Facial mask flaw automatic detection mechanism

Abstract

The invention discloses an automatic mask flaw detection mechanism which comprises a conveying belt device, an electronic induction device, a visual detection device, an automatic removal device, a control device and an upper computer. The conveyor belt device is used for conveying the mask to pass through the visual detection device; the electronic induction device is used for detecting whether the facial mask enters the visual detection device or not; the control device is used for controlling the visual detection device to collect image information after receiving the induction signal; the visual detection device is used for collecting a mask image; the upper computer is used for detecting flaws of the facial mask based on the facial mask image; the control device is further used for sending a rejecting signal to the automatic rejecting device after receiving the rejecting instruction of the upper computer; and the automatic removing device is used for removing the facial masks conveyed out from the visual detection device after receiving the removing signal. According to the invention, the accuracy and efficiency of mask flaw detection are improved.

Description

Technical Field

[0001] The invention belongs to the technical field of detection equipment, and in particular relates to an automatic detection mechanism for facial mask defects. Background Art

[0002] Traditional mask defect detection relies on manual visual judgment, which is not only time-consuming and labor-intensive, but also lacks unified standards. This manual detection method has multiple problems:

[0003] 1. Low production efficiency: Since manual picking is required, this directly affects the efficiency of the assembly line. After the mask is cut, it needs to be manually sorted before entering the automatic folding and bagging assembly line. This sorting process not only increases labor costs, but also reduces overall production efficiency.

[0004] 2. Difficulty in quality control: Managers cannot understand the quality of mask fabrics in real time. This information lag may lead to delayed handling of quality issues, which in turn affects the overall quality control of the product.

[0005] 3. Inconsistent test results: Manual testing is affected by human factors such as subjective will, emotions, and visual fatigue, resulting in large errors and slow speed in test results.

[0006] 4. Lack of standardization: Due to the lack of unified testing standards, the results of manual testing are difficult to quantify and standardize, which makes it difficult to ensure product quality. Summary of the invention

[0007] The technical problem to be solved by the present invention is to provide an automatic detection mechanism for facial mask defects in view of the deficiencies in the above-mentioned prior art, which can replace manual inspection with machine vision inspection equipment, avoid the shortcomings of manual inspection standards and subjective interference, standardize and quantify the inspection results, further improve the automation level of the entire production system, save labor costs, and achieve cost reduction and efficiency improvement for enterprises. This inspection mechanism not only improves the accuracy and efficiency of inspection, but also enables managers to monitor product quality in real time.

[0008] In order to solve the above technical problems, the technical solution adopted by the present invention is: an automatic detection mechanism for facial mask defects, including a conveyor belt device, an electronic sensing device, a visual detection device, an automatic rejection device, a control device and a host computer;

[0009] The conveyor belt device is used to transport the facial mask through the visual inspection device;

[0010] The electronic sensing device is used to detect whether the mask enters the visual detection device, and when it is detected that the mask enters the visual detection device, it sends a sensing signal to the control device;

[0011] The control device is used to control the visual detection device to collect image information after receiving the sensing signal;

[0012] The visual detection device is used to collect facial mask images;

[0013] The host computer is used to receive the mask image collected by the visual inspection device, detect the defects of the mask based on the mask image, and send a rejection instruction to the control device;

[0014] The control device is also used to send a rejection signal to the automatic rejection device after receiving a rejection instruction from the host computer;

[0015] The automatic rejecting device is used to reject the mask transported from the visual inspection device after receiving the rejecting signal.

[0016] The above-mentioned automatic mask defect detection mechanism, the conveying device is preferably a two-section conveyor belt, one of which is located inside the visual detection device, the visual detection device includes an area array camera and a line scan camera, the area array camera is used to capture the front image of the mask from top to bottom, and the line scan camera is used to capture the back image of the mask from bottom to top.

[0017] The conveyor belt located inside the visual inspection device can preferably be a transparent conveyor belt.

[0018] The above-mentioned automatic mask defect detection mechanism, the automatic rejection device is installed at the exit of the visual detection device, and a waste hopper is arranged directly below the visual detection device.

[0019] The above-mentioned automatic mask defect detection mechanism, when the host computer detects the defects of the mask based on the mask image, includes the following steps:

[0020] Step 1: Preprocess the mask image to reduce noise and highlight the features of the mask itself;

[0021] Step 2: Perform edge extraction on the mask image to remove the conveyor belt image outside the mask;

[0022] Step 3: Input the mask image processed in step 2 into a pre-trained deep learning model for feature recognition to identify the defect features of the mask.

[0023] In the above-mentioned automatic facial mask defect detection mechanism, the image preprocessing in step 1 includes reducing image noise by Gaussian blurring, and then converting the image into a multi-level grayscale image by multi-level threshold segmentation.

[0024] In the above-mentioned automatic mask defect detection mechanism, in step 2, an edge extraction operation is performed on the mask image, and the edge extraction is realized by using the Sobel operator.

[0025] In the above-mentioned automatic mask defect detection mechanism, the deep learning model in step 3 is a convolutional neural network (CNN). During pre-training, the deep learning model uses data enhancement technology to increase the diversity of training data. The data enhancement technology includes mirroring, rotating and / or translating the training data to increase the diversity of the training data.

[0026] In the above-mentioned automatic mask defect detection mechanism, in step 3, when the mask image processed in step 2 is input into a pre-trained deep learning model for feature recognition, a sliding window method is adopted, and a sliding window is redundantly slid on the mask image, and the image of each window is input into the deep learning model for feature recognition.

[0027] The above-mentioned automatic mask defect detection mechanism also includes a workbench, a conveyor belt device above the workbench, and a shooting chamber above and below the conveyor belt device. The area array camera is installed in the upper shooting chamber, and the line scan camera is installed in the lower shooting chamber.

[0028] The above-mentioned automatic mask defect detection mechanism, the electronic sensing device includes two photoelectric sensors, and the two photoelectric sensors are symmetrically installed on both sides of the starting end of the conveyor belt device. There is a gap between the bottom surface of the photoelectric sensor and the belt surface of the conveyor belt device for the mask to pass through.

[0029] The two photoelectric sensors may adopt a plate-shaped structure, which facilitates better sensing of the facial mask.

[0030] Compared with the prior art, the present invention has the following advantages: the innovation of the automatic defect detection mechanism of the facial mask lies in its integrated design and automated process. By combining a conveyor belt device, an electronic sensing device, a visual detection device, an automatic rejection device, a control device and a host computer, a fully automated operation from facial mask transportation, sensing, image acquisition, defect detection to automatic rejection is realized. This integrated system improves the detection efficiency and accuracy, reduces manual intervention, and improves production efficiency and product quality. The system improves the response speed and accuracy of detection through the linkage of electronic sensing and automatic control. The combination of visual detection and intelligent analysis of the host computer can identify minor defects and improve the accuracy of detection. The automatic rejection device reduces manual operation, improves rejection efficiency and accuracy, and reduces labor costs. In addition, the flexibility and adaptability of the system are enhanced, and it can adapt to the detection needs of facial masks of different specifications, reduce waste in production, and improve material utilization. At the same time, automated operation reduces safety risks in the production process and improves production safety. In general, the detection mechanism achieves high efficiency, high quality and low cost in facial mask production through technological innovation, and enhances the market competitiveness of the product.

[0031] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a circuit schematic diagram of the present invention.

[0033] Figure 2 The structure of the present invention is schematically shown Figure 1 .

[0034] Figure 3 The structure of the present invention is schematically shown Figure 2 .

[0035] Figure 4 It is a schematic diagram of the internal structure of the present invention.

[0036] Description of reference numerals:

[0037] 1—Conveyor belt device; 2—Electronic sensing device; 3—Visual detection device;

[0038] 4—Automatic rejection device; 5—Control device; 6—Upper computer;

[0039] 7—working table; 8—shooting chamber; 9—sound and light alarm;

[0040] 10—Waste hopper. DETAILED DESCRIPTION

[0041] like Figure 1 — Figure 4 As shown, an automatic detection mechanism for facial mask defects includes a conveyor belt device 1, an electronic sensing device 2, a visual detection device 3, an automatic rejection device 4, a control device 5 and a host computer 6;

[0042] The conveyor belt device 1 is used to transport the facial mask through the visual inspection device 3;

[0043] The electronic sensing device 2 is used to detect whether the mask enters the visual detection device 3, and when it is detected that the mask enters the visual detection device 3, a sensing signal is sent to the control device 5;

[0044] The control device 5 is used to control the visual detection device 3 to collect image information after receiving the sensing signal;

[0045] The visual detection device 3 is used to collect the mask image;

[0046] The host computer 6 is used to receive the mask image collected by the visual inspection device 3, detect the defects of the mask based on the mask image, and send a rejection instruction to the control device 5;

[0047] The control device 5 is also used to send a rejection signal to the automatic rejection device 4 after receiving the rejection instruction from the host computer 6;

[0048] The automatic rejecting device 4 is used to reject the mask transported from the visual inspection device 3 after receiving the rejecting signal.

[0049] In this embodiment, the conveying device is a two-section conveyor belt, one of which is located inside the visual inspection device 3. The conveyor belt located inside the visual inspection device 3 is preferably a transparent conveyor belt. The visual inspection device 3 includes an area array camera and a line scan camera. The area array camera is used to capture the front image of the mask from top to bottom, and the line scan camera is used to capture the back image of the mask from bottom to top.

[0050] In this embodiment, the automatic mask defect detection mechanism also includes a workbench 7, and the conveyor belt device 1 is above the workbench 7. There is a shooting chamber 8 above and below the conveyor belt device 1. The area array camera is installed in the upper shooting chamber 8, and the line scan camera is installed in the lower shooting chamber 8.

[0051] In this embodiment, the electronic sensing device 2 includes two photoelectric sensors, which are symmetrically installed on both sides of the starting end of the conveyor belt device 1. There is a gap between the bottom surface of the photoelectric sensor and the belt surface of the conveyor belt device 1 for the mask to pass through.

[0052] The photoelectric sensor is preferably a plate-shaped structure.

[0053] In this embodiment, a sound and light alarm 9 is installed near the photoelectric sensor. When the automatic mask defect detection mechanism finds that there are defects in consecutive masks, the control device 5 controls the sound and light alarm 9 to sound an alarm.

[0054] The conveyor belt device 1 is equipped with a code meter for monitoring the transmission speed of the conveyor belt and sending the collected conveyor belt speed information to the control device 5. The control device 5 controls the visual detection device 3 to collect the image of the mask at an appropriate time based on the conveyor belt speed information and the time when the mask is placed on the conveyor belt sensed by the electronic sensing device 2.

[0055] In this embodiment, the automatic rejection device 4 is installed at the exit of the visual inspection device 3 , and a waste hopper 10 is provided directly below the visual inspection device 3 .

[0056] It should be noted that the automatic rejection device 4 is a retractable card knife. When a defective mask that needs to be rejected is found, the control device 5 controls the automatic rejection device 4 to extend, and the unqualified defective mask is rejected and rejected into the waste hopper 10 directly below.

[0057] In this embodiment, when the host computer 6 detects defects of the mask based on the mask image, the following steps are included:

[0058] Step 1: Preprocess the mask image to reduce noise and highlight the features of the mask image;

[0059] Step 2: Perform edge extraction on the mask image to remove the conveyor belt image outside the mask;

[0060] Step 3: Input the mask image processed in step 2 into a pre-trained deep learning model for feature recognition to identify the defect features of the mask.

[0061] It should be noted that the image preprocessing in step 1 includes reducing image noise through Gaussian blur, and then converting the image into a multi-level grayscale image through multi-level threshold segmentation. Image preprocessing includes steps such as grayscale conversion, Gaussian blur, and threshold segmentation to reduce noise and highlight the characteristics of the mask itself. Gaussian blur is used to reduce image noise and make the edges of the defective area smoother, and then the image is converted into a multi-level grayscale image through multi-level threshold segmentation to more clearly identify the defective area.

[0062] It should be noted that in step 2, the edge extraction operation is performed on the mask image, and the sobel operator is used to realize the edge extraction, so as to remove the image of the conveyor belt itself outside the mask cloth (to avoid misidentification of conveyor belt stains).

[0063] It should be noted that the deep learning model in step 3 is a convolutional neural network (CNN). During pre-training, the deep learning model uses data enhancement technology to increase the diversity of training data. The data enhancement technology includes mirroring, rotating and / or translating the training data to increase the diversity of the training data.

[0064] It should be noted that the convolutional neural network (CNN) is used to automatically learn features in images and perform classification and recognition. Through the combination of convolutional layers, pooling layers, and fully connected layers, CNN can effectively capture local features in images, reduce dimensions, and ultimately achieve accurate judgment of image content. When training and optimizing the model: Use labeled data sets to train the model. During the training process, L2 regularization and Dropout are used to prevent overfitting. A specific loss function, Focal Loss, is designed to deal with the problem of class imbalance. After the model training is completed, the model needs to be evaluated to ensure its accuracy and robustness. The performance of the model can be evaluated by calculating the performance indicators of the model on the validation set, such as accuracy, recall, and F1 score. Based on the evaluation results, the model can be iteratively optimized.

[0065] It should be noted that in step 3, when the mask image processed in step 2 is input into a pre-trained deep learning model for feature recognition, a sliding window method is adopted, and the sliding window is redundantly slid on the mask image, and the image of each window is input into the deep learning model for feature recognition.

[0066] In actual implementation, the size and confidence of the defects detected can be set for different types of masks, thereby achieving different quality inspection standards. Various defects include creases, holes, color stains, black spots, hair, flying insects, poor cutting, etc.

[0067] When the present invention is used:

[0068] Step 1. Use electronic sensing device to detect the mask entering the testing equipment from the assembly line;

[0069] Step 2, based on the trigger signal, trigger the camera shooting device to shoot the front and back sides of the mask cloth;

[0070] Step 3: Perform image preprocessing to filter out noise through grayscale conversion, Gaussian blur and threshold segmentation;

[0071] Step 4: Identify the contour of the mask through edge extraction to avoid false detection of dirt on the conveyor belt;

[0072] Step 5. Submit the processed images to algorithm recognition (based on deep learning and trained algorithm model) to identify various defects in the images (creases, holes, color stains, black spots, hair, flying insects, poor cutting, etc.);

[0073] Step 6: Algorithm service combines the front and back test results and sends the final pass / fail result to PLC;

[0074] Step 7. Based on the accepted result, PLC triggers the automatic rejection device 4 to remove the unqualified mask cloth from the assembly line.

[0075] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any way. Any simple modification, change and equivalent structural change made to the above embodiment based on the technical essence of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A facial mask defect automatic detection mechanism, characterized in that: It includes a conveyor belt device, an electronic sensing device, a visual detection device, an automatic rejection device, a control device and a host computer; The conveyor belt device is used to transport the facial mask through the visual inspection device; The electronic sensing device is used to detect whether the mask enters the visual detection device, and when it is detected that the mask enters the visual detection device, a sensing signal is sent to the control device; The control device is used to control the visual detection device to collect image information after receiving the sensing signal; The visual detection device is used to collect facial mask images; The host computer is used to receive the mask image collected by the visual inspection device, detect the defects of the mask based on the mask image, and send a rejection instruction to the control device; The control device is also used to send a rejection signal to the automatic rejection device after receiving a rejection instruction from the host computer; The automatic rejecting device is used to reject the mask transported from the visual inspection device after receiving the rejecting signal.

2. The automatic facial mask defect detection mechanism according to claim 1, characterized in that: The conveyor belt of the conveying device located inside the visual inspection device is a two-section conveyor belt. The visual inspection device includes an area array camera and a line scan camera. The area array camera is used to capture the front image of the mask from top to bottom, and the line scan camera is used to capture the back image of the mask from bottom to top.

3. The automatic facial mask defect detection mechanism according to claim 1, characterized in that: The automatic rejection device is installed at the exit of the visual inspection device, and a waste hopper is arranged directly below the visual inspection device.

4. The automatic facial mask defect detection mechanism according to claim 1, characterized in that: When the host computer detects defects of the mask based on the mask image, the following steps are included: Step 1: Preprocess the mask image to reduce noise and highlight the features of the mask image; Step 2: Perform edge extraction on the mask image to remove the conveyor belt image outside the mask; Step 3: Input the mask image processed in step 2 into a pre-trained deep learning model for feature recognition to identify the defect features of the mask.

5. The automatic facial mask defect detection mechanism according to claim 4, characterized in that: The image preprocessing in step 1 includes reducing image noise by Gaussian blurring, and then converting the image into a multi-level grayscale image by multi-level threshold segmentation.

6. The automatic facial mask defect detection mechanism according to claim 4, characterized in that: In step 2, edge extraction is performed on the mask image using the Sobel operator.

7. The automatic facial mask defect detection mechanism according to claim 4, characterized in that: The deep learning model in step 3 is a convolutional neural network (CNN). During pre-training, the deep learning model uses data enhancement technology to increase the diversity of training data. The data enhancement technology includes mirroring, rotating and / or translating the training data to increase the diversity of the training data.

8. The automatic facial mask defect detection mechanism according to claim 4, characterized in that: In step 3, when the mask image processed in step 2 is input into a pre-trained deep learning model for feature recognition, a sliding window method is adopted, and a sliding window is redundantly slid on the mask image, and the image of each window is input into the deep learning model for feature recognition.

9. The automatic facial mask defect detection mechanism according to claim 2, characterized in that: It also includes a workbench, the conveyor belt device is above the workbench, and there is a shooting chamber above and below the conveyor belt device, the area array camera is installed in the upper shooting chamber, and the line scan camera is installed in the lower shooting chamber.

10. The automatic facial mask defect detection mechanism according to claim 2, characterized in that: The electronic sensing device comprises two photoelectric sensors, which are symmetrically mounted on both sides of the starting end of the conveyor belt device. There is a gap between the bottom surface of the photoelectric sensor and the belt surface of the conveyor belt device for the mask to pass through.

Images