Leak detection methods, production line operating methods, devices, systems and storage media

Abstract

This invention relates to a leak detection method, comprising: acquiring data of a test chamber; receiving specified parameters; forming a detection model based on the specified parameters; inputting the test chamber data into the detection model for identification, and obtaining a leak detection result; wherein, receiving the specified parameters includes: reading sample RGB image data; converting the sample RGB image data into sample HSV color space image data; and receiving parameter calibration of the sample HSV color space image data. This invention can realize the function of detecting missing parts in product components on the production line. It can be quickly deployed and, even without professional personnel, can be adjusted and deployed based on a very small number of samples, greatly saving R&D resources and deployment time, and improving efficiency. It solves the problem in existing leak detection algorithms that use AI methods for image recognition, which requires professional technicians, GPU computing resources, and a large amount of development and debugging time.

Description

Technical Field

[0001] This invention relates to the field of production line leak detection, specifically to a leak detection method, production line operating method, apparatus, system, and storage medium. Background Technology

[0002] In existing production lines, the missed detection algorithms used employ AI methods for image recognition, requiring hundreds or even thousands of training images. This necessitates specialized technicians and large GPU computing clusters for algorithm training and recognition. This significantly impacts production line operations, as the equipment cannot be used during image acquisition and algorithm training. Furthermore, when internal packaging materials are replaced, the AI ​​program needs to be readjusted, still requiring specialized technicians, GPU computing resources, and substantial development and debugging time. Summary of the Invention

[0003] In view of this, the present invention provides a leak detection method, production line operation method, apparatus, system and storage medium to solve the problem that the leak detection algorithm in the prior art uses AI methods for image recognition, which requires professional technicians, GPU computing resources and a lot of development and debugging time.

[0004] To achieve one, some, or all of the above objectives, or other objectives, the present invention provides a leak detection method, comprising:

[0005] Collect data from the enclosure under test;

[0006] Receive specified parameters;

[0007] A detection model is formed based on the specified parameters;

[0008] The data of the box to be tested is input into the detection model for identification to obtain the leak detection result;

[0009] The receiving of specified parameters includes:

[0010] Read sample RGB image data;

[0011] Convert the sample RGB image data into sample HSV color space image data;

[0012] Receive parameter calibration for the sample HSV color space image data.

[0013] Furthermore, the parameter calibration includes the following three items: specifying the image analysis area, extracting the data range of the V color space, and reading the area data.

[0014] Furthermore, the detection model includes the following two judgment function parts:

[0015] 1. Based on the specified parameters, perform image recognition of the enclosure to determine whether it is qualified;

[0016] Second: Conduct a missing parts inspection to determine whether it is qualified.

[0017] Furthermore, the collection of data from the test chamber includes:

[0018] The markings on the box that have entered the area to be detected have been identified.

[0019] The depth data of the area to be detected is acquired by capturing images;

[0020] Extract data within a specified interval, where the specified interval is defined within the depth data;

[0021] The data within the specified interval is normalized to form a grayscale image;

[0022] Convert the grayscale image data into RGB image data;

[0023] Store the RGB image data.

[0024] Further, the step of performing box image recognition based on the specified parameters to determine whether it is qualified includes:

[0025] Read the RGB image data;

[0026] Convert the RGB image data into HSV color space image data;

[0027] The HSV color space image data is binarized.

[0028] Image filtering is performed on the binarized HSV color space image data to retain pixels that meet the calibration parameters, thereby obtaining the box outline, the first image containing the box outline, and the filtered data area.

[0029] Determine whether the area of ​​the filtered data is less than a first preset value. If it is, the data is qualified; otherwise, it is unqualified.

[0030] Furthermore, the first preset value is 10% of the preset specified area data.

[0031] Furthermore, the process of detecting missing parts and determining whether the parts are qualified includes:

[0032] The first image is cropped according to the box outline, while retaining the image data within the outer outline of the box;

[0033] Adjust the first image to make it upright and horizontal;

[0034] The first image is cropped, retaining the image within a preset specified area data to form the second image;

[0035] The shape and outline of the accessory placed inside the box are obtained by identifying the second image through area filtering;

[0036] A minimum outer rectangle is used, which encompasses the outline of the accessory;

[0037] Calculate the center point of the component, and calculate the distance between the center point of the component and a preset designated center point;

[0038] Determine whether the distance between the center point of the accessory and a preset designated center point is less than a second preset value. If yes, it is qualified; otherwise, it is unqualified.

[0039] Furthermore, the second preset value is 10% of the diagonal distance of the housing.

[0040] Further, after determining whether the distance between the center point of the component and a preset designated center point is less than a second preset value, if yes, it is qualified; if no, it is unqualified, the process also includes:

[0041] Determine if there are multiple components that need to be inspected. If so, perform the step of cropping the first image according to the box outline for each component, retaining the image data within the outer outline of the box.

[0042] Furthermore, obtaining the leak detection result specifically includes: when the enclosure image recognition based on the specified parameters is qualified, and when the leak detection is qualified, the leak detection result is qualified; otherwise, it is unqualified.

[0043] A production line operating method, based on any of the above-described methods for detecting missed defects, includes the following steps:

[0044] T1: The chamber enters the testing area;

[0045] T2: Identify the markings on the box when entering the detection area;

[0046] T4: Capture the area to be detected;

[0047] T7: Detect whether there is a leak according to the leak detection method described above.

[0048] If not, issue a second display signal and enter T8.

[0049] If so, issue a third display signal and enter T9;

[0050] T8: Seal the box;

[0051] T9: Manually verify if any items are missing.

[0052] If so, manually replace the missing parts and proceed to T2.

[0053] If not, proceed to T8.

[0054] Furthermore, the interval between step T2 and step T4 includes:

[0055] T3: Determine whether this enclosure has been inspected.

[0056] If so, delete the identification data corresponding to this identifier from the previous step, and then proceed to T4.

[0057] If not, proceed directly to T4.

[0058] Furthermore, the interval between step T4 and step T7 includes:

[0059] T5: Determine if the box exists.

[0060] If so, issue the first display signal and enter T6.

[0061] If not, proceed to T7;

[0062] T6: After manually moving the box to the designated shooting area, repeat T2.

[0063] An apparatus for performing any of the above-described leak detection methods and / or any of the above-described production line operating methods, the apparatus comprising an imaging module, an image processing module, a manual specification module, a detection and judgment module, and a storage module;

[0064] The camera module is used to collect data from the box under test;

[0065] The image processing module is used to process the data acquired by the shooting module;

[0066] The manual specification module is used to receive manually calibrated or specified parameters, which include one or more of the following: specifying the image analysis area, extracting the data range of the V color space, reading area data, specifying area data, and specifying the center point;

[0067] The detection and judgment module identifies and judges the data processed by the image processing module and / or the data collected by the shooting module based on the manually calibrated or specified data.

[0068] The storage module is used to store one or more of the following: data collected by the shooting module, intermediate or result data from the image processing module, and manually calibrated or specified data.

[0069] Furthermore, the imaging module includes a depth camera and / or a lidar.

[0070] A system comprising the above-described apparatus, further comprising: an identification module and a display module;

[0071] The identification module is used to identify the markings on the box and trigger the shooting module to collect data of the box under test;

[0072] The display module is used to visualize the detection results given by the detection and judgment module, and the display module emits different display signals for different detection results.

[0073] Furthermore, the identification module is a barcode scanner and / or an RFID reader, and the markings on the box are graphic codes and / or RFID chips.

[0074] A readable storage medium containing a computer program, which, when executed by a processor, causes the processor to perform any of the leak detection methods described above.

[0075] This invention enables the detection of missing parts in production line components. It allows for rapid deployment and, even without specialized personnel, can adjust and deploy the leak detection method based on a very small sample size, significantly saving R&D resources and deployment time, and improving efficiency. The algorithm parameters can be adjusted to adapt to new production lines or packaging materials by simply collecting a few photos and following a sequential operation process. It requires neither specialized technical personnel nor GPU computing resources, and allows for rapid adjustment and deployment of the leak detection method.

[0076] This addresses the problem that existing image recognition algorithms using AI methods for missed detection require specialized technical personnel, GPU computing resources, and significant development and debugging time. Attached Figure Description

[0077] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0078] in:

[0079] Figure 1 This is a flowchart illustrating the leak detection method in some embodiments of this application;

[0080] Figure 2 This is a flowchart illustrating the leak detection method in some embodiments of this application;

[0081] Figure 3 This is a flowchart illustrating the production line operation method in some embodiments of this application;

[0082] Figure 4 This is a schematic diagram of the apparatus and system in some embodiments of this application;

[0083] Figure 5 This is a schematic diagram illustrating the interaction between a readable storage medium containing a computer program and a processor in certain embodiments of this application. Detailed Implementation

[0084] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0085] like Figure 1 As shown, this is a preferred embodiment of a leak detection method. This method can be applied to scenarios such as production lines. Products produced on production lines often have many small parts. These parts need to be placed in the grooves of the foam box before sealing. This method can detect leaks in the aforementioned parts. It should be noted that this embodiment is applied to the case where several parts are placed in a container, and is not limited to the box. For example, a tray or the like can also be used.

[0086] In existing production lines, the leak detection algorithms use AI-based image recognition, requiring hundreds or even thousands of training images. This necessitates specialized technicians and large GPU computing clusters for algorithm training and recognition. This significantly impacts production, preventing equipment use during image acquisition and algorithm training. Furthermore, when internal packaging materials are replaced, the AI ​​program needs readjustment, still requiring specialized technicians, GPU computing resources, and substantial development and debugging time. This leak detection method, however, uses specified parameters to form a detection model. Through efficient judgment steps involving box image recognition and leak detection, it allows for adjustment and deployment based on minimal samples. It requires neither specialized technicians nor GPU computing resources and allows for rapid adjustment.

[0087] One leak detection method in this embodiment includes:

[0088] Collect data from the enclosure under test;

[0089] Receive specified parameters;

[0090] A detection model is formed by manually collecting data based on the specified parameters; the data of the box to be tested is input into the detection model for identification, and the leak detection result is obtained.

[0091] like Figure 2 As shown, receiving specified parameters includes:

[0092] Read sample RGB image data. This sample RGB image data can be pre-selected sample image data or data collected from the above-mentioned data collection of the test box. For example, it is only necessary to extract data once from the first qualified image after the leak detection method is deployed.

[0093] The sample RGB (Red, Green, Blue) image data is converted into sample HSV (Hue, Saturation, Value) color space image data;

[0094] The parameter calibration of the sample HSV color space image data is received, and the parameter calibration can be performed manually.

[0095] like Figure 2 As shown, in another embodiment, the manual calibration parameters include the following three items: specifying the image analysis area, extracting the data range of the V (Value, Brightness) color space, and reading the area data.

[0096] The three manually specified calibration parameters mentioned above can be saved to a configuration file.

[0097] In another embodiment, the detection model includes the following two judgment function parts:

[0098] Based on the received specified parameters, the box image is recognized to determine whether it is qualified;

[0099] Perform a missing parts inspection to determine if the product is qualified.

[0100] In another embodiment, the acquisition of data from the test chamber includes:

[0101] The system identifies the markings on the cabinet that have entered the detection area. It should be noted that these markings can be graphic codes, such as barcodes, QR codes, or RFID (Radio Frequency Identification) chips, etc., which can be identified as electronic tags. As a preferred solution, this embodiment uses SN codes (Serial Number or Serial No., product serial number) and uses a barcode scanner to identify the SN codes. The SN codes serve as a unique number for the cabinet recorded by the system. At the same time, scanning the SN codes can trigger the camera to collect data from the cabinet.

[0102] The depth data of the area to be inspected is acquired by capturing images. This area may be part of a production line, such as a section on a conveyor belt. In this embodiment, a depth camera is preferably used to capture images of the area to be inspected. The data collected by the depth camera is the depth data obtained by infrared sensors from various points on the housing from the camera. Each pixel of the image will collect one depth data point. Since the camera's acquisition range needs to be greater than 50cm, the range of the acquired depth data is from 10000 to positive infinity.

[0103] Data within a specified interval is extracted, where the specified interval is defined within the depth data. It should be noted that, in this embodiment, a possible application scenario includes placing a box to be detected on a conveyor belt. For this scenario or similar scenarios, since the information to be detected in the detection model is a very narrow depth interval from the conveyor belt plane under the camera to the top of the product box, the specified interval is this interval. Therefore, this step filters the data within the depth interval, setting all other data to 0.

[0104] The data within the specified interval is normalized to form a grayscale image. When the application scenario is that the box to be detected is placed on the conveyor belt, the data range with a value greater than 0 in the previous step can be normalized to map the data between 1 and 255 to form grayscale image data.

[0105] Convert the grayscale image data into RGB image data;

[0106] The RGB image data can be stored, for example, in the system hard drive.

[0107] In another embodiment, the step of performing box image recognition based on the received specified parameters to determine whether it is qualified includes:

[0108] Read the RGB image data;

[0109] Convert the RGB image data into HSV color space image data;

[0110] The HSV color space image data is binarized; common binarization threshold selection methods include bimodal method, p-parameter method, large law method, maximum entropy threshold method, and iterative method.

[0111] Perform image filtering on the HSV color space image data after the binarization process, only retain the pixel points that meet the manually calibrated parameters, and obtain the box contour, the first image where the box contour is located, and the filtered data area; determine whether the filtered data area is less than a first preset value, if so, it is qualified, if not, it is unqualified. As a preferred solution, the first preset value is 10% of the preset manually specified area data.

[0112] In another embodiment, the performing missing part detection and determining whether it is qualified includes:

[0113] Crop the first image according to the outer contour of the box, and retain the image data within the outer contour of the box;

[0114] Adjust the first image to make it upright and horizontal, and the adjustment includes one or more of, but is not limited to, mirroring, translation, rotation, shearing, and deformation;

[0115] Crop the first image, and only retain the image within the preset manually specified area data;

[0116] Identify the shape and contour of the accessory through area filtering;

[0117] Adopt the minimum bounding rectangle, and the minimum bounding rectangle contains the contour of the accessory therein, and the minimum bounding rectangle is as close as possible to the contour of the accessory;

[0118] Calculate the center point of the accessory, and calculate the distance between the center point of the accessory and the preset manually specified center point;

[0119] Determine whether the distance between the center point of the accessory and the preset manually specified center point is less than a second preset value, if so, it is qualified, if not, it is unqualified.

[0120] As a preferred solution, the second preset value is 10% of the diagonal distance of the box.

[0121] If there are multiple accessories to be detected in one box, repeat the above steps for each accessory to perform missing part detection and determine whether it is qualified.

[0122] Preferably, when the box image recognition based on the received specified parameters is determined to be qualified and when the missing part detection is determined to be qualified, the leak detection result is qualified, otherwise it is unqualified.

[0123] As Figure 2 shown in, a leak detection method in a preferred embodiment of the present application. For the convenience of understanding, the leak detection method can be understood in four parts, namely data acquisition, box image recognition, missing part detection, and result feedback. Among them, the box judgment includes two parts: receiving specified parameters and box image recognition.

[0124] like Figure 2 As shown in the figure, the specific steps of this preferred embodiment are as follows:

[0125] I. Data Collection:

[0126] S11: Identify the markings on the box that have entered the area to be detected;

[0127] S12: Capture depth data of the area to be detected;

[0128] S13: Extract data within a specified interval, wherein the specified interval is specified within the depth data;

[0129] S14: Normalize the data within the specified interval to form a grayscale image;

[0130] S15: Convert the grayscale image data into RGB image data;

[0131] S16: Store the RGB image data.

[0132] II. Box-shaped structure assessment:

[0133] 1. Receive specified parameters:

[0134] S211: Read sample RGB image data;

[0135] S212: Convert the sample RGB image data into sample HSV color space image data;

[0136] S213: Receive manually calibrated parameters for the sample HSV color space image data (manually specify the image analysis area, manually specify the data range for extracting the V color space, and manually specify the area data to be read);

[0137] S214: Save the manually calibrated parameters to a configuration file.

[0138] This process only requires extracting data once from the first qualified image after the leak detection method is deployed.

[0139] 2. Box image recognition:

[0140] S221: Read the RGB image data;

[0141] S222: Convert the RGB image data into HSV color space image data;

[0142] S223: Binarize the HSV color space image data;

[0143] S224: Perform image filtering on the HSV color space image data, retaining only pixels that meet the manually calibrated parameters, to obtain the box outline and the first image containing the box outline;

[0144] S225: Determine whether the area of ​​the filtered data is less than the first preset value (10% of the manually specified area data). If it is, it is qualified; otherwise, it is unqualified.

[0145] III. Leaking Components Detection

[0146] S31: Crops the first image according to the outer contour of the box, while retaining the image data within the outer contour of the box;

[0147] S32: Adjust the first image to make it upright and horizontal;

[0148] S33: Crop the first image, retaining only the image within the preset manually specified area data;

[0149] S34: Obtain the shape and outline of the accessory through area filtering;

[0150] S35: A minimum outer rectangle is adopted, wherein the minimum outer rectangle encompasses the outline of the accessory;

[0151] S36: Calculate the center point of the part and the distance between the center point of the part and the manually specified center point;

[0152] S37: Determine whether the distance between the center point of the accessory and the preset artificially designated center point is less than the second preset value (10% of the diagonal distance of the box). If yes, it is qualified; otherwise, it is unqualified.

[0153] If multiple parts need to be inspected in the process, the missing part inspection can be repeated.

[0154] IV. Results Feedback

[0155] S4: When the box image recognition based on the manual data collection is qualified (i.e., qualified in step S225) and when the leak detection is qualified (i.e., qualified in step S37), the leak detection result is qualified; otherwise, it is unqualified.

[0156] Another embodiment of the present invention provides a production line operation method, based on the missed detection method described in any of the above embodiments, the production line operation method comprising the following steps:

[0157] T1: The chamber enters the testing area;

[0158] T2: Identify the markings on the cabinets entering the detection area, such as using a barcode scanner to identify the SN code on the cabinets. Once the scan is successful, the depth camera is triggered to take a picture.

[0159] T4: Capture the area to be detected;

[0160] T7: Detect whether there is a leak according to the leak detection method described above.

[0161] If not, issue a second display signal and enter T8.

[0162] If so, issue a third display signal and enter T9;

[0163] T8: Seal the box;

[0164] T9: Manually verify if any items are missing.

[0165] If so, manually replace the missing parts and proceed to T2.

[0166] If not, proceed to T8.

[0167] In another embodiment, the step between step T2 and step T4 further includes:

[0168] T3: Determine whether this enclosure has been inspected.

[0169] If so, delete the identification data corresponding to this identifier from the previous step, and then proceed to T4.

[0170] If not, proceed directly to T4.

[0171] In another embodiment, the step between step T4 and step T7 further includes:

[0172] T5: Determine if the box exists.

[0173] If so, issue the first display signal and enter T6.

[0174] If not, proceed to T7;

[0175] T6: After the worker manually moves the container to the designated shooting area, T2 is repeated. The worker can move the container to the designated shooting area, trigger the barcode scanner to scan the SN code, and after scanning the code again, the camera will be triggered again to collect data.

[0176] As a preferred embodiment, the first display signal, the second display signal, and the third display signal are emitted by a three-color alarm light, with different colored lights representing different signals. For example, the first display signal, the second display signal, and the third display signal correspond to yellow, green, and red, respectively.

[0177] like Figure 3 The diagram shows a preferred embodiment of a production line operation method according to the present invention, and the specific steps are as follows:

[0178] T1: The chamber enters the testing area;

[0179] T2: Identify the markings on the box when entering the detection area;

[0180] T3: Determine whether this enclosure has been inspected.

[0181] If so, delete the identification data corresponding to this identifier from the previous step, and then proceed to T4.

[0182] If not, proceed directly to T4;

[0183] T4: Capture the area to be detected;

[0184] T5: Determine if the box exists.

[0185] If so, issue the first display signal and enter T6.

[0186] If not, proceed to T7;

[0187] T6: After manually moving the box to the designated shooting area, repeat T2;

[0188] T7: Detect whether there is a leak according to the leak detection method described above.

[0189] If not, issue a second display signal and enter T8.

[0190] If so, issue a third display signal and enter T9;

[0191] T8: Seal the box;

[0192] T9: Manually verify if any items are missing.

[0193] If so, manually replace the missing parts and proceed to T2.

[0194] If not, proceed to T8.

[0195] like Figure 4 As shown in the figure, another embodiment of the present invention provides an apparatus including a shooting module, an image processing module, a manual designation module, a detection and judgment module, and a storage module;

[0196] The camera module is used to collect data from the box under test; for example, to acquire depth data, RGB image data, etc. of the area to be tested.

[0197] The image processing module is used to process the data acquired by the shooting module; for example, it performs image processing operations such as converting RGB image data into HSV color space image data, performing binarization processing on HSV color space image data, performing normalization calculation on data within a specified range to form a grayscale image, converting grayscale image data into RGB image data, performing image filtering on the HSV color space image data to retain only pixels that meet manually calibrated parameters, and obtaining the box outline and the first image where the box outline is located, etc.

[0198] The manual specification module is used to receive manually calibrated or specified parameters, which include one or more of the following: specifying the image analysis area, extracting the data range of the V color space, reading area data, specifying area data, and specifying the center point;

[0199] The detection and judgment module identifies and judges the data processed by the image processing module and / or the data collected by the shooting module based on the manually calibrated or specified data.

[0200] The storage module is used to store one or more of the following: data collected by the shooting module, intermediate or result data from the image processing module, and manually calibrated or specified data.

[0201] like Figure 4 As shown in the figure, another embodiment of the present invention provides a system that, in addition to all the modules in the above-described device, further includes: an identification module and a display module;

[0202] The identification module is used to identify the markings on the box and trigger the shooting module to collect data of the box under test;

[0203] The display module is used to visualize the detection results given by the detection and judgment module. The display module emits different display signals corresponding to different detection results. For example, the display module includes alarm lights, which provide three different colored lights at T5 and T7 when the system is applied to the production line operation method.

[0204] In another embodiment, the identification module is a barcode scanner and / or an RFID reader, and the markings on the housing are graphic codes and / or RFID chips. As a preferred option, an SN code and a Bluetooth wireless barcode scanner are used.

[0205] like Figure 5 As shown, another embodiment of the present invention provides a readable storage medium containing a computer program, characterized in that, when the computer program is executed by a processor, the processor performs any of the leak detection methods described above. A Raspberry Pi motherboard can be used as the edge computing power.

[0206] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary hardware. With this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product, or it can be manifested through the implementation process of data migration. This computer software product can be stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, mobile terminal, server, or network device, etc.) to execute the methods described in various embodiments or certain parts of the embodiments of this application.

[0207] Some modules of the apparatus or system described in this application can be described in the general context of computer-executable instructions that are executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, classes, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0208] In summary, this invention enables the detection of missing parts in production line components. It allows for rapid deployment and, even without specialized personnel, can adjust and deploy the leak detection method based on a minimal sample size, significantly saving R&D resources and deployment time, thus improving efficiency. The algorithm parameters can be adjusted to adapt to new production lines or packaging materials by simply collecting a few photos and following a sequential operation process. It requires neither specialized technical personnel nor GPU computing resources, and allows for rapid adjustment and deployment of the leak detection method. This solution is applicable not only to pre-packaging leak detection but also to structural defect detection during production.

[0209] This addresses the problem that existing image recognition algorithms using AI methods for missed detection require specialized technical personnel, GPU computing resources, and significant development and debugging time.

[0210] The above description is merely a preferred embodiment of this application and is not intended to limit this application in any way. Although this application has disclosed the preferred embodiment as above, it is not intended to limit this application. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the technical solution of this application. Any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of this application without departing from the content of the technical solution of this application shall still fall within the scope of the technical solution of this application.

Claims

1. A leak detection method, characterized in that, include: Collect data from the enclosure under test; The data collected from the test chamber includes: The markings on the box that have entered the area to be detected have been identified. The depth data of the area to be detected is acquired by capturing images; Extract data within a specified interval, wherein the specified interval is defined within the depth data, and the specified interval is the depth interval corresponding to the distance from the conveyor belt plane to the top of the product box; Convert the data within the specified interval into RGB image data; The RGB image data is converted into HSV color space image data after binarization, and image filtering is performed to retain pixels that meet the calibration parameters, thereby obtaining the box outline, the first image containing the box outline, and the filtered data area. The first image is cropped according to the box outline, while retaining the image data within the outer outline of the box; Adjust the first image to make it upright and horizontal; The first image is cropped, retaining the image within a preset specified area data to form the second image; The shape and outline of the accessory placed inside the box are obtained by identifying the second image through area filtering; A minimum outer rectangle is used, which encompasses the outline of the accessory; Calculate the center point of the component, and calculate the distance between the center point of the component and a preset designated center point; Determine whether the distance between the center point of the accessory and a preset designated center point is less than a second preset value. If yes, it is qualified; otherwise, it is unqualified. Receive specified parameters; A detection model is formed based on the specified parameters; The data of the enclosure to be tested is input into the detection model for identification to obtain the leak detection result; the detection model is used to determine whether the area of ​​the filtered data is less than a first preset value. If it is, the enclosure is qualified; otherwise, it is unqualified; the detection model is also used to perform leak detection on the enclosure to determine whether it is qualified; the leak detection result is determined based on the result of the enclosure and the leak detection. The receiving of specified parameters includes: Read sample RGB image data; Convert the sample RGB image data into sample HSV color space image data; Receive parameter calibration for the sample HSV color space image data.

2. The leak detection method as described in claim 1, characterized in that: The parameter calibration includes the following three items: specifying the image analysis area, extracting the data range of the V color space, and reading the area data.

3. The leak detection method as described in claim 1 or 2, characterized in that: The step of converting the data within the specified interval into RGB image data includes: The data within the specified interval is normalized to form a grayscale image; Convert the grayscale image data into RGB image data; Store the RGB image data.

4. The leak detection method as described in claim 3, characterized in that: The process of converting the RGB image data into binarized HSV color space image data, performing image filtering, retaining pixels that conform to the calibration parameters, and obtaining the box outline, the first image containing the box outline, and the filtered data area includes: Read the RGB image data; Convert the RGB image data into HSV color space image data; The HSV color space image data is binarized. Image filtering is performed on the binarized HSV color space image data to retain pixels that conform to the calibration parameters, thereby obtaining the box outline, the first image containing the box outline, and the filtered data area.

5. The leak detection method as described in claim 4, characterized in that: The first preset value is 10% of the preset specified area data.

6. The leak detection method as described in claim 4, characterized in that: The second preset value is 10% of the diagonal distance of the box.

7. The leak detection method as described in claim 4, characterized in that: After determining whether the distance between the center point of the component and a preset designated center point is less than a second preset value, if yes, it is qualified; otherwise, it is unqualified, the following steps are also included: Determine if there are multiple components that need to be inspected. If so, perform the step of cropping the first image according to the box outline for each component, retaining the image data within the outer outline of the box.

8. The leak detection method as described in claim 1 or 2, characterized in that: The process of obtaining the leak detection result specifically includes: when the box image recognition based on the specified parameters is deemed qualified, and when the leak detection is deemed qualified, the leak detection result is qualified; otherwise, it is unqualified.

9. A production line operating method, characterized in that, Based on the leak detection method according to any one of claims 1 to 8, the production line operation method includes the following steps: T1: The chamber enters the testing area; T2: Identify the markings on the box when entering the detection area; T4: Capture the area to be detected; T5: Determine if the box exists. If so, enter T7. If not, issue the first display signal and proceed to T6; T6: After manually moving the box to the designated shooting area, repeat T2; T7: Detect whether there is a leak according to the leak detection method described above. If not, issue a second display signal and enter T8. If so, issue a third display signal and enter T9; T8: Seal the box; T9: Manually verify if any items are missing. If so, manually replace the missing parts and proceed to T2. If not, proceed to T8.

10. The production line operation method as described in claim 9, characterized in that: Between step T2 and step T4, the following also applies: T3: Determine whether this enclosure has been inspected. If so, delete the identification data corresponding to this identifier from the previous step, and then proceed to T4. If not, proceed directly to T4.

11. An apparatus, characterized in that: The device is used to perform the leak detection method according to claims 1 to 8 and / or the production line working method according to claims 9 to 10, and the device includes a shooting module, an image processing module, a manual designation module, a detection and judgment module, and a storage module. The imaging module is used to collect data of the box to be tested; specifically, the imaging module is used to identify the markings on the box that enter the area to be tested; specifically, the imaging module is used to capture and obtain the depth data of the area to be tested. The image processing module is used to process the data collected by the shooting module; specifically, the image processing module is used to extract data within a specified interval, which is specified within the depth data, and the specified interval is the depth interval corresponding to the distance from the conveyor belt plane to the top of the product box; convert the data within the specified interval into RGB image data; convert the RGB image data into binarized HSV color space image data, and perform image filtering to retain pixels that meet the calibration parameters, thereby obtaining the box outline, the first image containing the box outline, and the filtered data area; The image processing module is used to crop the first image according to the outline of the box, retaining the image data within the outer outline of the box; adjust the first image to make it upright and horizontal; crop the first image, retaining the image within a preset specified area data to form a second image; identify the shape and outline of the accessory placed inside the box by area filtering; and use a minimum outer rectangle, which includes the outline of the accessory. Calculate the center point of the component and the distance between the center point of the component and a preset designated center point; determine whether the distance between the center point of the component and the preset designated center point is less than a second preset value. If yes, it is qualified; otherwise, it is unqualified. The manual specification module is used to receive manually calibrated or specified parameters, which include one or more of the following: specifying the image analysis area, extracting the data range of the V color space, reading area data, specifying area data, and specifying the center point; The detection and judgment module identifies and judges the data processed by the image processing module and / or the data collected by the shooting module based on the manually calibrated or specified data. Specifically, the detection and judgment module is used to determine whether the area of ​​the filtered data is less than a first preset value; if so, it is qualified; otherwise, it is unqualified. The detection and judgment module is also used to perform leak detection on the housing and determine whether it is qualified. The leak detection result is determined based on the result of the housing and the leak detection. The storage module is used to store one or more of the following: data collected by the shooting module, intermediate or result data from the image processing module, and manually calibrated or specified data.

12. The apparatus as claimed in claim 11, characterized in that: The imaging module includes a depth camera and / or a lidar.

13. A system, characterized in that: The device, comprising the apparatus of claim 11, further includes: an identification module and a display module; The identification module is used to identify the markings on the box and trigger the shooting module to collect data of the box under test; The display module is used to visualize the detection results given by the detection and judgment module, and the display module emits different display signals for different detection results.

14. The system as described in claim 13, characterized in that: The identification module is a barcode scanner and / or an RFID reader, and the markings on the box are graphic codes and / or RFID chips.

15. A readable storage medium containing a computer program, characterized in that, When the computer program is executed by the processor, the processor performs the leak detection method according to any one of claims 1 to 8.

Images