412) for pushing defective products located on the conveyor belt (12) into the defective product collection box (42).
The on-line mask detection system according to claim 9, wherein the good product processing mechanism (43) includes: a fifth optical fiber sensor (431) and a second cylinder (432), and the good product collection mechanism (44) ) Includes: a good product collection belt (441) disposed below the conveyor belt (12), and a drive motor (442) for driving the good product collection belt (441) to operate, and the second cylinder (432) When pushing the good product onto the good product collection belt (441), the fifth fiber sensor (431), the second cylinder (432), the drive motor (442) and the PLC control system (7) Electrical connection.
A mask online detection method, which is characterized by comprising:

S1, through the image acquisition mechanism (6) electrically connected with the image processing software (6), collect the background image of No. 1 at the first detection position without the mask to be detected and collect the background of the No. 2 detection position at the second detection position Figure 1, and convert the background image No. 1 and the background image No. 2 into a gray background image No. 1 and a gray background image No. 2;

S2. The conveying mechanism (1) conveys the mask to be detected to the No. 1 detection position. The image acquisition mechanism (2) collects the No. 1 detection map containing the mask to be detected at the No. 1 detection position. The image processing software (6) The No. 1 inspection chart is identified and compared with the preset mask standard. If it is judged to be a defective product, the mask to be tested is collected as a defective product through a sorting mechanism (4), if it is judged to be a good product, then Go to the next step;

S3. The good product is turned 180 ° through the turning mechanism (3), and the conveying mechanism (1) conveys the good product turned 180 ° to the second detection position;

S4. The image acquisition mechanism (2) collects the second inspection map containing the good product at the second inspection location, and the image processing software (6) recognizes the second inspection map and matches the preset mask standard For comparison, if the product is judged to be defective, the product will be collected, and if the product is judged to be good, the product will be collected.
The online mask detection method according to claim 11, wherein the step S2 includes the following processing steps: S21, the four vertices of the No. 1 detection map are marked as P1 (0, 0), P2 ( w, 0), P3 (0, h), P4 (w, h) and the central point Pc0 (w / 2, h / 2), and the image No. 1 detection map is converted into a Number detection grayscale;

S22. Perform image foreground processing on the first detection gray image and the first background gray image to obtain a front gray image of the mask, and perform binarization processing on the front gray image of the mask to obtain a front two of the mask Valued image;

S23. Perform a fixed number of image dilation and image erosion processing on the frontal binarized image of the mask, and use a contour extraction algorithm to extract the front contour of the mask, select the front contour of the mask with the largest area, The wheel uses the minimum circumscribed rectangle algorithm to extract the first minimum circumscribed rectangle on the front of the mask;

S24. If the extraction of the first minimum circumscribed rectangle fails, it is judged that there is no mask processing;

If the extraction of the first minimum circumscribed rectangle is successful, the length and width of the first minimum circumscribed rectangle are calculated, and the length and width are subtracted from the preset mask standard to obtain the length difference and height difference Value, if the length difference and / or height difference exceeds the predetermined mask standard range, it is judged as a defective product, otherwise, proceed to the next step:

S25. Output the four vertex coordinates P11, P12, P13, P14 of the first smallest rectangle, and the center point coordinates Pc1 (cx, cy), calculate the connection line L1 (P13, P14) and the connection line L2 (P3, P4) between the angle A1, calculate the horizontal offset Hx1 and the vertical offset Vy1 between the center point Pc1 (cx, cy) relative to the center point Pc0 (w / 2, h / 2);

S26. Translate the grayscale image of No. 1 detection by Hx1 in the horizontal direction and Vy1 in the vertical direction, so that the center point Pc1 (cx, cy) and the center point Pc0 (w / 2, h / 2) coincide. Get the No.1 detection gray shift map;

S27. Rotate the first detection grayscale translation map with the center point Pc0 (w / 2, h / 2) as the origin, and the rotation angle is the degree of the included angle A1 to obtain the first detection grayscale rotation map;

S28. Compare and judge the grayscale rotation image of No. 1 detection with the preset mask standard. If it is judged as a defective product, collect the defective mask through the sorting mechanism (4) for the defective product. Good products are collected as good products.
The online mask detection method according to claim 12, characterized in that:

Step S28 includes the following judgment steps:

S281. Use the preset earband detection rule frame (E1) to detect the earband defects on the front of the mask: use the contour extraction algorithm to calculate and judge the earband image outline attributes of the earband detection area. If it is not within the predetermined mask standard range, it is determined that the ear strap is bad;

S282. Use the preset dirt detection rule frame (D1) to detect the dirt defects on the front of the mask: use a fixed threshold binarization algorithm to process the dirt detection area, and count the number of zeros in the image after binarization , If the number of zero points is not within the predetermined mask standard range, it is judged as dirty;

S283. Detect the nose strip defects on the front of the mask using the preset nose strip detection rule frame (N1): use the contour extraction algorithm to calculate and judge the contour attributes of the nose strip image in the nose strip detection area. If it is not within the preset mask standard range, it is judged that the nose strip is bad. If the attribute of the nose strip image contour is within the preset mask standard range, the smallest circumscribed rectangle algorithm is used to calculate the smallest circumscribed rectangle of the nose strip and output The coordinates of the four vertices and the center point of the smallest circumscribed rectangle of the nose are used to determine whether the coordinates of the four vertices and the center point of the smallest circumscribed rectangle of the nose bar are within the standard range of the mask, if they are not within the standard range of the mask Inside, it is judged as a bad nose;

When none of the masks to be detected at the detection position No. 1 has bad ear straps, dirty dirt, and bad nose strips, the masks to be tested are good products.
The on-line mask detection method according to claim 13, characterized in that: the processing procedures and methods of step S4 and step S2 are the same; when the mask to be tested at the second detection position has no bad earband or dirt In case of bad or bad nose strips, the mask to be tested is good.
An online mask detection method according to claim 14, wherein the image foreground processing includes: subtracting the pixel values at the same position of the grayscale image No. 1 and the grayscale background image No. 1 to obtain the first An absolute value, subtracting the pixel value at the same position of the grayscale image No. 2 and the grayscale image No. 2 background to obtain the second absolute value;

If the first absolute value is within the preset mask standard range, the pixel value at that position is set to 255; otherwise, if the first absolute value is greater than the preset mask standard range, the pixel value at that position is The pixel value of the degree graph;

If the second absolute value is within the preset mask standard range, the pixel value at that position is set to 255; otherwise, if the second absolute value is greater than the preset mask standard range, then the pixel value at that position uses the No. 2 detection gray The pixel value of the degree graph.