CN114527075A - Mask defect detection device and method - Google Patents

Abstract

The invention relates to the technical field of mask production, in particular to a mask defect detection device and a method, wherein the device comprises a device cabinet, and the device cabinet is respectively provided with: the feeding mechanism is used for providing the mask to be detected; the three-section type belt line conveying mechanism is used for conveying the mask to be detected; the image acquisition mechanism is used for acquiring an image of the mask to be detected; the image processing software judges the collected mask image according to a preset mask standard and a detection algorithm and gives a good product or defective product result; the rejecting mechanism rejects the defective products given by the image processing software; and the control system is used for electrically connecting the feeding mechanism, the conveying mechanism, the image acquisition mechanism, the image processing software and the rejecting mechanism. The mask defect detection method provided by the invention combines the traditional BLOB detection method and the deep learning mode to finish the detection of all items to be detected, and has the advantages of high stability, high detection efficiency and high detection precision, so that the quality of products detected by the device is guaranteed.

Description

A kind of mask defect detection device and method

technical field

The invention relates to the technical field of mask production, in particular to a mask defect detection device and method.

Background technique

With the acceleration of industrialization and urbanization, air pollution has become more and more intensified. As an important protective product for epidemic prevention and control, masks have become an indispensable daily dependency. In the process of mass production of masks, due to Due to various reasons such as raw materials, production machines, and personnel operations, there will be many types of defective mask products. Manual testing is to rely on a large number of online quality inspectors to conduct offline testing of masks. Although a certain quality control can be achieved, it is easy to Staff fatigue, time consuming and inefficiencies arise. The method of manual visual inspection is easily affected by subjective factors such as the technical quality, experience, visual resolution and fatigue of inspectors, lacks accuracy and standardization, and cannot guarantee normal product quality. These problems will directly lead to the inefficiency of missing detection and reduce the profits of manufacturers, which in turn will reduce the competitiveness of mask manufacturers' products, and may eventually be eliminated by the market. Therefore, the research and development of automatic defect detection device is imminent and urgent.

SUMMARY OF THE INVENTION

Aiming at the deficiencies of the prior art, the present invention provides a mask defect detection device and method, which solves the technical problems of low efficiency, high cost, and unguaranteed quality of the current mask detection through manual detection.

The present invention is achieved through the following technical solutions: a mask defect detection device, comprising a device cabinet, and the device cabinet is respectively installed with: a feeding mechanism, which provides masks to be tested; a three-section belt line conveying mechanism, which transports masks to be tested The image acquisition mechanism collects the image of the mask to be detected; the image processing software judges the collected mask image according to the preset mask standard and detection algorithm and gives the result of good or bad product; the rejection mechanism provides the image processing software The defective products are rejected; the control system is used to electrically connect the feeding mechanism, the conveying mechanism, the image acquisition mechanism, the image processing software and the rejection mechanism.

Preferably, the image acquisition mechanism includes: a bracket for fixing a camera; a 4K line scan camera installed on the bracket and electrically connected to the image processing software; an interface with the 4K line scan camera A matched fixed-focus FA lens with fine-tuning aperture and focal length; and a photoelectric detection switch and an encoder module electrically connected to the 4K line scan camera; wherein two groups of the 4K line scan cameras are located in each adjacent Just below and above the gap between the two belts.

Preferably, the image acquisition mechanism further includes: four groups of high-brightness linear light sources; wherein two groups of the high-brightness linear light sources are located directly above and below the gap between each adjacent two belts, and the other two groups of the high-brightness linear light sources are located respectively. The linear light sources are located at 45° on the side of the gap between each two adjacent belts, that is, the high-brightness linear light sources are in the backlight mode with the irradiation direction facing the FA lens and 45° from the central axis of the 4K line scan camera. ° front lighting method; complete the lighting of the mask to be detected.

Preferably, the feeding mechanism is used for docking with the upstream mask production conveyor line to receive the masks to be tested conveyed by the upstream mask production conveyor line.

Preferably, the conveying mechanism includes: a first conveying belt, a second conveying belt and a third conveying belt; and a belt for driving the first conveying belt, the second conveying belt and the third conveying belt to operate Servo motor; the mask to be tested is conveyed between the first conveying belt, the second conveying belt and the third conveying belt.

Preferably, the rejecting mechanism comprises: an air blowing cylinder located at the discharge end of the conveying mechanism and a defective product collection box located directly below the air blowing nozzle of the air blowing cylinder.

The present invention also provides the following technical solutions: a mask defect detection method, which is carried out by using the above-mentioned mask defect detection device, and the detection defect items of the detection method include but are not limited to the following five categories: mask empty bag/twin Mask/half, mask folds, mask membrane joints, mask earband integrity detection and mask pressure earband defect detection.

Preferably, the detection method comprises the following steps:

S1. Collect the background image of the mask to be detected and the mask image to be detected through a 4K line scan camera, and make a difference between the two images to obtain a difference image, which is recorded as Image_Sub;

S2. Crop the difference image Image_Sub from the original collected image to obtain the target mask area image Image_MaskRegion;

Among them, the steps of the detection method for the two detection items of mask empty bag/twin mask/half mask and mask fold include:

(a) to the image of the target mask region Image_MaskRegion through image grayscale, image binarization and other preprocessing operations, and use the region area and geometric shape features to screen out the image region of the type of defect to be determined, denoted as ToJudgeRegion, and the above The image area of the defect type to be determined is cropped from the original captured image;

(b) For the image area ToJudgeRegion of the type of defect to be judged, if the area and geometrical features of the area do not meet the preset feature range of the mask, it is directly judged as a defective product, if the area and geometrical characteristics of the area meet The preset feature range of the mask, the image of the type of defect to be determined is cached, and reserved for the use of deep learning to determine whether it is a highly overlapping twin mask situation, to avoid the risk of misjudgment of highly overlapping twin masks;

S3. For the cached image of the type of defect to be determined after the detection in the above steps, when detecting the integrity of the earband of the mask and the defect of the pressure earband of the mask, since the position of the earband of the mask relative to the mask body is relatively random, the traditional detection method will In case of misjudgment or undetectable situation, it is considered to use deep learning method to detect the integrity of mask earbands and the defects of mask earbands.

Preferably, the steps of the detection method of the mask membrane joint detection item specifically include:

(a) considering the color characteristics of the mask membrane joint, on the basis of completing the above-mentioned detection items, when carrying out the membrane joint detection, to the target mask region image Image_MaskRegion, use a color space conversion algorithm to convert the collected image from RGB Convert the space into HSV space, use the S channel image for the channel-separated image, perform image binarization processing on the channel image, and filter out the membrane junction area using the area area and geometric shape features, which is recorded as MoJieTouRegion;

(b) if the membrane joint region MoJieTouRegion screened out from the above-mentioned target mask region is not empty, then the detection result is judged as a defective product, if the membrane joint region MoJieTouRegion screened out from the above-mentioned target mask region is empty, then cache this pending decision The image of the defect type is reserved for subsequent inspection steps.

Preferably, the steps of using the deep learning method to detect the integrity of the mask earband and the detection method for the defect detection item of the mask pressure earband specifically include:

(a) Collect a large number of images of masks with qualified and unqualified ear straps under various light intensities and postures, and store them in the OK and NG folders as a sample image set of the deep learning classification model;

(b) Use the pre-trained binary deep learning model, set the main hyperparameters of model training, such as batch_size and learning_rate, use 70% of the total sample image set as the training image set of the training model, and train the pre-trained model ;

(c) Use 20% of the total sample image set as the test images for the trained model to test the model trained with hyperparameters.

(d) Use 10% of the total sample image set as the verification image of the training model to verify the classification effect of the model trained by setting hyperparameters.

Described step (a) also comprises the following processing steps:

Collect the cached images of the types of defects to be determined, and store them in the NG folder as a sample image set for the deep learning classification model; so that the entire detection method can be compatible with twin masks to detect highly overlapping twin masks.

The beneficial effects of the present invention are:

The invention provides a mask defect detection device and method, which realizes automatic feeding through a feeding mechanism and a conveying mechanism, and transports the mask to be detected under the image acquisition mechanism, and the image acquisition mechanism collects the image of the mask to be detected and transmits it to the image processing software. Through the detection algorithm processing, the processing information is fed back to the electrical control system, and then the electrical control system controls the rejection of defective products to realize the automatic detection of the masks to be detected. The innovation of the present invention lies in that the detection device provided by the present invention completes the mask defect detection by means of a line scan camera, and can be quickly and flexibly deployed in a modular manner, and the mask defect detection method provided by the present invention uses the traditional BLOB detection method and depth The combination of learning methods can complete the detection of all items to be detected. The detection method has high stability, high detection efficiency, and high detection accuracy, so that the product quality after the device detection is guaranteed.

Description of drawings

In order to explain the embodiments of the present invention or the technical solutions in the prior art more clearly, the following briefly introduces the accompanying drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained according to these drawings without creative efforts.

Fig. 1 is the structural representation of the mask defect detection device provided in the embodiment of the present invention;

Fig. 2 is the overall flow chart of the mask detection method provided by the embodiment of the present invention.

In the figure: 1-feeding mechanism; 2-conveying mechanism; 21-first conveyor belt; 22-second conveyor belt; 23-second conveyor belt; 3-image acquisition mechanism; 31-camera 1; 32-lens 1 33-light source 1; 34-light source 2; 35-light source 3; 36-light source 4; 37-photoelectric detection sensor; 38-encoder module; 39-camera 2; 310-lens 2; 4-rejection mechanism; Defective product blowing nozzle; 42-defective product collection box; 5-image processing software; 6-electrical control system; 7-mask to be tested.

Detailed ways

In order to make the purposes, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments These are some embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

Example 1:

Please refer to Figures 1-2: a mask defect detection device provided by the present invention includes: a device cabinet (not shown in the figure); a feeding mechanism 1, which provides masks to be tested; a three-section belt line conveying mechanism 2 , convey the masks to be detected; the image acquisition mechanism 3, collect the images of the masks to be detected; the image processing software 5, according to the preset mask standards and detection algorithms, judge the collected mask images and give the results of good or bad products; The mechanism 4 is used to reject the defective products given by the image processing software;

The device cabinet acts as the main body of the device, and the feeding mechanism 1 is connected to the upstream mask production and conveying mechanism to receive the masks to be tested conveyed by the upstream mask production and conveying mechanism. The control of the belt lines 21, 22 and 23 adopts the synchronous belt method, the three-phase terminals of the servo motor controlling the three-section belt lines are controlled by a controller, and the level of the installation of the belt lines is strictly controlled to achieve the best transmission effect. The mask to be inspected can be transmitted stably; the image acquisition mechanism 3 is used to collect the image of the mask to be inspected 7 and transmit the collected image to the image processing software 5, and the image processing software 5 is installed in the computer equipped with the device. The detection algorithm detects the collected images and outputs the result of whether the mask to be inspected is good or bad. The detection result is transmitted to the electrical control mechanism 6 , and the electrical control mechanism 6 controls the blowing nozzle 41 in the rejection mechanism 4 to reject defective products, and stores the masks for rejected defective products in the defective product collection box 42 .

Specifically, in this embodiment, the image acquisition mechanism 3 includes: 4K line scan cameras 31 and 310 mounted on the device cabinet through fixed brackets, and FA lenses 32 and 39 connected to 31 and 310, and FA lenses 32 and 39, respectively The illumination direction of the light source is directly facing the high-brightness line light sources 34 and 35 of the backlight mode of the FA lens and the high-brightness line light sources 33 and 36 of the front-light mode with the central axis of the camera at 45°. In addition, the image acquisition mechanism is equipped with a photoelectric detection switch 37 and an encoder module 38, which are used to control the timing of image acquisition and realize the acquisition of a complete mask image. The detection device provided by the present invention has a total of two sets of image acquisition mechanisms, which are respectively installed in the gaps of the three belt lines, and are used to flexibly complete the detection of multi-faceted defects of various types of masks.

In addition, the present invention also provides a method for detecting defects of masks. The detection defect items of the detection method include but are not limited to the following five categories: empty mask bags/twin masks/half masks, mask folds, mask membrane joints, mask earband integrity Detection and mask pressure earband defect detection, the specific process of the detection algorithm is shown in Figure 2, and the specific steps of the detection algorithm are as follows:

S1, collect the background image of the mask to be detected by the camera that is connected to the image processing software and collect the mask image to be detected, and make a difference between the two images, obtain the difference image, and denote it as Image_Sub;

S2, the difference image Image_Sub is cut out from the original collection image, and the target mask area image Image_MaskRegion is obtained;

The steps of the detection method for the two detection items of mask empty bag/twin mask/half mask and mask fold include:

S3, perform preprocessing operations such as image grayscale, image binarization, etc. on the target mask region image Image_MaskRegion, and use the region area and geometric shape features to screen out the image region of the type of defect to be judged and record it as ToJudgeRegion, and the above to be judged The defect type image area is cropped from the original acquired image;

S4. For the image area ToJudgeRegion of the type of defect to be determined, if the area and geometric features of the area do not meet the preset feature range of the mask, it is directly judged as a defective product, if the area and geometric features of the area conform to the mask. For the preset feature range, the image of the type of defect to be determined is cached and reserved for the use of deep learning to determine whether it is a situation of highly overlapping twin masks, so as to avoid the risk of misjudgment of highly overlapping twin masks;

The steps of the detection method of the mask membrane joint detection item include:

S5, taking into account the color characteristics of the mask membrane joint, on the basis of completing the above detection items, when performing membrane joint detection, use a color space conversion algorithm for the target mask region image Image_MaskRegion to convert the collected image from the RGB space. Convert it into HSV space, use the S channel image for the channel-separated image, perform image binarization processing on the channel image, and filter out the membrane junction area using the area area and geometric shape features, which is recorded as MoJieTouRegion;

S6. If the membrane joint region MoJieTouRegion screened from the above-mentioned target mask region is not empty, the detection result is judged as a defective product, and if the membrane joint region MoJieTouRegion screened out from the above-mentioned target mask region is empty, then cache the to-be-determined defect Type images, reserved for subsequent detection steps;

S7. For the cached image of the type of defect to be determined after the detection in the above steps, when detecting the integrity of the earband of the mask and the defect of the pressure earband of the mask, because the position of the earband of the mask relative to the mask body is relatively random, the traditional detection method will In case of misjudgment or undetectable situation, consider using deep learning method to detect the integrity of mask earbands and the defects of mask earbands;

The main steps of using the deep learning method to detect the integrity of the mask earbands and the defects of the mask earbands include;

S8. Collect a large number of images of qualified and unqualified masks with ear straps under various light intensities and postures, and store them in the OK and NG folders as sample image sets for the deep learning classification model;

S9. Use the pre-trained two-class deep learning model, set the main hyperparameters of model training, such as batch_size and learning_rate, and use 70% of the total sample image set as the training image set of the training model to train the pre-training model;

S10. Use 20% of the total sample image set as the test image of the training model, and test the model trained by setting the hyperparameters.

S11. Use 10% of the total sample image set as a verification image of the training model, and verify the classification effect of the model trained by setting hyperparameters.

The step S8 also includes the following processing steps:

S81. Collect the cached images of the types of defects to be determined, and store them in the NG folder as a sample image set of the deep learning classification model; so that the entire detection method can be compatible with twin masks to detect highly overlapping twin masks.

The above embodiments are only used to illustrate the technical solutions of the present invention, but not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: The recorded technical solutions are modified, or some technical features thereof are equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims (10)

1. a mask defect detection device, comprising a device cabinet, is characterized in that, is installed on the described device cabinet respectively: The feeding mechanism provides masks to be tested; Three-stage belt line conveying mechanism to convey the masks to be tested; an image acquisition mechanism, which collects the image of the mask to be detected; The image processing software judges the collected mask images according to the preset mask standards and detection algorithms and gives the results of good or bad products; Removal mechanism to remove defective products given by image processing software; A control system is used for electrically connecting the feeding mechanism, the conveying mechanism, the image acquisition mechanism, the image processing software and the rejecting mechanism. 2. A mask defect detection device according to claim 1, characterized in that: the image acquisition mechanism comprises: a bracket for fixing a camera; A 4K line scan camera that is electrically connected; a fixed-focus FA lens that matches the interface of the 4K line scan camera and that can fine-tune the aperture and focal length; and a photoelectric detection switch and code that are electrically connected to the 4K line scan camera device module; The two groups of the 4K line scan cameras are respectively located directly below and directly above the gap between two adjacent belts. 3. A mask defect detection device according to claim 2, characterized in that: the image acquisition mechanism further comprises: four groups of high-brightness linear light sources; wherein the two groups of the high-brightness linear light sources are respectively located in every adjacent two The other two sets of high-brightness linear light sources are located at 45° to the side of the gap of each adjacent two-segment belt, namely, the high-brightness linear light sources use the irradiation direction to face the The backlight mode of the FA lens and the front lighting mode at 45° to the central axis of the 4K line scan camera; complete the lighting of the mask to be detected. 4. A mask defect detection device according to claim 1, characterized in that: the feeding mechanism is used for docking with the upstream mask production conveyor line, and receives the mask to be detected conveyed by the upstream mask production conveyor line. 5. A mask defect detection device according to claim 1, characterized in that: the conveying mechanism comprises: a first conveying belt, a second conveying belt and a third conveying belt; The servo motor for the operation of the belt, the second conveyor belt and the third conveyor belt; the mask to be tested is transported between the first conveyor belt, the second conveyor belt and the third conveyor belt . 6. A mask defect detection device according to claim 1, characterized in that: the rejecting mechanism comprises: a blowing cylinder located at the discharge end of the conveying mechanism and a blowing nozzle located directly below the blowing cylinder defective product collection box. 7. a mask defect detection method, adopts a kind of mask defect detection device described in any one of claim 1-6 to carry out, it is characterized in that, the detection defect item of described detection method includes but is not limited to following five categories: Empty bag/twin masks/half masks, mask folds, mask membrane joints, mask earband integrity detection and mask pressure earband defect detection. 8. a kind of mask defect detection method according to claim 7 is characterized in that: The detection method comprises the following steps: S1. Collect the background image of the mask to be detected and the mask image to be detected through a 4K line scan camera, and make a difference between the two images to obtain a difference image, which is recorded as Image_Sub; S2. Crop the difference image Image_Sub from the original collected image to obtain the target mask area image Image_MaskRegion; Among them, the steps of the detection method for the two detection items of mask empty bag/twin mask/half mask and mask fold include: (a) to the target mask region image Image_MaskRegion through image grayscale, image binarization preprocessing operation, and use the area area and geometric shape features to screen out the image region of the type of defect to be judged, denoted as ToJudgeRegion, and the above-mentioned pending The image area for determining the type of defect is cropped from the original captured image; (b) For the image area ToJudgeRegion of the type of defect to be judged, if the area and geometrical features of the area do not meet the preset feature range of the mask, it is directly judged as a defective product, if the area and geometrical characteristics of the area meet The preset feature range of the mask, the image of the type of defect to be determined is cached, and reserved for the use of deep learning to determine whether it is a highly overlapping twin mask situation, to avoid the risk of misjudgment of highly overlapping twin masks; S3. For the cached image of the type of defect to be determined after the detection in the above steps, when detecting the integrity of the earband of the mask and the defect of the earband of the mask, a deep learning method is used to detect the integrity of the earband of the mask and the defect of the earband of the mask. 9. a kind of mask defect detection method according to claim 8, is characterized in that: The steps of the detection method of the mask membrane joint detection item specifically include: (a) considering the color characteristics of the mask membrane joint, on the basis of completing the above-mentioned detection items, when carrying out the membrane joint detection, to the target mask region image Image_MaskRegion, use a color space conversion algorithm to convert the collected image from RGB Convert the space into HSV space, use the S channel image for the channel-separated image, perform image binarization processing on the channel image, and filter out the membrane junction area using the area area and geometric shape features, which is recorded as MoJieTouRegion; (b) if the membrane joint region MoJieTouRegion screened out from the above-mentioned target mask region is not empty, then the detection result is judged to be a defective product, if the membrane joint region MoJieTouRegion screened out from the above-mentioned target mask region is empty, then cache the to-be-determined The image of the defect type is reserved for subsequent inspection steps. 10. a kind of mask defect detection method according to claim 9, is characterized in that: The steps of using the deep learning method to detect the integrity of the mask earbands and the defect detection items of the mask pressure earbands include: (a) Collect a large number of images of masks with qualified and unqualified ear straps under various light intensities and postures, and store them in two folders, OK and NG, as a sample image set of the deep learning classification model; (b) Use the pre-trained binary deep learning model, set the main hyperparameters of model training, such as batch_size and learning_rate, use 70% of the total sample image set as the training image set of the training model, and train the pre-trained model ; (c) Use 20% of the total sample image set as the test images for the trained model to test the model trained with hyperparameters. (d) Use 10% of the total sample image set as the verification image of the training model to verify the classification effect of the model trained by setting hyperparameters. Described step (a) also comprises the following processing steps: Collect the cached images of the types of defects to be determined, and store them in the NG folder as a sample image set of the deep learning classification model; so that the entire detection method can be compatible with twin masks to detect highly overlapping twin masks.

Images