Airtightness detection method, device and system for sealed container

Abstract

The present invention provides a method, device and system for detecting the airtightness of a sealed container, which belongs to the field of industrial visual inspection technology, and includes: obtaining at least one first video stream and a second video stream that are aligned in time and space; obtaining a starting video frame in the first video stream, and intercepting a first video frame sequence starting from the starting video frame, wherein the starting video frame contains a bubble image; intercepting a second video frame sequence from the second video stream according to the timestamp of the starting video frame; comparing the first bubble displacement information determined by the first video frame sequence and the second bubble displacement information determined by the second video frame sequence, and determining the detection result. The method, device and system for detecting the airtightness of a sealed container provided by the present invention, and the proposed collaborative bubble trajectory tracking can effectively reduce false alarms and missed alarms caused by interference such as suspended particles in water and bubbles attached to the surface of a sealed container, and improve the accuracy of airtightness detection.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial visual inspection, and in particular to an airtightness inspection method, device and system for a sealed container. Background Art

[0002] Air tightness detection technology, such as air tightness water detection technology, is currently widely used for air tightness detection of closed containers such as metal liners. In this process, a certain high-pressure gas is filled into the closed container and pressed into water. Then, whether bubbles are generated in the water is observed manually or by machine vision to determine whether the closed container is leaking.

[0003] Generally, the entire side wall of the water tank is replaced with transparent tempered glass to facilitate manual or camera imaging observation from the outside of the water tank. Since some small bubbles usually adhere to the surface of the sealed container during the water entry process, and the material of the sealed container itself has optical properties such as reflection, human eye observation or conventional image recognition is prone to false detection or missed detection of air leaks. Summary of the invention

[0004] The present invention provides a method, device and system for detecting the airtightness of a sealed container, which are used to solve the defects of false detection or missed detection in watertightness detection in the prior art and realize high-precision airtightness detection of the sealed container.

[0005] The present invention provides a method for detecting the airtightness of a sealed container, comprising the following steps:

[0006] Acquire at least one first video stream and a second video stream that are aligned in time and space, wherein the first video stream is obtained by shooting a sealed container to be tested placed in a tank, and the second video stream is obtained by shooting an upper space in the tank located at the sealed container to be tested; the tank is loaded with a liquid, and the liquid immerses the sealed container to be tested;

[0007] Acquire a starting video frame in the first video stream, and capture a first video frame sequence starting from the starting video frame, wherein the starting video frame includes a bubble image;

[0008] intercepting a second video frame sequence from the second video stream according to the timestamp of the starting video frame;

[0009] The first bubble displacement information determined by the first video frame sequence and the second bubble displacement information determined by the second video frame sequence are compared to determine a test result of the airtightness test on the sealed container to be tested.

[0010] According to an airtightness detection method for a sealed container provided by the present invention, determining the first bubble displacement information from the first video frame sequence specifically includes:

[0011] Inputting the first video frame sequence into a video classification model to obtain an output result of the video classification model;

[0012] The output result includes whether a dynamic bubble is identified and the first bubble displacement information of the dynamic bubble; the first bubble displacement information is used to indicate the running track of the dynamic bubble;

[0013] The video classification model is obtained after training with 3D CNN as the initial model.

[0014] According to a method for detecting airtightness of a sealed container provided by the present invention, the step of obtaining a starting video frame in the first video stream includes:

[0015] Extracting a video frame to be tested from the first video stream according to a preset sampling frequency;

[0016] Inputting the video frames to be tested into the target recognition model in sequence, and obtaining the detection results output by the target recognition model;

[0017] When it is determined that any detection result is an image with bubbles, obtaining a video frame to be tested corresponding to any detection result as the starting video frame;

[0018] The target recognition model is obtained after training using multiple video frame samples and a detection result label corresponding to each video frame sample.

[0019] According to an airtightness detection method for a sealed container provided by the present invention, determining the second bubble displacement information by the second video frame sequence specifically includes:

[0020] Performing bubble image detection on each video frame in the second video frame sequence, screening out all target video frames containing bubble images, and marking an area of ​​interest related to the bubble image in each target video frame;

[0021] For any region of interest, the region of interest is divided into a plurality of image grids, and displacement information between corresponding image grids in two adjacent target video frames is traversed and calculated to determine a velocity field of the region of interest;

[0022] The velocity fields of all the regions of interest are acquired to determine the second bubble displacement information; the second bubble displacement information is used to indicate the running trajectory of the dynamic bubble.

[0023] According to a method for detecting airtightness of a sealed container provided by the present invention, before intercepting a second video frame sequence from the second video stream according to the timestamp of the starting video frame, the method further comprises:

[0024] The second video stream is subjected to image grayscale processing.

[0025] According to a method for detecting airtightness of a sealed container provided by the present invention, before dividing the region of interest into a plurality of image grids, the region of interest is normalized, specifically comprising:

[0026] Calculate the gray value of each pixel in the region of interest to calculate the average gray intensity;

[0027] Calculating the sum of squares of differences between the grayscale value of each pixel and the average grayscale intensity, and determining a normalization factor according to the sum of squares of differences and the total number of pixels;

[0028] The normalized pixel value of each pixel is determined according to the relationship between the difference between the grayscale value of each pixel and the average grayscale intensity and the normalization factor.

[0029] According to a method for detecting airtightness of a sealed container provided by the present invention, the step of comparing first bubble displacement information determined by the first video frame sequence and second bubble displacement information determined by the second video frame sequence to determine a detection result of airtightness detection of the sealed container to be detected specifically includes:

[0030] Determine at least one first running trajectory according to the first bubble displacement information, and determine at least one second running trajectory according to the second bubble displacement information;

[0031] If it is determined that at least one of the first running trajectories and any of the second running trajectories are running trajectories of the same dynamic bubble, then determining that the detection result is unqualified;

[0032] Otherwise, the test result is determined to be qualified.

[0033] According to a method for detecting airtightness of a sealed container provided by the present invention, after acquiring the second video stream, the method further comprises:

[0034] Extracting at least one video frame in the second video stream as a liquid quality detection image;

[0035] Acquiring the image brightness of the liquid quality detection image;

[0036] If it is determined that the image brightness is less than a preset brightness threshold, sending a cleaning instruction to the window cleaning device;

[0037] The cleaning instruction is used to drive the window cleaning device to clean the shooting window of the tank body;

[0038] The preset brightness threshold is the image brightness of the water quality detection sample image, and the water quality detection sample image is obtained by pre-shooting the tank body in a clean state.

[0039] The present invention also provides an airtightness detection device for a sealed container, comprising:

[0040] A video acquisition unit, used to acquire at least one first video stream and a second video stream that are aligned in time and space, wherein the first video stream is obtained by shooting a sealed container to be tested placed in a slot, and the second video stream is obtained by shooting an upper space in the slot located at the sealed container to be tested; the slot is loaded with a liquid, and the liquid immerses the sealed container to be tested;

[0041] a video preprocessing unit, configured to obtain a starting video frame in the first video stream, and intercept a first video frame sequence starting from the starting video frame, wherein the starting video frame includes a bubble image; and intercept a second video frame sequence from the second video stream according to a timestamp of the starting video frame;

[0042] The detection operation unit is used to compare the first bubble displacement information determined by the first video frame sequence and the second bubble displacement information determined by the second video frame sequence to determine the detection result of the airtightness detection of the sealed container to be tested.

[0043] The present invention also provides an airtightness detection system for a sealed container, comprising: a tank body, at least two cameras and a processor; the tank body is loaded with liquid, and the liquid immerses the sealed container to be tested;

[0044] A transparent window is provided on at least one side wall of the trough body, and the camera respectively shoots a first video stream and a second video stream through the transparent window, wherein the first video stream is obtained by shooting the sealed container to be tested placed in the trough body, and the second video stream is obtained by shooting the upper space of the sealed container to be tested in the trough body;

[0045] The camera is communicatively connected to the processor to synchronously transmit the first video stream and the second video stream to the processor;

[0046] When the processor is running, it executes any one of the above-mentioned methods for detecting the airtightness of a sealed container.

[0047] According to the airtightness detection system for a sealed container provided by the present invention, the number of the transparent window is one, and it is a rectangular window; the size of the rectangular window is determined based on the following method:

[0048] Determine the width of the rectangular window according to a first distance between the camera lens and the rectangular window, a second distance between the camera lens and the sealed container to be measured in the tank, and a width of the sealed container to be measured;

[0049] The length of the rectangular window is determined according to the first distance, the second distance and the width of the sealed container to be measured.

[0050] According to the airtightness detection system for a sealed container provided by the present invention, the number of the transparent windows is two, and the first transparent window is located below the second transparent window;

[0051] The first transparent window is used by the first camera to capture the first video stream;

[0052] The second transparent window is used by the second camera to capture the second video stream.

[0053] According to the airtightness detection system for a sealed container provided by the present invention, the number of the transparent windows is multiple, and the number of the cameras is also multiple, wherein at least one transparent window is provided for one of the cameras to collect the second video stream;

[0054] The number of the remaining multiple transparent windows is determined by the camera pixel utilization ratio determined by the aspect ratio of the effective photosensitive area of ​​the camera and the aspect ratio of the sealed container to be measured.

[0055] According to the airtightness detection system for a sealed container provided by the present invention, reinforcing ribs are additionally provided on the outside of the transparent window;

[0056] The arrangement of the reinforcing ribs is determined based on the maximum deflection of the center point of the transparent window.

[0057] The airtightness detection system for a sealed container provided according to the present invention further includes: determining a shooting interval for shooting the upper space of the sealed container to be tested; and setting a background for the shooting interval, wherein there is a visual contrast between the background and the bubbles.

[0058] The airtightness detection system for a closed container provided by the present invention further comprises a window cleaning device; when the window cleaning device is in operation, it is used to clean the transparent window.

[0059] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for detecting the airtightness of a sealed container as described above is implemented.

[0060] The present invention also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the method for detecting the airtightness of a sealed container as described in any one of the above is implemented.

[0061] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method for detecting the airtightness of a closed container as described above is implemented.

[0062] The airtightness detection method, device and system for closed containers provided by the present invention and the proposed collaborative bubble trajectory tracking can effectively reduce false alarms and missed alarms caused by interference such as suspended particles in water and bubbles attached to the surface of closed containers, thereby improving the accuracy of airtightness detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0064] Figure 1 It is one of the flow diagrams of the airtightness detection method of the airtight container provided by the present invention.

[0065] Figure 2 This is the second flow chart of the airtightness detection method for a sealed container provided by the present invention.

[0066] Figure 3 It is a working schematic diagram of the window cleaning device provided by the present invention.

[0067] Figure 4 It is a control flow diagram of the window cleaning device provided by the present invention.

[0068] Figure 5 It is a structural schematic diagram of the airtightness detection device for a closed container provided by the present invention.

[0069] Figure 6 It is a schematic structural diagram of a single transparent window provided by the present invention.

[0070] Figure 7 It is a schematic diagram of the imaging optical path provided by the present invention.

[0071] Figure 8 This is one of the schematic diagrams of the shape of the transparent window provided by the present invention;

[0072] Fig. 9 This is the second schematic diagram of the shape of the transparent window provided by the present invention.

[0073] Fig.10 It is a schematic diagram of the distribution of two transparent windows provided by the present invention.

[0074] Fig.11 It is a schematic diagram of dividing the observation areas of multiple cameras using an inner tank with a high aspect ratio provided by the present invention.

[0075] Fig.12 It is a schematic diagram of imaging of two transparent windows provided by the present invention.

[0076] Fig.13 It is a schematic diagram of setting an independent bubble tracking window on two transparent windows provided by the present invention.

[0077] Fig.14 It is a schematic diagram of the structure of multiple transparent windows provided by the present invention.

[0078] Fig.15 It is one of the schematic diagrams of the reinforcement arranging method provided by the present invention.

[0079] Fig.16 This is the second schematic diagram of the reinforcement arranging method provided by the present invention.

[0080] Fig.17 A schematic structural diagram of an electronic device provided by the invention. DETAILED DESCRIPTION

[0081] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0082] It should be noted that in the description of the present invention, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "include one..." do not exclude the existence of other identical elements in the process, method, article or device including the elements. The orientation or position relationship indicated by the terms "upper", "lower" and the like is based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present invention. Unless otherwise clearly specified and limited, the terms "installed", "connected" and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be a connection between the two elements. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0083] The terms "first", "second", etc. in the present invention are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present invention can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are generally of the same type, and the number of objects is not limited. For example, the first object can be one or more.

[0084] Combine the following Figure 1-Figure 17 The invention describes the method, device and system for detecting the airtightness of a sealed container.

[0085] Figure 1 It is one of the flow diagrams of the airtightness detection method of the sealed container provided by the present invention, such as Figure 1 As shown, including but not limited to the following steps:

[0086] Step 101, obtaining at least one first video stream and second video stream that are time-space aligned, wherein the first video stream is obtained by shooting a sealed container to be tested placed in a tank, and the second video stream is obtained by shooting an upper space in the tank located at the sealed container to be tested. The tank is loaded with liquid, and the liquid immerses the sealed container to be tested.

[0087] In this embodiment, the execution subject can be an on-site industrial computer, a remote computer, or a cloud server or other processor with image processing, data storage, and data calculation functions, and there is no specific limitation on this.

[0088] Among them, the closed container refers to a container with a formed outer wall and a closed cavity inside. The main purpose of the present invention is to confirm whether the internal cavity of the closed container is leaking by liquid detection. In order to simplify the specific description of the embodiment, the airtightness detection of the metal liner is used as an example in the subsequent embodiments, which is not regarded as a specific limitation on the scope of protection of the present invention.

[0089] The tank is a container that can hold liquids. By immersing a sealed container in liquid, if the sealed container (usually the sealed container is first inflated so that the gas pressure inside is greater than the atmospheric pressure to a certain extent) has an airtightness defect, gas will be released outward from the leaking part, and bubbles will be generated in the liquid. In this way, by observing the formation of bubbles and the state of the bubbles (including size, displacement speed, etc.), it is possible to accurately determine whether the sealed container has an airtightness defect.

[0090] Generally speaking, water testing is performed by injecting ordinary tap water into the tank body, but sometimes other liquids such as alcohol, oil, etc. may be used in consideration of the corrosiveness of water on the sealed container to be tested or the influence of other properties, which is not specifically limited in the present invention. In the case of using tap water (or purified water) for water testing, the tank body can be simply referred to as a water tank or a water tank.

[0091] Specifically, two or more high-resolution industrial cameras can be used to perform the video acquisition work of the air tightness test, at least one of which is used as a metal liner observation window camera, which is mainly used to completely shoot the metal liner to be tested placed in the tank, obtain the state of the metal liner under water, and detect whether there are bubbles caused by air leakage on the surface of the metal liner, as well as other abnormalities on the surface of the liner. This camera can be referred to as camera 1. Another camera is used as a bubble independent window camera, which is mainly used for the space above the metal liner, especially focusing on the rising trajectory of the bubble in the water. It can capture the rising trajectory and characteristics of the bubble by shooting the bubble dynamics in the space above the metal liner, which is used for subsequent bubble trajectory tracking and bubble property analysis. This camera can be referred to as camera 2.

[0092] In order to ensure that the video streams captured by all cameras are aligned in time, a high-precision time synchronization device, such as a GPS time synchronization module or a network time protocol (NTP) server, can be used to ensure that the clock error of multiple cameras is within milliseconds. Alternatively, a simple trigger can be used to trigger synchronous acquisition.

[0093] In addition, it is also necessary to ensure that the video streams collected by each camera are spatially aligned. This is because it is necessary to ensure that the complete image of the metal liner spliced ​​together from the photos of the metal liner taken by each camera at the same time is consistent with the actual situation. Taking the alignment of the video streams of two cameras as an example, at the same moment, if the video frames in the video stream taken by camera 1 and the video frames in the video stream taken by camera 2 have no intersection at all, it means that there are areas that are not fully observed by the two cameras, and there may be missed detections at this time. Therefore, in actual use, it is necessary to ensure that the shooting areas of multiple cameras overlap slightly, and then use the relevant spatial alignment algorithm to spatially align the video streams taken by multiple cameras to remove the overlapping areas.

[0094] The video stream captured by camera 1 is called the first video stream, and the video stream captured by camera 2 is called the second video stream. Step 101 provides synchronous and corresponding video data for subsequent bubble detection, trajectory tracking and leakage judgment by collecting the two video streams aligned in time and space.

[0095] Step 102, obtaining a starting video frame in the first video stream, and capturing a first video frame sequence starting from the starting video frame, wherein the starting video frame includes a bubble image.

[0096] First, a frame in which bubbles first appear or reach a specific detection threshold is identified in the first video stream captured by camera 1 as a starting video frame.

[0097] Then, starting from the start video frame, a series of video frames are captured according to a predetermined time interval or frame number to form a first video frame sequence.

[0098] Step 103: intercept a second video frame sequence from the second video stream according to the timestamp of the starting video frame.

[0099] At the same time, according to the timestamp of the starting video frame, a corresponding video frame sequence is synchronously intercepted from the second video stream captured by camera 2 as the second video frame sequence to ensure the time correspondence between the two video frame sequences.

[0100] For example, at the 10th second, the generation of the first bubble is detected in the first video stream of camera 1, and this frame is marked as the starting video frame. Subsequently, starting from the 10th second, a frame is captured every 50ms to form the first video frame sequence containing the rising process of the bubble.

[0101] When the timestamp of the starting video frame is determined to be 10 seconds, a frame can be captured every 50 ms starting from the 10th second of the second video stream captured by camera 2 to form a second video frame sequence. This sequence includes the trajectory of the bubble floating up in the upper space.

[0102] Through the above steps 102-103, it can be ensured that the video frame sequences captured by camera 1 and camera 2 are aligned in time and space, which is crucial for subsequent bubble trajectory tracking, false alarm rate reduction and fully automated detection. At the same time, the shooting objects and functions of camera 1 and camera 2 are clarified, which helps to better understand the operating principle of the entire detection system.

[0103] Step 104, comparing the first bubble displacement information determined by the first video frame sequence and the second bubble displacement information determined by the second video frame sequence, and determining the test result of the airtightness test on the sealed container to be tested.

[0104] For each pair of corresponding video frames in the first video frame sequence and the second video frame sequence, image processing techniques (such as edge detection, feature extraction, etc.) can be used to extract displacement information of the bubbles. Then, by comparing the displacement information, it is possible to accurately determine whether the rising trajectories of the bubbles are consistent, thereby determining whether there is air leakage in the liner.

[0105] For example, for each pair of video frames in the first video frame sequence and the second video frame sequence, an edge detection algorithm is used to extract the outline of the bubble and calculate the coordinates of its center point. Then, the change trends of these coordinates are compared.

[0106] If it is found that the rising trajectory of the bubble in camera 2 is basically the same as the floating trajectory in camera 1, it means that the bubble is a dynamic bubble rather than a static bubble brought in when the metal liner enters the water, so the bubble is generated by leakage of the liner.

[0107] The airtightness detection method of the sealed container provided by the present invention successfully utilizes the video streams captured by two cameras and realizes high-precision detection of the airtightness of the sealed container through the method of time-space alignment and displacement information comparison. This method not only improves the accuracy and reliability of detection, but also reduces false detection and missed detection caused by factors such as optical characteristics and bubble adhesion.

[0108] Based on the content of the above embodiment, as an optional embodiment, determining the first bubble displacement information by the first video frame sequence specifically includes:

[0109] Inputting the first video frame sequence into a video classification model to obtain an output result of the video classification model;

[0110] The output result includes whether a dynamic bubble is identified and the first bubble displacement information of the dynamic bubble; the first bubble displacement information is used to indicate the running track of the dynamic bubble;

[0111] The video classification model is obtained after training with 3D CNN as the initial model.

[0112] Figure 2 FIG. 2 is a flow chart of the airtightness detection method of the sealed container provided by the present invention. Figure 2 As shown, this embodiment focuses on how to use a three-dimensional convolutional neural network model (3D CNN) to combine image processing and deep learning technology to process the first video frame sequence to determine the first bubble displacement information, aiming to improve the accuracy and efficiency of airtightness detection.

[0113] First, the method provided in the above embodiment is used to analyze the first video stream collected by camera 1 (i.e., the inner liner observation window camera) to extract the first video frame sequence. If there is a gas leak in the metal liner, the first video frame sequence will record continuous images of the bubble from generation to rise.

[0114] The first video frame sequence may be preprocessed, including but not limited to image denoising, graying, resizing, etc., to reduce the computational complexity of subsequent processing and improve the recognition accuracy of the video classification model.

[0115] The preprocessed first video frame sequence is used as input to the trained 3D CNN video classification model. The video classification model is based on a three-dimensional convolutional layer and can capture the spatiotemporal features in the video, namely the motion information of the bubbles in the time dimension and the position changes in the space dimension.

[0116] The output result of the obtained video classification model mainly includes whether dynamic bubbles are identified and the first bubble displacement information of the dynamic bubbles. Among them, whether dynamic bubbles are identified is a binary classification result, indicating whether dynamic bubbles generated by air leakage can be detected on the surface of the metal liner in the first video frame sequence. For the identified dynamic bubbles, the video classification model will output their running trajectory information, including the change of the position coordinates in the video frame over time.

[0117] According to the first bubble displacement information of the dynamic bubbles obtained, combined with the observation results of camera 2 (i.e., the bubble independent tracking window camera), mainly including the second bubble displacement information obtained by identifying the dynamic bubbles, the two are aligned in time and space and the bubble trajectory is tracked. By comparing whether the bubble trajectories under the perspectives of the two cameras are consistent, it is determined whether the detected bubbles are dynamic bubbles, so that it is possible to accurately determine whether the metal liner has a leak.

[0118] Assume that during the air tightness test of the liner, camera 1 captures a bubble suspected to be caused by a leak and generates a corresponding first video frame sequence. After the sequence is fed into the 3D CNN model, the output shows that the dynamic bubble is identified and the first bubble displacement information of the bubble is given (equivalent to the running trajectory of the bubble). At the same time, camera 2 also observes the liner in the same time period and can determine the second bubble displacement information. By comparing the bubble trajectories under the perspectives of the two cameras, it is found that the two are consistent, thus judging that there is a leak in the liner.

[0119] The airtightness detection method provided by the present invention uses a 3D CNN model to process a video frame sequence, which can more accurately identify dynamic bubbles and extract their displacement information, thereby reducing false positives and missed positives.

[0120] Based on the content of the above embodiment, as an optional embodiment, obtaining the starting video frame in the first video stream includes:

[0121] Extracting a video frame to be tested from the first video stream according to a preset sampling frequency;

[0122] Inputting the video frames to be tested into the target recognition model in sequence, and obtaining the detection results output by the target recognition model;

[0123] When it is determined that any detection result is an image with bubbles, obtaining a video frame to be tested corresponding to any detection result as the starting video frame;

[0124] The target recognition model is obtained after training using multiple video frame samples and a detection result label corresponding to each video frame sample.

[0125] This embodiment describes in detail how to automatically detect and obtain the starting video frame in the first video stream, that is, the video frame where the bubble first appears or changes significantly. The video frame to be tested is extracted by a preset sampling frequency, and the bubble detection is performed using a trained target recognition model, so as to accurately determine the starting video frame. The specific implementation steps may include but are not limited to:

[0126] First, according to a preset sampling frequency (such as 10 frames per second), the video frames to be tested are uniformly extracted from the first video stream (taken by camera 1). This can reduce the amount of data to be processed while retaining enough information to detect the appearance of bubbles.

[0127] Each extracted video frame to be tested is sequentially input into a pre-trained target recognition model. The target recognition model is a classifier trained by deep learning and can identify whether there are bubbles in the image and its bubble characteristics.

[0128] After processing each video frame to be tested, the target recognition model will output a detection result indicating whether the video frame to be tested contains a bubble image. The detection results of all video frames to be tested are traversed until the first video frame to be tested marked as containing a bubble image is found.

[0129] When any detection result is determined to be a bubble image, the video frame to be tested corresponding to the detection result is immediately obtained as the starting video frame, that is, the video frame where the bubble first appears or changes significantly, which is the starting point for subsequent bubble trajectory tracking.

[0130] It should be noted that the above target recognition model is obtained after training with multiple video frame samples containing bubbles and without bubbles, as well as the detection result label (bubble / no bubble) corresponding to each video frame sample. During the training process, the model learns the feature representation of bubbles and optimizes its classification performance to ensure that bubbles can be accurately identified in practical applications.

[0131] For example, during the air tightness test of the inner liner, camera 1 captures the first video stream at a speed of 30 frames per second. The video frame to be tested is extracted from the video stream according to the preset sampling frequency (10 frames per second). For each frame of the video to be tested, the system inputs it into the target recognition model for detection. When the 5th frame is detected, the model outputs an image with bubbles. Therefore, the system determines the 5th frame as the starting video frame, and uses it as the starting point for subsequent bubble trajectory tracking.

[0132] The airtightness detection method for a closed container provided by the present invention reduces the amount of data to be processed by presetting the sampling frequency, and at the same time utilizes the fast processing capability of a deep learning model to improve the efficiency of the entire detection process.

[0133] Based on the content of the above embodiment, as an optional embodiment, determining the second bubble displacement information by the second video frame sequence specifically includes:

[0134] Performing bubble image detection on each video frame in the second video frame sequence, screening out all target video frames containing bubble images, and marking an area of ​​interest related to the bubble image in each target video frame;

[0135] For any region of interest, the region of interest is divided into a plurality of image grids, and displacement information between corresponding image grids in two adjacent target video frames is traversed and calculated to determine a velocity field of the region of interest;

[0136] The velocity fields of all the regions of interest are acquired to determine the second bubble displacement information; the second bubble displacement information is used to indicate the running trajectory of the dynamic bubble.

[0137] The region of interest (ROI) refers to a key area in an image that requires special attention or analysis in image processing, and in this embodiment, refers to an area containing a bubble image.

[0138] The image grid is a plurality of small blocks or regions into which the image is divided for the convenience of analysis and calculation. In the present embodiment, the purpose of the division is to capture the displacement information of the bubble between consecutive frames in a more fine-grained manner.

[0139] The displacement information refers to the amount of change in the position of the bubble between consecutive frames, and in this embodiment includes information such as the position coordinates, velocity magnitude and direction of the bubble at different time points.

[0140] The velocity field is a two-dimensional vector field, which represents the velocity components of each point in the region in different directions. In this embodiment, it is used to describe the motion state of the bubble in the region of interest.

[0141] Combination Figure 2 As shown, this embodiment further refines the specific steps of how to determine the second bubble displacement information based on the second video frame sequence (taken by camera 2, focusing on the detailed observation of the bubble floating trajectory). By combining bubble image detection, region of interest annotation, image grid division and fine displacement calculation and other technologies, high-precision tracking of the dynamic bubble running trajectory is achieved, and the specific steps include but are not limited to:

[0142] First, bubble image detection and region of interest annotation are performed. Bubble image detection can be performed on each video frame in the second video frame sequence by applying a trained bubble detection algorithm or bubble detection model to identify whether the video frame includes a bubble image. All target video frames containing bubble images are screened out, and these target video frames are the basis for subsequent airtightness detection and analysis. In each target video frame, image processing techniques (such as contour detection, edge recognition, etc.) are used to annotate the regions of interest related to the bubble image. These regions of interest should be closely surrounded by the bubble image for subsequent accurate analysis.

[0143] Furthermore, image grid division and bubble displacement information calculation are performed. For any region of interest, it is divided into multiple image grids. The size and shape of the grid can be adjusted according to the actual situation to ensure that the tiny displacement of the bubble between consecutive frames can be accurately captured. The displacement information between the corresponding image grids in two adjacent target video frames is traversed and calculated. This is usually achieved by comparing the pixel value changes of the grids at the same position in adjacent target video frames. Optical flow method, feature matching and other techniques can be used to calculate the displacement vector between grids. Based on the displacement vector between grids, the velocity field of the region of interest is determined. The velocity field is a two-dimensional vector field that represents the velocity components of each pixel point in the region of interest in different directions.

[0144] Finally, determine the second bubble displacement information. After obtaining the velocity fields of all regions of interest, perform a comprehensive analysis. Since bubbles may present complex motion patterns in consecutive frames, it is necessary to fuse or weighted average the velocity fields of each region of interest to obtain the overall bubble displacement information. The second bubble displacement information finally determined is used to indicate the trajectory of the dynamic bubble. This information can be used for subsequent tasks such as bubble behavior analysis, velocity measurement, or trajectory prediction.

[0145] For example, suppose that in the second video frame sequence, camera 2 captures a certain area of ​​the bubble's floating trajectory. First, perform bubble image detection on each video frame, and select the target video frame containing the bubble image. Then, mark the bubble's region of interest in each target video frame. Next, divide each region of interest into multiple small grids, and calculate the displacement information between corresponding grids in adjacent frames. Finally, by comprehensively analyzing the velocity fields of all regions of interest, the displacement information of the bubble in this section is obtained, thereby drawing the bubble's trajectory.

[0146] The airtightness detection method of the sealed container provided by the present invention can accurately capture the tiny displacement of bubbles between consecutive frames through image grid division and displacement calculation, thereby obtaining accurate bubble displacement information. The method is not affected by the shape, size or movement mode of the bubble and can be applied to bubble displacement analysis in a variety of complex scenarios.

[0147] Based on the content of the above embodiment, as an optional embodiment, before intercepting the second video frame sequence from the second video stream according to the timestamp of the starting video frame, the method further includes: performing image grayscale processing on the second video stream.

[0148] This embodiment grayscales each color video frame in the second video frame sequence, retains key information in the video frame, such as brightness, contrast, etc., to facilitate subsequent operations such as feature extraction, target detection or image matching, which can significantly reduce the amount of data required for subsequent processing and improve the processing speed. In addition, after removing the color information, the image analysis algorithm can focus more on the image's brightness, texture and other features, simplifying the processing flow. Under certain conditions, the grayscale process can also enhance the contrast of the video frame, making the target features more prominent, thereby improving the accuracy of subsequent processing steps (such as target detection, tracking, etc.).

[0149] The method of graying the video frame may be implemented by a variety of algorithms, for example, a weighted average method may be used, that is, the gray value is calculated according to different weights of the three color components of red, green and blue in the color image.

[0150] Based on the content of the above embodiment, as an optional embodiment, before dividing the region of interest into a plurality of image grids, normalizing the region of interest specifically includes:

[0151] Calculate the gray value of each pixel in the region of interest to calculate the average gray intensity;

[0152] Calculating the sum of squares of differences between the grayscale value of each pixel and the average grayscale intensity, and determining a normalization factor according to the sum of squares of differences and the total number of pixels;

[0153] The normalized pixel value of each pixel is determined according to the relationship between the difference between the grayscale value of each pixel and the average grayscale intensity and the normalization factor.

[0154] When processing video frames containing bubble images, the zero-mean normalized sum of squared difference (ZNSSD) is used to normalize the region of interest where the bubbles are located. This is to adjust the range of the image data to have a uniform scale, which is convenient for subsequent processing and analysis.

[0155] This embodiment will elaborate on the specific steps of how to perform normalization processing before dividing the region of interest into multiple image grids, and will illustrate this with reference to a specific video frame as an example.

[0156] First, the region of interest is cropped from the video frame, and each pixel in the region of interest is traversed to calculate its grayscale value.

[0157] Then, add up the grayscale values ​​of all pixels and divide by the total number of pixels to get the average grayscale intensity of the region of interest:

[0158] ;

[0159] in, is the average gray intensity, N is the total number of pixels in the region of interest, For the i The gray value of a pixel.

[0160] For each pixel in the region of interest, calculate the difference between its grayscale value and the average grayscale intensity, and find the sum of the squares of these differences:

[0161] ;

[0162] in, is the sum of the squares of the differences.

[0163] The SSD and the total number of pixels are used to determine the normalization factor (NF). The normalization factor should be chosen to ensure that the difference in pixel values ​​can be scaled to a range suitable for subsequent processing, for example:

[0164] ;

[0165] in, is the normalization factor, It is a scaling factor adjusted according to specific needs, which is used to further control the range of normalized pixel values.

[0166] For each pixel in the region of interest, the normalized pixel value is calculated using the difference between its grayscale value and the average grayscale intensity combined with the normalization factor:

[0167] ;

[0168] in, For the i The normalized pixel value of each pixel.

[0169] The calculated normalized pixel value is reassigned to each pixel in the region of interest to complete the normalization process. At this point, the bubble image in the region of interest will be presented in a more uniform grayscale value distribution that is more suitable for subsequent processing.

[0170] The airtightness detection method for a closed container provided by the present invention helps to highlight the difference between bubbles and the background by normalizing the pixel points, making the outline and features of the bubbles more obvious, and can improve the execution efficiency and accuracy of subsequent algorithms.

[0171] Based on the content of the above embodiment, as an optional embodiment, the comparing the first bubble displacement information determined by the first video frame sequence and the second bubble displacement information determined by the second video frame sequence to determine the test result of the airtightness test of the sealed container to be tested specifically includes:

[0172] Determine at least one first running trajectory according to the first bubble displacement information, and determine at least one second running trajectory according to the second bubble displacement information;

[0173] If it is determined that at least one of the first running trajectories and any of the second running trajectories are running trajectories of the same dynamic bubble, then determining that the detection result is unqualified;

[0174] Otherwise, the test result is determined to be qualified.

[0175] In this embodiment, how to combine the bubble displacement information determined in the first video frame sequence and the second video frame sequence to determine the airtightness test result of the sealed container to be tested will be described in detail. This process combines image processing, trajectory tracking and comparative analysis technology to accurately determine whether the container has a leak.

[0176] Combination Figure 2 As shown, in this embodiment, at least two cameras are arranged on the transparent window, wherein camera 1 is the inner liner observation window camera, and camera 2 is the bubble independent tracking window camera, to respectively perform synchronous video acquisition of the metal inner liner. The acquired data form a first video frame sequence and a second video frame sequence respectively. Image processing technology (such as grayscale processing, normalization processing, background difference, frame difference, etc.) is used to extract bubbles from each video frame sequence respectively, and the position change of the bubbles in the continuous video frames is tracked, so as to obtain the respective bubble displacement information.

[0177] According to the first bubble displacement information determined in the first video frame, the continuous change of the bubble in time can be plotted, that is, the first running trajectory, which represents the running status of the bubble observed from the inner tank observation window.

[0178] Similarly, according to the bubble displacement information in the second video frame sequence, a second running trajectory is drawn, which represents the bubble movement observed by the bubble independent tracking window.

[0179] By comparing the first running trajectory and the second running trajectory, it is analyzed whether they have the same motion characteristics, such as direction, speed, etc. If there is at least one first running trajectory that is highly consistent with any second running trajectory in time and space, it is determined to be the running trajectory of the same dynamic bubble.

[0180] If at least one such match is determined, that is, at least one bubble is observed by two windows at the same time and presents a consistent movement trajectory, then it can be considered that the bubble is generated by leakage in the closed container to be tested, and the test result is determined to be unqualified (recorded as NG).

[0181] If no matching running track is found, that is, all observed bubbles fail to form a consistent track in the two windows, then it can be considered that these bubbles may be caused by non-leakage factors such as suspended particles in the water and bubbles attached to the surface of the inner tank, so the test result is determined to be qualified (recorded as OK).

[0182] When the metal liner enters the water, it is easy for some small bubbles to adhere to the surface. In addition, the optical properties of the metal material such as reflection can cause false detection and missed detection of air leaks. To overcome this problem, the present invention provides a new air tightness detection method, which effectively reduces false alarms and missed detections caused by factors such as suspended particles in water and reflection of the liner surface through multi-window synchronous observation and bubble trajectory tracking, and improves the accuracy of air tightness detection.

[0183] Based on the content of the above embodiment, as an optional embodiment, after acquiring the second video stream, the method further includes:

[0184] Extracting at least one video frame in the second video stream as a liquid quality detection image;

[0185] Acquiring the image brightness of the liquid quality detection image;

[0186] If it is determined that the image brightness is less than a preset brightness threshold, sending a cleaning instruction to the window cleaning device;

[0187] The cleaning instruction is used to drive the window cleaning device to clean the shooting window of the tank body;

[0188] The preset brightness threshold is the image brightness of the water quality detection sample image, and the water quality detection sample image is obtained by pre-shooting the tank body in a clean state.

[0189] As the running time accumulates, the welding slag and oil stains in the inner tank will continuously pollute the water quality of the tank body. After long-term use, the observation window is easily attached with dirt, which affects the observation and causes false detection and missed detection. In order to effectively solve this problem, the present invention provides a new airtightness detection method. Figure 3 Schematic diagram of the working of the window cleaning device provided by the present invention. Figure 3 As shown, the present invention adds a window cleaning device 11 similar to a vehicle windshield wiper that can clean the outer wall of the entire transparent window, and uses a dirt recognition algorithm to automatically perform water purification and window self-cleaning, thereby achieving fully automatic airtightness detection without human intervention.

[0190] Figure 4 : is a control flow diagram of the window cleaning device provided by the present invention, such as Figure 4 As shown in the figure, the algorithm flow of dirt recognition mainly includes but is not limited to:

[0191] Since the quality of water will directly affect the imaging quality of the camera, and thus affect the accuracy of bubble detection, it is very necessary to perform water quality detection. After acquiring the second video stream captured by the camera 2, this embodiment can extract at least one video frame from the second video stream as a liquid quality detection image. This liquid quality detection image is used to evaluate the cleanliness of the water in the current tank.

[0192] Image brightness is a quantitative value that reflects the overall brightness of the image. In practical applications, image processing software or algorithms can be used to calculate the brightness value of the image. The image brightness of the extracted liquid quality detection image is compared with a preset brightness threshold. The preset brightness threshold is a preset standard value that represents the image brightness when the water quality is good. This threshold can be determined by pre-shooting a water quality detection sample image of the tank in a clean state. If the image brightness is less than the preset brightness threshold, it means that the current water quality is poor and there may be dirt or impurities, which will affect the detection effect of the bubbles. Therefore, the system sends a cleaning instruction to the window cleaning device 11.

[0193] After receiving the cleaning instruction, the window cleaning device 11 will start working to clean the shooting window of the tank body. According to the description of the embodiment, this cleaning device may be similar to a car windshield wiper, which can regularly wipe the transparent window to ensure the imaging quality of the camera.

[0194] Assume that at a certain time point, a frame of liquid quality detection image is extracted from the second video stream, and the calculated image brightness is 150 units, while the preset brightness threshold is 180 units, which is obtained by shooting and calculating the tank body in a clean state. Since the current image brightness is less than the preset brightness threshold, the system determines that the water quality is poor and immediately sends a cleaning instruction to the window cleaning device. The window cleaning device is then activated to wipe the shooting window of the tank body.

[0195] As an optional embodiment, since direct judgment based on image brightness is greatly affected by ambient light, the window cleaning device 11 can also be controlled to operate based on the brightness difference.

[0196] Specifically, the brightness difference can be calculated based on the image brightness collected before and after a preset time interval (generally speaking, the image brightness will decrease). When this brightness difference is greater than a certain threshold, it means that the window needs to be cleaned. At this time, a cleaning instruction is sent to the execution device related to window cleaning.

[0197] As an optional embodiment, a water filtration system may be added to the tank body to remove impurities in the water in cooperation with the automatic water filtration system during the cleaning process of the window cleaning device 11, thereby ensuring that subsequent bubble detection is not affected by water quality.

[0198] The airtightness detection method for a closed container provided by the present invention ensures the stability of the camera imaging quality by real-time monitoring and cleaning of the shooting window of the trough body. The dynamic cleaning function reduces system failures and downtime caused by water pollution, thereby improving the accuracy of bubble detection and reducing the probability of false alarms and missed alarms.

[0199] Figure 5 Schematic diagram of the structure of the airtightness detection device for a closed container provided by the present invention, such as Figure 5 As shown, mainly including but not limited to:

[0200] The video acquisition unit 51 is mainly used to acquire at least one first video stream and a second video stream that are aligned in time and space, wherein the first video stream is obtained by shooting the sealed container to be tested placed in the tank, and the second video stream is obtained by shooting the upper space of the sealed container to be tested in the tank; the tank is loaded with liquid, and the liquid immerses the sealed container to be tested;

[0201] The video preprocessing unit 52 is mainly used to obtain a starting video frame in the first video stream, and intercept a first video frame sequence from the starting video frame, wherein the starting video frame includes a bubble image; and intercept a second video frame sequence from the second video stream according to a timestamp of the starting video frame;

[0202] The detection operation unit 53 is mainly used to compare the first bubble displacement information determined by the first video frame sequence and the second bubble displacement information determined by the second video frame sequence to determine the detection result of the airtightness detection of the sealed container to be tested.

[0203] It should be noted that the airtightness detection device for a sealed container provided by the present invention can execute the airtightness detection method for a sealed container described in any of the above embodiments during specific operation, which will not be elaborated in this embodiment.

[0204] The airtightness detection device for a closed container provided by the present invention and the proposed collaborative bubble trajectory tracking can effectively reduce false alarms and missed alarms caused by interference such as suspended particles in water and bubbles attached to the surface of the closed container, thereby improving the accuracy of airtightness detection.

[0205] The present invention also provides an airtightness detection system for a sealed container, which mainly comprises: a tank body, at least two cameras and a processor; the tank body is loaded with liquid, and the liquid immerses the sealed container to be tested.

[0206] A transparent window is provided on at least one side wall of the trough body, and the camera respectively captures a first video stream and a second video stream through the transparent window. The first video stream is obtained by capturing the sealed container to be tested placed in the trough body, and the second video stream is obtained by capturing the upper space in the trough body located at the sealed container to be tested.

[0207] The camera is in communication with the processor to synchronously transmit the first video stream and the second video stream to the processor. When the processor is running, it executes the airtightness detection method of the sealed container provided in any of the above embodiments.

[0208] The tank is generally made of stainless steel with good corrosion resistance and structural strength. The size of the tank can be designed according to the size of the closed container to be tested, ensuring that the container can be completely immersed in water and leaving enough upper space for observation. A drain valve is provided at the bottom of the tank to facilitate the discharge of liquid after testing.

[0209] A large acrylic transparent window can be installed on at least one side wall of the tank body, and the edge of the window is tightly fitted to the side wall of the tank body through a sealing strip to ensure that the liquid inside the tank body does not leak. The outflowing transparent window provides a clear viewing angle for each camera.

[0210] At least two high-definition industrial cameras are configured, for example, installed on one side of the tank or on both sides of the tank, and the camera lenses must be aimed at the closed container to be tested through the transparent window. The center line of the lens is parallel to the center line of the tank, and the lens height is adjustable to ensure that the entire surface and upper space of the closed container to be tested can be clearly photographed. The camera can be fixed to the outside of the tank by a bracket and maintain at least a certain distance from the tank. The bracket is designed to be stable to prevent the camera from shaking during shooting.

[0211] The selected camera has good imaging capabilities under low illumination and supports high-speed continuous shooting to capture subtle changes in bubbles. The camera is connected to the processor via a data cable to achieve real-time data transmission.

[0212] A high-performance industrial computer can be used as a processor, which is equipped with a large-capacity memory and a high-speed hard disk to store and analyze video data. The processor runs customized airtightness detection software to achieve synchronous reception, processing and result determination of video streams.

[0213] All cameras are connected to the video capture card of the processor via data cables to ensure that the video stream can be transmitted to the processor synchronously. The software module inside the processor is responsible for the synchronous processing of the video stream to ensure the accuracy of the detection results.

[0214] As an optional embodiment, a lighting device, such as an LED fill light, may be added to the slot to improve the shooting effect under low light conditions.

[0215] After starting the airtightness detection system, first inject an appropriate amount of water into the tank until the water level completely covers the airtight container to be tested. Then, start the camera to start shooting the video stream and transmit the video data to the processor in real time. The processor analyzes the video stream according to the preset algorithm to detect whether there are bubbles escaping from the airtight container to be tested or forming in the upper space. If bubbles are found, the airtightness of the container is judged to be unqualified; otherwise, it is judged to be qualified.

[0216] The airtightness detection system provided by the present invention can capture the dynamic changes of bubbles in all directions by synchronously shooting the sealed container to be tested and the upper space of the sealed container to be tested, thereby significantly improving the accuracy and reliability of airtightness detection. The use of such collaborative bubble trajectory tracking can effectively reduce false alarms and missed alarms caused by interference such as suspended particles in water and bubbles attached to the surface of the sealed container, thereby improving the accuracy of airtightness detection.

[0217] Figure 6 is a schematic diagram of the structure of a single transparent window provided by the present invention, as an optional embodiment, such as Figure 6 As shown, the number of the transparent window is one, and it is a rectangular transparent window 12; the size of the transparent window 12 can be determined according to the camera field of view 13, and can be specifically determined based on the following method:

[0218] Determine the width of the rectangular transparent window 12 according to a first distance between the camera lens and the transparent window 12, a second distance between the camera lens and the sealed container to be measured in the tank, and a width of the sealed container to be measured;

[0219] The length of the rectangular transparent window 12 is determined according to the first distance, the second distance and the width of the sealed container to be measured.

[0220] At present, it is widely used in the airtightness detection system of closed containers such as metal liners. During the detection process, the product under inspection is filled with a certain high-pressure gas and pressed into water, and then the leak of the product under inspection is judged by manual or machine vision. In order to facilitate observation, it is generally necessary to change the entire side wall of the tank into transparent tempered glass to facilitate camera imaging observation.

[0221] However, the design of large-area tempered glass is used for observing large-volume metal tanks such as electric water heaters. Unqualified metal tanks have the risk of explosion after pressurization, and the transparent side walls of the tank have a high risk of fatigue failure under impact loads. In addition, the tempered glass material itself has a 3‰~0.3‰ probability of self-explosion. Once it fails, the production line is at risk of stopping.

[0222] In view of this, the airtightness detection system for a closed container provided by the present invention reduces the size of the transparent window as much as possible by reasonably determining the minimum window opening size, so as to minimize the failure risk of the transparent window (including tempered glass self-explosion, fatigue failure under impact load, etc.).

[0223] Figure 7 is a schematic diagram of the imaging optical path provided by the present invention, such as Figure 7 As shown, the size corresponding to the specific range 16 of the rectangular transparent window can be reasonably determined according to the imaging principle, the imaging range of the lens and the size 15 of the observed object.

[0224] Assume that the first distance between the camera lens and the rectangular window is l (Unit: mm), the second distance between the camera lens and the sealed container to be tested in the tank is L (Unit: mm), the width of the closed container to be tested is W (Unit: mm). The goal is to determine the width of the rectangular window w and length h .

[0225] Since the camera's imaging range needs to cover the entire width of the sealed container to be tested W And the length of the entire closed container to be tested H , so the required rectangular window width can be calculated by the following formula (1) and formula (2): w and the length of the rectangular window h :

[0226]

[0227]

[0228] Furthermore, let the horizontal imaging angle of the lens be , and its vertical imaging angle is , then according to Figure 7 Select the camera based on the imaging light path shown.

[0229] The minimum horizontal imaging angle required by the camera is:

[0230]

[0231] The minimum vertical imaging angle required by the camera is:

[0232]

[0233] By combining the imaging angle with the camera's resolution, the focal length of the lens can be determined to achieve the selection of the camera lens. Generally, the shorter the camera's focal length, the larger its imaging range, but the corresponding fisheye distortion phenomenon is more serious. The role of the imaging angle is to calculate the required camera focal length parameters. For example, if the calculated value is 12.5mm, then choose a 12mm lens. If the calculated value is 5.6mm, then considering the fisheye distortion problem, directly choosing a 5mm lens may not be the best solution. You can use a coms target surface with a larger resolution for calculation, and the calculated lens parameters will fall into a range greater than 8mm.

[0234] As an optional embodiment, the present invention also provides a structural design in which the window material can be easily replaced, so as to solve the problem that the material turns yellow after aging and affects observation.

[0235] Figure 8 This is one of the schematic diagrams of the transparent window provided by the present invention. The transparent window of the present invention adopts a rectangular window, and a window is opened on the metal side wall of the trough body, such as Figure 8 As shown in the blue rectangle in the figure. The white area in the middle is the cut-through area, and the orange area retains a certain metal thickness, which is determined by the thickness of the metal side wall minus the material thickness of the transparent window. A sealing strip is placed in the orange area, and the locking structure is used to lock it after the transparent material is installed.

[0236] Fig. 9 FIG. 2 is a second schematic diagram of the shape of the transparent window provided by the present invention. The transparent window of the present invention can also adopt a circular structure, that is, by drawing the smallest circumscribed circle on the smallest rectangular window determined in the above embodiment. Fig. 9 As shown, a circular hole of the same size as the circumscribed circle is opened on the metal side wall of the tank body. A thread design is adopted on the edge of the circular hole. The transparent material can be directly installed and removed by rotation, so as to achieve the purpose of rapid replacement.

[0237] Furthermore, if the size of the sealed container to be observed is too large, the size of the single transparent window will also increase accordingly. In this case, the size of the window can be reduced by using multiple cameras to perform imaging in different areas.

[0238] Fig.10 is a schematic diagram of the distribution of two transparent windows provided by the present invention, such as Fig.10 As shown, there are two transparent windows, and the first transparent window is located below the second transparent window; the first transparent window is for the first camera to capture the first video stream; and the second transparent window is for the second camera to capture the second video stream.

[0239] The airtightness detection system in this embodiment uses two transparent windows, namely the first transparent window and the second transparent window, which are used for the first camera and the second camera to collect video streams respectively. Both transparent windows are made of acrylic material with good light transmittance and high toughness to avoid the risk of self-explosion of tempered glass.

[0240] The first transparent window is located below the second transparent window, forming a vertical layout. This layout helps to obtain all-round information of the metal liner through the upper and lower perspectives during observation, thereby improving the accuracy of detection. Among them, the first transparent window serves as the main observation window to ensure that the key areas of the metal liner being observed can be fully covered. The second transparent window is located above the first transparent window and can be slightly smaller than the first transparent window, or adjusted according to observation needs, so as to observe the floating trajectory of the bubble.

[0241] As an optional embodiment, the number of transparent windows can be multiple, and the number of cameras can also be multiple, wherein there is at least one transparent window for one of the cameras to capture the second video stream; the number of the remaining multiple transparent windows is determined by the camera pixel utilization rate determined by the aspect ratio of the camera's effective photosensitive area and the aspect ratio of the sealed container to be tested.

[0242] When a transparent window is used for a camera to achieve complete coverage observation of the metal liner to be tested, the large size of the metal liner will cause the window size of the transparent window to be too large, and even fail to achieve complete coverage of the metal liner. In this case, you can consider using multiple cameras to observe part of the metal liner separately, and then stitch the video frames taken by these cameras to achieve the purpose of completing the observation of complete coverage of the metal liner while effectively reducing the window size of the transparent window.

[0243] As another optional embodiment, when designing multiple windows according to the size of the metal liner to be measured, the corresponding relationship between the aspect ratio of the camera CMOS target surface and the aspect ratio of the metal liner can also be considered. When the aspect ratio of the metal liner deviates too much from the aspect ratio of the camera CMOS target surface, the utilization rate of the camera resolution will be too low, thereby affecting the imaging effect of the camera.

[0244] For example, the aspect ratio of the CMOS target surface of a camera is usually 4:3 or 3:2, so we can consider dividing the entire metal liner into multiple areas close to 4:3 or 3:2, and then use multiple cameras to perform imaging observations in each area, so as to maximize the utilization of the camera resolution.

[0245] Fig.11 Schematic diagram of the inner tank with a high aspect ratio provided by the present invention dividing the observation area of ​​multiple cameras, such as Fig.11As shown in the figure, if the size of the metal liner to be tested is 696mm long and 340mm in diameter, it can be divided into two areas, the length-width ratio of the two areas is close to 4:3, and then two cameras are used to shoot the areas separately. Accordingly, when multiple cameras are used to perform the observation task of the metal liner, multiple transparent windows need to be opened on the side wall of the tank.

[0246] Fig.12 is a schematic diagram of imaging of two transparent windows provided by the present invention, Fig.12 Taking the example shown, two transparent windows are opened in parallel on a side wall of the tank body, respectively denoted as a first transparent window 17 and a second transparent window 18, and a camera (also referred to herein as camera 1 and camera 2) is configured for each transparent window. Then, after the first field of view 19 of camera 1 and the second field of view 20 of camera 2 are spliced, they completely cover the entire metal liner.

[0247] Fig.13 The present invention provides a schematic diagram of providing independent bubble tracking windows on two transparent windows. In the above embodiment, it is only explained that multiple transparent windows are opened at the same time to complete the complete coverage observation of the metal liner itself. Fig.13 It is shown in the figure how to set a transparent window on this basis for other cameras to complete the shooting of the upper space of the metal liner to achieve the purpose of tracking the bubble trajectory.

[0248] Of course, the above embodiments include Fig.12 and Fig.13 It only shows that two transparent windows are set on the side wall of the trough body for coordinated shooting of the metal liner, and another transparent window is added above these two transparent windows to realize the solution of shooting the bubble trajectory. However, in the actual detection process, more transparent windows can be opened on the same side wall of the trough body according to the size of the metal liner to be tested.

[0249] In addition, based on actual conditions and while ensuring that the metal liner can be fully observed, transparent windows can be opened on different side walls to reduce the impact of too many windows on the same side wall on its strength.

[0250] Fig.14 is a schematic diagram of the structure of multiple transparent windows provided by the present invention, such as Fig.14As shown, two transparent windows are respectively arranged on two opposite side walls of the tank body, and a camera is configured for each transparent window, which are respectively called the first camera 14-1, the second camera 14-2, the third camera 14-3 and the fourth camera 14-4. The distance between the lens of each camera and the side wall is controlled at 350±30mm, so that the camera does not contact the side wall to eliminate the vibration transmission of the tank body during the water test. A protective cover is installed for each of the two cameras on each side, as shown in FIG. Fig.14 The base protective cover 14-5 and the second protective cover 14-6 are shown to protect the camera.

[0251] In this embodiment, the two cameras on the same side can complete the full coverage observation of the metal liner, but considering the three-dimensional structure of the metal liner, the purpose of setting a group of cameras on the two opposite side walls of the trough body is to observe the metal liner more comprehensively to ensure the accuracy of the detection results as much as possible.

[0252] It should be noted that, in this embodiment, a bracket can be configured for each camera, and the bracket can be drilled into the ground for fixation without directly contacting the tank body, in order to eliminate the vibration transmission of the tank body during the water testing process.

[0253] Based on the contents of the above embodiments, as an optional embodiment, the present invention further adds reinforcing ribs on the outside of the transparent window; the arrangement of the reinforcing ribs is determined based on the maximum deflection of the center point of the transparent window.

[0254] Based on the above embodiments, the present invention can significantly reduce the deflection of the transparent material under the action of hydrostatic pressure and shock waves and reduce the risk of failure and damage by adding reinforcing ribs.

[0255] As an optional embodiment, the method of adding reinforcing ribs can also be based on the idea of ​​lean design. On the basis of the existing large window, only by selecting the solution of reinforcing ribs, the deflection of the transparent material under the action of hydrostatic pressure and shock waves can be reduced to a certain extent, thereby reducing the risk of failure and damage.

[0256] Fig.15 It is one of the schematic diagrams of the reinforcement rib arrangement method provided by the present invention, such as Fig.15 As shown, it is the arrangement mode of the reinforcing ribs when a transparent window 12 is opened on the side wall of the tank body 1 for water inspection of the sealed container 10 to be tested. In this embodiment, a first reinforcing rib 15-1, a second reinforcing rib 15-2, a third reinforcing rib 15-3 and a fourth reinforcing rib 15-4 are provided, and the above four reinforcing ribs are fixed on the outer surface of the transparent window in a tic-tac-toe shape.

[0257] Fig.16 This is a second schematic diagram of the reinforcement rib arrangement method provided by the present invention, such as Fig.16 As shown, it is the arrangement mode of the reinforcing ribs when two transparent windows 12 are opened on the side wall of the tank body 1 for water inspection of the sealed container 10 to be tested.

[0258] In this embodiment, a fifth reinforcing rib 16-1, a sixth reinforcing rib 16-2, and a seventh reinforcing rib 16-3 are used, wherein the fifth reinforcing rib 16-1 and the sixth reinforcing rib 16-2 are arranged at equal intervals in the horizontal direction in the height direction of the transparent window 12, and the seventh reinforcing rib 16-3 is arranged vertically along the dividing portion between the two transparent windows 12 and welded to the contact portion of the fifth reinforcing rib 16-1 and the sixth reinforcing rib 16-2.

[0259] Of course, the premise of the solution of adding reinforcement ribs is that the added reinforcement ribs cannot block the camera's field of view. At the same time, the safety requirements of the transparent window are met in the least and most economical way.

[0260] As an optional embodiment, the present invention provides a method for determining the number of reinforcement ribs to be arranged according to the maximum deflection of the center point of the transparent window, and the calculation formula is:

[0261] ;

[0262] in, is the maximum deflection of the center point of the transparent window, indicating the maximum deformation degree of the transparent window under the maximum load; n Indicates the number of reinforcement bars installed; P is the concentrated load applied on the transparent window; is the section inertia moment, which represents the resistance of the material section of the transparent window to bending; is the effective elastic modulus, which represents the stiffness of the material of the transparent window.

[0263] Based on the contents of the above embodiments, as an optional embodiment, the airtightness detection system for a closed container provided by the present invention may further include:

[0264] Determine a shooting interval for shooting the upper space of the sealed container to be tested;

[0265] A background is set for the shooting interval, and there is a visual contrast between the background and the bubbles. For example, a dark matte background design is added within the field of view of the camera that shoots the upper space of the sealed container to be tested. Against the dark matte background, the bubbles glow due to the strong refraction of light, forming a strong contrast with the background, thereby strengthening the characteristics of the bubbles. The dark matte background can be fixed to the area near the independent observation window of the water inspection tank body by means of waterproof stickers or spraying.

[0266] As an optional embodiment, a dark matte background plate may be provided in the shooting area, or the side wall of the groove body corresponding to the shooting area may be painted.

[0267] It should be noted that the background plate can be arranged on the outer side of the side wall of the tank body, and when the side wall of the tank body is painted, the outer side of the side wall of the tank body can also be painted.

[0268] Optionally, since the observation of bubbles will be affected to a certain extent by the surrounding environment, such as the lighting conditions of the environment, other surrounding objects (even the body color of the sealed container to be tested), etc., will affect the visual difference between the bubbles and the background plate to a certain extent. Therefore, multiple background plates of different colors, materials and sizes can be prepared, and before the actual detection, bubbles can be artificially created, and the contrast, brightness and other parameters between the background plate and the bubbles can be accurately calculated to design the background that best suits the current detection environment. For example, image processing software can be used to simulate the bubble recognition effect under different background colors, and then the optimal solution can be selected, and the background plate with the largest visual contrast can be selected for setting.

[0269] The present invention adopts this design, and the optimized background design can significantly reduce false positives and false negatives, and improve the accuracy and stability of bubble recognition. And because the contrast between the background and the bubbles is enhanced, the dirt detection algorithm can also benefit from it. Dirt usually appears as spots or stripes with different colors from the background. The optimized background makes these anomalies easier to be captured by the algorithm.

[0270] Fig.17 is a schematic diagram of the structure of the electronic device provided by the present invention, such as Fig.17As shown, the electronic device may include: a processor (processor) 710 , a communication interface (Communications Interface) 720 , a memory (memory) 730 and a communication bus 740 , wherein the processor 710 , the communication interface 720 , and the memory 730 communicate with each other through the communication bus 740 . The processor 710 can call the logic instructions in the memory 730 to execute the airtightness detection method of the sealed container, which method includes: obtaining at least one first video stream and a second video stream that are aligned in time and space, the first video stream is obtained by shooting the sealed container to be tested placed in a trough, and the second video stream is obtained by shooting the upper space of the sealed container to be tested in the trough; the trough is loaded with liquid, and the liquid immerses the sealed container to be tested; obtaining the starting video frame in the first video stream, and intercepting a first video frame sequence starting from the starting video frame, and the starting video frame contains a bubble image; intercepting a second video frame sequence from the second video stream according to the timestamp of the starting video frame; comparing the first bubble displacement information determined by the first video frame sequence and the second bubble displacement information determined by the second video frame sequence to determine the detection result of the airtightness detection of the sealed container to be tested.

[0271] In addition, the logic instructions in the above-mentioned memory 730 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0272] On the other hand, the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the airtightness detection method for a closed container provided in the above embodiments, the method comprising: obtaining at least one first video stream and a second video stream that are aligned in time and space, the first video stream is obtained by shooting the closed container to be tested placed in a trough body, and the second video stream is obtained by shooting the upper space of the closed container to be tested in the trough body; the trough body is loaded with liquid, and the liquid immerses the closed container to be tested; obtaining a starting video frame in the first video stream, and intercepting a first video frame sequence starting from the starting video frame, the starting video frame containing a bubble image; intercepting a second video frame sequence from the second video stream according to the timestamp of the starting video frame; comparing the first bubble displacement information determined by the first video frame sequence and the second bubble displacement information determined by the second video frame sequence, and determining the detection result of the airtightness detection of the closed container to be tested.

[0273] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented by a processor to execute the airtightness detection method for a sealed container provided in the above-mentioned embodiments, the method comprising: obtaining at least one first video stream and a second video stream that are aligned in time and space, the first video stream being obtained by shooting the sealed container to be tested placed in a trough, and the second video stream being obtained by shooting the upper space of the sealed container to be tested in the trough; the trough is loaded with a liquid, and the liquid immerses the sealed container to be tested; obtaining a starting video frame in the first video stream, and intercepting a first video frame sequence starting from the starting video frame, the starting video frame containing a bubble image; intercepting a second video frame sequence from the second video stream according to the timestamp of the starting video frame; comparing the first bubble displacement information determined by the first video frame sequence with the second bubble displacement information determined by the second video frame sequence, and determining the detection result of the airtightness detection of the sealed container to be tested.

[0274] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0275] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0276] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting the airtightness of a closed container, characterized in that: include: Acquire at least one first video stream and a second video stream that are aligned in time and space, wherein the first video stream is obtained by shooting a sealed container to be tested placed in a tank, and the second video stream is obtained by shooting an upper space in the tank located at the sealed container to be tested; the tank is loaded with a liquid, and the liquid immerses the sealed container to be tested; Acquire a starting video frame in the first video stream, and capture a first video frame sequence starting from the starting video frame, wherein the starting video frame includes a bubble image; intercepting a second video frame sequence from the second video stream according to the timestamp of the starting video frame; Comparing the first bubble displacement information determined by the first video frame sequence and the second bubble displacement information determined by the second video frame sequence to determine the test result of the airtightness test on the sealed container to be tested specifically includes: Determine at least one first running trajectory according to the first bubble displacement information, and determine at least one second running trajectory according to the second bubble displacement information; If it is determined that at least one of the first running trajectories and any of the second running trajectories are running trajectories of the same dynamic bubble, then determining that the detection result is unqualified; Otherwise, determining that the test result is qualified; The determining of the second bubble displacement information by the second video frame sequence specifically includes: Performing bubble image detection on each video frame in the second video frame sequence, screening out all target video frames containing bubble images, and marking an area of ​​interest related to the bubble image in each target video frame; For any region of interest, the region of interest is divided into a plurality of image grids, and displacement information between corresponding image grids in two adjacent target video frames is traversed and calculated to determine a velocity field of the region of interest; The velocity fields of all the regions of interest are acquired to determine the second bubble displacement information; the second bubble displacement information is used to indicate the running trajectory of the dynamic bubble.

2. The method for detecting airtightness of a sealed container according to claim 1, characterized in that: Determining the first bubble displacement information from the first video frame sequence specifically includes: Inputting the first video frame sequence into a video classification model to obtain an output result of the video classification model; The output result includes whether a dynamic bubble is identified and the first bubble displacement information of the dynamic bubble; the first bubble displacement information is used to indicate the running track of the dynamic bubble; The video classification model is obtained after training with 3D CNN as the initial model.

3. The method for detecting airtightness of a sealed container according to claim 1, characterized in that: The obtaining of a starting video frame in the first video stream comprises: Extracting a video frame to be tested from the first video stream according to a preset sampling frequency; Inputting the video frames to be tested into the target recognition model in sequence, and obtaining the detection results output by the target recognition model; When it is determined that any detection result is an image with bubbles, obtaining a video frame to be tested corresponding to any detection result as the starting video frame; The target recognition model is obtained after training using multiple video frame samples and a detection result label corresponding to each video frame sample.

4. The method for detecting airtightness of a sealed container according to claim 1, characterized in that: Before intercepting a second video frame sequence from the second video stream according to the timestamp of the starting video frame, the method further includes: The second video stream is subjected to image grayscale processing.

5. The method for detecting airtightness of a sealed container according to claim 4, characterized in that: Before dividing the region of interest into a plurality of image grids, normalizing the region of interest specifically includes: Calculate the gray value of each pixel in the region of interest to calculate the average gray intensity; Calculating the sum of squares of differences between the grayscale value of each pixel and the average grayscale intensity, and determining a normalization factor according to the sum of squares of differences and the total number of pixels; The normalized pixel value of each pixel is determined according to the relationship between the difference between the grayscale value of each pixel and the average grayscale intensity and the normalization factor.

6. The method for detecting airtightness of a sealed container according to any one of claims 1 to 5, characterized in that: After acquiring the second video stream, the method further includes: Extracting at least one video frame in the second video stream as a liquid quality detection image; Acquiring the image brightness of the liquid quality detection image; If it is determined that the image brightness is less than a preset brightness threshold, sending a cleaning instruction to the window cleaning device; The cleaning instruction is used to drive the window cleaning device to clean the shooting window of the tank body; The preset brightness threshold is the image brightness of the water quality detection sample image, and the water quality detection sample image is obtained by pre-shooting the tank body in a clean state.

7. An airtightness detection device for a sealed container, characterized in that: include: A video acquisition unit, used to acquire at least one first video stream and a second video stream that are aligned in time and space, wherein the first video stream is obtained by shooting a sealed container to be tested placed in a slot, and the second video stream is obtained by shooting an upper space in the slot located at the sealed container to be tested; the slot is loaded with a liquid, and the liquid immerses the sealed container to be tested; a video preprocessing unit, configured to obtain a starting video frame in the first video stream, and intercept a first video frame sequence starting from the starting video frame, wherein the starting video frame includes a bubble image; and intercept a second video frame sequence from the second video stream according to a timestamp of the starting video frame; The detection operation unit is used to compare the first bubble displacement information determined by the first video frame sequence and the second bubble displacement information determined by the second video frame sequence to determine the detection result of the airtightness detection of the sealed container to be tested, specifically including: Determine at least one first running trajectory according to the first bubble displacement information, and determine at least one second running trajectory according to the second bubble displacement information; If it is determined that at least one of the first running trajectories and any of the second running trajectories are running trajectories of the same dynamic bubble, then determining that the detection result is unqualified; Otherwise, determining that the test result is qualified; The determining of the second bubble displacement information by the second video frame sequence specifically includes: Performing bubble image detection on each video frame in the second video frame sequence, screening out all target video frames containing bubble images, and marking an area of ​​interest related to the bubble image in each target video frame; For any region of interest, the region of interest is divided into a plurality of image grids, and displacement information between corresponding image grids in two adjacent target video frames is traversed and calculated to determine a velocity field of the region of interest; The velocity fields of all the regions of interest are acquired to determine the second bubble displacement information; the second bubble displacement information is used to indicate the running trajectory of the dynamic bubble.

8. An airtightness detection system for a sealed container, characterized in that: include: A tank body, at least two cameras and a processor; the tank body is loaded with liquid, and the liquid immerses the sealed container to be tested; A transparent window is provided on at least one side wall of the trough body, and the camera respectively shoots a first video stream and a second video stream through the transparent window, wherein the first video stream is obtained by shooting the sealed container to be tested placed in the trough body, and the second video stream is obtained by shooting the upper space of the sealed container to be tested in the trough body; The camera is communicatively connected to the processor to synchronously transmit the first video stream and the second video stream to the processor; When the processor is running, it executes the airtightness detection method for the sealed container as described in any one of claims 1 to 6.

9. The airtightness detection system for a sealed container according to claim 8, characterized in that: The number of the transparent window is one, and the transparent window is a rectangular window; the size of the rectangular window is determined based on the following method: Determine the width of the rectangular window according to a first distance between the camera lens and the rectangular window, a second distance between the camera lens and the sealed container to be measured in the tank, and a width of the sealed container to be measured; The length of the rectangular window is determined according to the first distance, the second distance and the width of the sealed container to be measured.

10. The airtightness detection system for a sealed container according to claim 8, characterized in that: The number of the transparent windows is two, and the first transparent window is located below the second transparent window; The first transparent window is used by the first camera to capture the first video stream; The second transparent window is used by the second camera to capture the second video stream.

11. The airtightness detection system for a sealed container according to claim 8, characterized in that: There are multiple transparent windows and multiple cameras, and at least one transparent window is provided for one of the cameras to capture the second video stream. The number of the remaining multiple transparent windows is determined by the camera pixel utilization ratio determined by the aspect ratio of the effective photosensitive area of ​​the camera and the aspect ratio of the sealed container to be measured.

12. The airtightness detection system for a sealed container according to any one of claims 8 to 11, characterized in that: A reinforcing rib is added to the outside of the transparent window; The arrangement of the reinforcing ribs is determined based on the maximum deflection of the center point of the transparent window.

13. The airtightness detection system for a closed container according to any one of claims 8 to 11, characterized in that: Also includes: Determine a shooting interval for shooting the upper space of the sealed container to be tested; A background of the shooting interval is set, and there is a visual contrast between the background and the bubbles.

14. The airtightness detection system for a closed container according to any one of claims 8 to 11, characterized in that: Also included is a window cleaning device; When the window cleaning device is in operation, it is used to clean the transparent window.

15. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the airtightness detection method of the sealed container as described in any one of claims 1 to 6 is implemented.

16. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the airtightness detection method of the sealed container as claimed in any one of claims 1 to 6 is implemented.

17. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the airtightness detection method of the sealed container as claimed in any one of claims 1 to 6 is implemented.

Images