Container airtightness detection method and device, electronic equipment and storage medium

Abstract

The invention provides a container airtightness detection method and device, electronic equipment and a storage medium, and belongs to the field of intelligent detection, and the method comprises the steps: determining a contour region of a target container in a to-be-detected product image; extracting a first sound source signal in the contour area; determining a welding seam position area in the contour area; extracting a second sound source signal in the welding seam position area from the first sound source signal; and based on the second sound source signal, positioning a leakage point possibly existing on the target container. According to the container airtightness detection method and device, the electronic equipment and the storage medium provided by the invention, visual information and acoustic information are creatively and deeply fused, and a detected sound source signal is strictly limited in a welding seam position area where leakage is most likely to occur through two-stage progressive spatial filtering, so that the detection accuracy is greatly improved. Therefore, interference of environmental noise, signal reflection and non-critical area signals is greatly eliminated, and high-precision and high-reliability positioning of the leakage point of the container is realized.

Description

Technical Field

[0001] The present invention relates to the field of intelligent detection, and in particular to a container air tightness detection method, device, electronic equipment and storage medium. Background Art

[0002] During the production of household appliances, pressurized containers like water heater tanks and compressors have very strict airtightness requirements, requiring them to be tested before they leave the production line. For applications with more stringent airtightness testing requirements, a more accurate solution is to test for helium or hydrogen and nitrogen leaks. For applications with less stringent airtightness testing requirements, microphone array testing technology offers a promising alternative to traditional water immersion bubble detection. Microphone array airtightness testing works by locating the ultrasonic signal generated by a gas leak to locate the leak source.

[0003] The current microphone array airtightness detection technology has the following problems: 1) The sound source signal generated by gas leakage will reflect between objects to generate false sound source signals, causing the microphone array to falsely alarm; 2) The rapid sealing and inflation mechanism and pipelines used in container detection will leak during use, causing interference with the microphone array's detection of real container leaks; 3) The current microphone array leak detection and positioning technology can only locate leaks within the dynamic range of the largest leak source, and cannot fully identify multiple leak sources with greatly varying leak degrees, resulting in leakage problems that still exist after subsequent repairs.

[0004] In view of this, it is necessary to provide a new microphone array airtightness detection method to overcome the above-mentioned shortcomings. Summary of the Invention

[0005] The present invention provides a container air tightness detection method, device, electronic equipment and storage medium, which are used to prevent the occurrence of false detection and missed detection in the current microphone array air tightness detection.

[0006] The present invention provides a method for detecting the air tightness of a container, comprising the following steps: Determine the contour area of ​​the target container in the image of the product to be tested; Extracting a first sound source signal within the contour area; determining a weld position region within the contour region; Extracting a second sound source signal in the weld position area from the first sound source signal; Based on the second sound source signal, a possible leakage point on the target container is located.

[0007] According to a container air tightness detection method provided by the present invention, locating a possible leakage point on the target container based on the second sound source signal includes: Obtaining an acoustic spectrum corresponding to the second sound source signal; Determining the point with the highest signal intensity on the acoustic spectrum; When it is determined that the highest signal intensity point is valid, determining a leakage point of the target container according to the position of the highest signal intensity point in the acoustic spectrum; Shielding the influence range of the point with the highest signal intensity in the acoustic spectrum; Iteratively executing the step of determining the highest signal strength point on the acoustic spectrum, shielding the influence range of the highest signal strength point in the acoustic spectrum, and determining that the highest signal strength point is invalid; Obtaining a set of leakage points of the target container determined during each iteration; The influence range of the highest signal strength point refers to all connected areas centered on the highest signal strength point where the signal strength is greater than a preset signal strength.

[0008] According to a container air tightness detection method provided by the present invention, whether the highest signal intensity point is valid is determined by judging whether the signal-to-noise ratio of the highest signal intensity point is greater than a preset signal-to-noise ratio threshold, and judging whether the influence range of the highest signal intensity point is greater than a preset minimum area threshold.

[0009] According to a container air tightness detection method provided by the present invention, the signal-to-noise ratio at the highest signal intensity point is determined based on the following steps: Determine the signal strength at the highest signal strength point; Determining a noise assessment area, where the noise assessment area is other areas in the acoustic spectrum after shielding the influence range of the point with the highest signal intensity; determining a background noise intensity associated with the noise assessment area; The signal-to-noise ratio of the highest signal strength point is determined based on the signal strength of the highest signal strength point and the background noise strength.

[0010] According to a container air tightness detection method provided by the present invention, determining the contour area of ​​the target container in the image of the product to be tested includes: Using a first-order differential operator to convolve the image of the product to be tested in the horizontal and vertical directions to obtain the gradient amplitude and gradient direction of each pixel point on the image of the product to be tested; Pixels whose gradient amplitude is greater than a preset upper amplitude threshold are defined as strong edge points, and pixels whose gradient amplitude is less than or equal to the preset upper amplitude threshold but greater than a preset lower amplitude threshold are defined as weak edge points; Based on the gradient direction of each of the weak edge points, determining a portion of the weak edge points connected to any of the strong edge points among all the weak edge points; sequentially connecting the strong edge points and some of the weak edge points to construct a closed contour of the target container; Based on the closed contour, a contour area of ​​the target container is determined.

[0011] According to a container air tightness detection method provided by the present invention, the step of determining the contour area of ​​the target container in the image of the product to be detected further includes: Determining a gradient magnitude map of the image of the product to be tested; performing binarization processing on the gradient amplitude map to obtain a binarized gradient amplitude map; determining a closed contour of the target container from the binary gradient amplitude map; Based on the closed contour, a contour area of ​​the target container is determined.

[0012] According to a container air tightness detection method provided by the present invention, determining the contour area of ​​the target container based on the closed contour includes: Determining a target closed contour based on the contour area, contour level, and contour position of each of the closed contours; The area where the target closed contour is located is used as the contour area of ​​the target container.

[0013] According to a container air tightness detection method provided by the present invention, the step of determining the contour area of ​​the target container in the image of the product to be detected further includes: Inputting the image of the product to be tested into a contour annotation model, and obtaining a contour area probability map output by the contour annotation model; Performing differential assignment on pixel points in the contour area probability map that are greater than a preset probability threshold and less than or equal to the preset probability threshold to obtain a binary mask map; determining a contour area of ​​the target container based on the binary mask image; The contour annotation model is obtained by training based on product image samples, and each of the product image samples is pre-annotated with a contour region probability map label.

[0014] According to a container air tightness detection method provided by the present invention, extracting the first sound source signal within the contour area includes: Determine a coordinate system transformation matrix between a microphone sound source coordinate system and a camera image coordinate system; the microphone sound source coordinate system is the spatial coordinate system of a microphone array used for container air tightness testing, and the camera image coordinate system is the coordinate system of a camera used to capture images of the product to be tested; Based on the acoustic signals collected by the microphone array, a full-space acoustic spectrum covering the airtightness detection space of the container is generated; Traversing each potential sound source position point in the full-space acoustic map; Using the coordinate system transformation matrix, projecting the microphone sound source coordinates of each potential sound source position point in the microphone sound source coordinate system to the camera image coordinate system to obtain corresponding image pixel coordinates; Determining whether the image pixel coordinates corresponding to each potential sound source position point are located within the contour area of ​​the target container; If the image pixel coordinates are within the contour area, retaining the area corresponding to the potential sound source position point in the full-space acoustic map; The first sound source signal is generated based on all retained areas in the full-space acoustic spectrum.

[0015] According to a container airtightness detection method provided by the present invention, determining the coordinate system transformation matrix between the microphone sound source coordinate system and the camera image coordinate system includes: Selecting a calibration object, wherein the calibration object is provided with a visual calibration pattern and an acoustic beacon, and the physical position relationship of the acoustic beacon relative to the visual calibration pattern is predetermined; photographing the calibration object with the camera and identifying the visual calibration pattern, and determining the position and pose of the visual calibration pattern in the camera image coordinate system; collecting the acoustic signal emitted by the acoustic beacon through the microphone array, and locating the coordinates of the acoustic beacon in the microphone sound source coordinate system; The coordinate system transformation matrix is ​​calculated based on the pose of the visual calibration pattern in the camera image coordinate system, the coordinates of the acoustic beacon in the microphone sound source coordinate system, and the physical position relationship.

[0016] According to a container air tightness detection method provided by the present invention, the step of determining the weld position area within the contour area includes: Obtaining a process design model of the target container; Based on the process design model, a three-dimensional coordinate point sequence of all welds of the target container is obtained; Projecting the three-dimensional coordinate point sequence onto the image of the product to be tested based on a pose transformation matrix between the model coordinate system of the process design model and the camera image coordinate system, and obtaining weld line segments of all welds of the target container on the image of the product to be tested; generating a weld mask on the image of the product to be tested based on the weld line segment; An intersection of the weld mask and the contour area is determined as the weld position area.

[0017] According to a container air tightness detection method provided by the present invention, extracting a second sound source signal in the weld position area from the first sound source signal includes: Traversing each potential sound source position point in the acoustic spectrum corresponding to the first sound source signal; Using the coordinate system transformation matrix, projecting the microphone sound source coordinates of each potential sound source position point in the microphone sound source coordinate system to the camera image coordinate system to obtain corresponding image pixel coordinates; If the image pixel coordinates are located within the weld location area, retaining the area corresponding to the potential sound source location point in the acoustic spectrum corresponding to the first sound source signal; The second sound source signal is generated based on all retained regions in the acoustic spectrum corresponding to the first sound source signal.

[0018] According to a container air tightness detection method provided by the present invention, the target container is an inner tank of an electric water heater.

[0019] According to a container air tightness detection method provided by the present invention, a solvent with rust-proof, strong tension and volatile properties is sprayed on the surface of the target container.

[0020] The present invention also provides a container airtightness detection device, comprising: A container positioning unit, used to determine the contour area of ​​the target container in the image of the product to be tested; a sound source signal primary screening unit, configured to extract a first sound source signal within the contour area; a weld seam locating unit, configured to determine a weld seam location region within the contour region; a sound source signal re-screening unit, configured to extract a second sound source signal in the weld position area from the first sound source signal; A leakage point locating unit is used to locate a possible leakage point on the target container based on the second sound source signal.

[0021] 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, any of the above-described methods for detecting air tightness of a container is implemented.

[0022] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for detecting air tightness of a container as described above is implemented.

[0023] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the above-mentioned methods for detecting air tightness of a container.

[0024] The container airtightness detection method, device, electronic device and storage medium provided by the present invention creatively and deeply integrate visual information with acoustic information. Through two-stage progressive spatial filtering, the detected sound source signal is strictly limited to the weld location area where leakage is most likely to occur, thereby greatly eliminating interference from environmental noise, signal reflections and signals in non-critical areas, and achieving high-precision and high-reliability positioning of container leakage points. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0026] Figure 1 It is a flow chart of the container air tightness detection method provided by the present invention.

[0027] Figure 2 This is one of the flow diagrams for locating possible leakage points on a target container provided by the present invention.

[0028] Figure 3 This is the second flow diagram of locating possible leakage points on a target container provided by the present invention.

[0029] Figure 4 This is one of the flow charts of determining the contour area of ​​a target container in an image of a product to be tested provided by the present invention.

[0030] Figure 5 This is the second flow chart of determining the contour area of ​​the target container in the image of the product to be tested provided by the present invention.

[0031] Figure 6 This is the third flow chart of determining the contour area of ​​a target container in an image of a product to be tested provided by the present invention.

[0032] Figure 7 It is a schematic diagram of the process of extracting the first sound source signal within the target container contour area provided by the present invention.

[0033] Figure 8 It is a schematic diagram of a process for determining a coordinate system transformation matrix between a microphone sound source coordinate system and a camera image coordinate system provided by the present invention.

[0034] Figure 9 It is a schematic diagram of the process of determining the weld position area within the contour area provided by the present invention.

[0035] Figure 10 It is a schematic diagram of a process for extracting a second sound source signal in a weld position area from a first sound source signal provided by the present invention.

[0036] Figure 11 It is a structural schematic diagram of the container air tightness detection device provided by the present invention.

[0037] Figure 12 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0038] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0039] It should be noted that, in the description of the present invention, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0040] The terms "first," "second," and the like in the present invention are used to distinguish similar objects, and are not used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, such that embodiments of the present invention can be implemented in orders other than those illustrated or described herein. Furthermore, the objects distinguished by "first," "second," and the like generally refer to a class of objects, and do not limit the number of objects. For example, the first object may be one or more.

[0041] The following combination Figures 1-12 The present invention describes a method, device, electronic device and storage medium for detecting air tightness of a container.

[0042] Figure 1 FIG. 1 is a flow chart of the container air tightness detection method provided by the present invention, as shown in FIG. Figure 1 As shown, including but not limited to the following steps: Step 11: Determine the contour area of ​​the target container in the image of the product to be tested.

[0043] An image of the product under test is a digital image captured by an image acquisition device (such as an industrial CCD or CMOS camera) while the product is at the inspection station on the production line. Specifically, an industrial camera (for example, a CMOS camera with a resolution of 1920x1080 pixels) can be used to capture the surface of the product under test at a fixed station on the production line.

[0044] To ensure image quality, the image acquisition device can be equipped with a ring-shaped LED light source to provide a uniform, shadow-free lighting environment to enhance the contrast between the container outline and the background.

[0045] The collected images of the product to be tested are usually RGB three-channel color images. In order to reduce the computational complexity and facilitate subsequent processing, the collected color images of the product to be tested can be converted into single-channel grayscale images. The conversion formula used is: Gray=0.299 R+0.587 G+0.114 B. G, R, and B refer to the brightness values ​​of the red, green, and blue color channels of any pixel in the color image of the product to be tested, respectively. Gray refers to the single-channel grayscale value of any pixel in the grayscale image obtained after weighted summation.

[0046] Considering that dust and subtle changes in lighting in the production environment of the product being tested may introduce noise into the product image, the grayscaled product image can be smoothed before contour detection. This embodiment uses a Gaussian filter, for example, convolution of the image with a 5x5 or 7x7 Gaussian kernel, to effectively filter out Gaussian noise while preserving edge information as much as possible.

[0047] The target container is the object of the airtightness test, which is usually a pressure container with airtightness requirements. In the preferred scenario of the present invention, the target container can be an electric water heater tank, an air conditioner compressor housing, etc. in household appliances.

[0048] The contour area refers to a two-dimensional plane area in the image of the product to be tested that can completely surround the target container body and is separated from the background and surrounding environment (such as conveyor belts, clamps, etc.). It can be defined by one or more closed sets of pixel coordinates.

[0049] In step 11, machine vision technology can be used to separate the inspection subject (target container) in the image of the product to be tested from the irrelevant background, thereby determining the contour area of ​​the target container as the key inspection area. The implementation of this step can be achieved based on a variety of image processing technologies.

[0050] For example, by analyzing image features such as color, brightness, and texture, edge detection algorithms, threshold segmentation algorithms, region growing algorithms, or deep learning-based image segmentation models can be used to identify the boundaries of the target container and then divide the contour area of ​​the target container. The output form of the contour area can be a mask or coordinate set that defines the range occupied by the target container in the two-dimensional image space.

[0051] Step 12: extract the first sound source signal within the contour area.

[0052] The first sound source signal refers to the acoustic signal collected by an acoustic acquisition device (such as a microphone array) after undergoing preliminary spatial filtering. This preliminary spatial filtering retains only those signals whose sound source locations physically correspond to the contour area of ​​the target container as determined by the visual system. Specifically, the sound source signal is obtained by filtering out interference clearly originating from outside the target container (such as leakage from the inflation pipe and wall reflections) from the full-space acoustic signal collected by the acoustic acquisition device.

[0053] The purpose of step 12 is to achieve the initial guidance and screening of acoustic information by visual information. Specifically, a spatial coordinate mapping relationship can be pre-established between an acoustic imaging system (e.g., a microphone array) and a visual imaging system (e.g., a camera). After obtaining this spatial coordinate mapping relationship, any spatial sound source located by the acoustic system can be accurately projected onto the image captured by the camera. Therefore, the present invention can screen the full-space acoustic signal by determining whether the projected position of the sound source falls within the contour area of ​​the determined target container. All acoustic signals whose projection positions fall within this contour area are retained and collectively constitute the first sound source signal.

[0054] Step 13: Determine the weld position area within the contour area.

[0055] Welds are the joints between different components during the manufacturing process of a target vessel. Based on process experience, they are the most vulnerable links where leaks are most likely to occur. The weld location area is a two-dimensional planar region within the target vessel's contour area, corresponding to the locations of all welds or specific critical welds on the target vessel.

[0056] As an optional embodiment, after acquiring the target container's contour area and performing a first spatial filtering to obtain the first sound source signal, the present invention further narrows the detection range by focusing on the most critical areas of the target container for airtightness testing. The weld location area can be determined by identifying the specific model of the target container and then retrieving the process design model (e.g., CAD drawing) of that model from a pre-established process database. Because the process design model accurately records the geometric position information of all welds, the weld location area can be precisely calibrated on the contour area by registering and aligning the weld geometric position information in the process design model with a real-time image of the product to be tested.

[0057] Step 14: extract a second sound source signal in the weld position area from the first sound source signal.

[0058] The second sound source signal is the final acoustic signal used for positioning, obtained after a more refined second spatial filtering based on the first sound source signal. The second spatial filtering means that only those sound source signals whose sound source positions correspond to the weld location area in physical space are retained.

[0059] Specifically, the present invention determines whether the projection position of each sound source component in the first sound source signal falls within the weld location area determined in the previous step. Only sound source components falling within the weld location area are retained, forming the second sound source signal. This second sound source signal thus largely eliminates interference from non-weld areas (such as friction noise from the shell connection gap), resulting in an extremely high signal-to-noise ratio.

[0060] Step 15: Locate possible leakage points on the target container based on the second sound source signal.

[0061] The possible leakage points on the target container refer to the final airtightness test results, that is, the coordinates of one or more locations where gas leakage occurs that are identified in the weld position area of ​​the target container.

[0062] When detecting leakage points, the present invention analyzes the second sound source signal obtained after completing the second spatial filtering. For example, by analyzing the acoustic spectrum formed by the second sound source signal, the areas with the most concentrated energy or the highest sound pressure level can be found. The center positions of these areas are determined as possible leakage points.

[0063] The container airtightness detection method provided by this invention first performs a first spatial filtering of the full-space acoustic signal using the contour area of ​​the target container, effectively filtering out noise interference from the external environment and the detection equipment itself. A second spatial filtering is then performed using the weld location area of ​​the target container, precisely focusing the detection on the most critical high-risk leakage area. This dual, progressive spatial filtering method can greatly improve the signal-to-noise ratio of the leakage signal, fundamentally solving the problem of false alarms and missed alarms caused by signal reflections and environmental interference in traditional microphone array detection technology, significantly improving the accuracy and reliability of container airtightness detection.

[0064] Figure 2 This is one of the flow diagrams of locating possible leakage points on a target container provided by the present invention, such as Figure 2 As shown, the present invention provides a method for locating a possible leakage point on the target container based on the second sound source signal, which mainly includes but is not limited to the following steps: Obtaining an acoustic spectrum corresponding to the second sound source signal; Determining the point with the highest signal intensity on the acoustic spectrum; When it is determined that the highest signal intensity point is valid, determining a leakage point of the target container according to the position of the highest signal intensity point in the acoustic spectrum; Shielding the influence range of the point with the highest signal intensity in the acoustic spectrum; Iteratively executing the step of determining the highest signal strength point on the acoustic spectrum, shielding the influence range of the highest signal strength point in the acoustic spectrum, and determining that the highest signal strength point is invalid; Obtaining a set of leakage points of the target container determined during each iteration; The influence range of the highest signal strength point refers to all connected areas centered on the highest signal strength point where the signal strength is greater than a preset signal strength.

[0065] An acoustic atlas, also known as an acoustic camera image or sound pressure level map (SPL map), is a two-dimensional or three-dimensional image that visualizes sound source information. Each pixel or voxel on the acoustic atlas represents a specific location in space, and its brightness, color, or value indicates the sound pressure level (SPL) or sound energy at that location. Acoustic atlases are generated by processing sound source signals using acoustic localization algorithms (such as beamforming).

[0066] The acoustic spectrum corresponding to the second sound source signal in the present invention is calculated by using an acoustic localization algorithm on the second sound source signal obtained after two spatial filterings to generate an acoustic spectrum that only reflects the distribution of the sound source in the weld position area.

[0067] Among all the pixels (or voxels) in the acquired acoustic spectrum, the pixel with the maximum sound pressure level or sound energy value physically corresponds to the most significant sound source location in the current sound field. This pixel is referred to as the peak signal intensity point in this paper. Therefore, the method for determining the peak signal intensity point on the acoustic spectrum can be a numerical optimization process. By traversing the numerical matrix of the acoustic spectrum, the coordinates corresponding to the vertex are found. These coordinates are the most likely leak point.

[0068] After finding the highest signal strength point, it must be verified against the actual situation to determine whether it is a leak signal or an accidental noise spike. Validity can be determined based on various criteria, such as the signal-to-noise ratio and the area covered by the signal. Only if it is deemed valid will the location of the highest signal point appear in the acoustic spectrum and, through coordinate transformation, be officially recorded as a leak point on the target container.

[0069] In traditional acoustic detection, the sound field generated by a leak point with greater intensity will often mask the sound field signals of the surrounding leak points with weaker intensity, resulting in only one bright spot being observed on the acoustic spectrum, thus causing missed detection. In order to solve the defect that traditional microphone arrays can only locate the leak point with the highest sound pressure level, while ignoring other leak points with smaller leakage that may exist at the same time, the present invention will shield the influence range of the highest signal intensity point that has been determined in the acoustic spectrum. This influence range is specifically defined as: a connected area with all signal intensity values ​​on the acoustic spectrum greater than a preset signal intensity, centered on the highest signal intensity point just determined. The shielding operation can be, for example, setting the signal intensity values ​​of all pixels within this influence range to zero, or marking them as unavailable in subsequent searches.

[0070] After the shielding is completed, the Figure 2 The iterative process shown in the figure repeats the "Determine the highest signal strength point on the acoustic spectrum" step on the partially blocked acoustic spectrum to find a new highest signal strength point. The validity of this new highest signal strength point is then determined again. If the new highest signal strength point is still valid, the above process is repeated: it is determined as a new leak point, and its corresponding impact range is blocked.

[0071] This iterative process continues until, after a search, the point with the highest signal strength found is deemed invalid (for example, its signal strength falls below a preset validity threshold, or no valid signal points can be found in the unmasked area of ​​the acoustic spectrum). At this point, the iterative loop terminates.

[0072] Finally, the set of all leak points identified in each valid iteration process is obtained to form a complete report containing the locations of all identified leak points on the target container, thereby achieving comprehensive and accurate positioning of multiple possible leak points.

[0073] The container air tightness detection method provided by the present invention adopts the innovative process of "search-determine-shield-iterate". After locating a point with the highest signal intensity, it actively shields its influence range, so that the sound field signal of the originally masked and weaker leakage point can be highlighted. In this way, through iterative execution, all valid leakage points can be traversed and found, thereby overcoming the technical masking effect and achieving comprehensive and accurate positioning of multiple leakage points on the target container.

[0074] The present invention determines whether the highest signal strength point is effective mainly by judging whether the signal-to-noise ratio of the highest signal strength point is greater than a preset signal-to-noise ratio threshold, and judging whether the influence range of the highest signal strength point is greater than a preset minimum area threshold. Based on this, the present invention provides an implementation method for locating possible leak points on a target container, specifically as follows: Figure 3 shown.

[0075] By inflating the target container, the internal pressure is raised to the preset pressure threshold, ensuring that the target container meets the working conditions required for detection.

[0076] During the actual inspection process, images of the product under test and raw sound source data collected by the microphone array are acquired to provide the foundation for subsequent analysis. Contour extraction technology is used to extract image features from the acquired product image to obtain the contour area of ​​the target container. This contour area is then used as a mask to perform a first spatial filtering on the full-space sound source signal collected by the microphone array, generating the first sound source signal.

[0077] The target container model information can be combined to further locate the weld position area of ​​the target container, which can be used as another mask to perform a first spatial filtering on the first sound source signal to obtain a second sound source signal.

[0078] Next, an acoustic spectrum corresponding to the second sound source signal is constructed, and the leakage point positioning of the highest signal intensity point of the acoustic spectrum is performed on it. For each located signal intensity point, a judgment is performed: by calculating the signal-to-noise ratio of the highest signal intensity point and comparing it with the preset signal-to-noise ratio threshold, it is determined whether the quality and clarity of the signal meet the requirements; by evaluating the area of ​​the influence range of the highest signal intensity point and comparing it with the preset minimum area threshold, it is determined whether the spatial characteristics of the signal conform to the spatial distribution characteristics of the actual leakage point. Only when both of the above conditions are met will the highest signal intensity point be determined to be valid, and then the influence range of the highest signal intensity point in the acoustic spectrum will be shielded to avoid its interference with subsequent positioning.

[0079] The above positioning and shielding process is iterated until the point with the highest signal strength in the current image is no longer valid (i.e., the signal-to-noise ratio falls below the threshold or the affected area is less than the preset minimum area threshold). At this point, the iteration stops. Finally, the leak point coordinates obtained from each valid iteration are fused into the image of the product under test to achieve spatially fused annotation of the leak points on the target container, visually presenting all potential leak points in the target container.

[0080] The container air tightness detection method provided by the present invention performs dual judgment of the signal-to-noise ratio and area threshold for the highest signal intensity point screened out each time, which can effectively improve the accuracy and reliability of leak point positioning. It can not only reject false signals caused by noise or artifacts, but also ensure the accurate capture of the spatial characteristics of the actual leakage point, greatly improving the confidence and practical value of the air tightness detection results.

[0081] As an optional embodiment, the present invention provides a method for calculating the signal-to-noise ratio at the highest signal strength point, which mainly includes but is not limited to the following steps: Determine the signal strength at the highest signal strength point; Determining a noise assessment area, where the noise assessment area is other areas in the acoustic spectrum after shielding the influence range of the point with the highest signal intensity; determining a background noise intensity associated with the noise assessment area; The signal-to-noise ratio of the highest signal strength point is determined based on the signal strength of the highest signal strength point and the background noise strength.

[0082] During the iterative leak location process on the target container, the present invention determines its effectiveness by calculating the signal-to-noise ratio (SNR) after locating the point with the highest signal intensity in the acoustic spectrum during each iteration. The core of the SNR quantifies the ratio between signal intensity (Signal) and background noise intensity (Noise). In this context, signal refers to the ultrasonic signal generated by the leak, while noise refers to the background ultrasonic signal generated by the environment, the device itself, and other factors when no leak is present.

[0083] The highest signal intensity point found on the acoustic spectrum is recorded as L_max, and the sound pressure level value SPL_max of the highest signal intensity point is defined as the signal intensity S, for example, S=SPL_max=75dB.

[0084] The key to calculating the signal-to-noise ratio is not to include the signal itself in the background noise. Therefore, before calculating the background noise intensity, we first determine the influence range of the highest signal strength point L_max. For example, we expand the search outward from the highest signal strength point L_max, and consider all connected areas adjacent to the highest signal strength point L_max and with sound pressure levels above a certain relative threshold (for example, SPL_max-6dB) as the influence range of the highest signal strength point.

[0085] The noise evaluation area can be defined as all other areas within the weld position area of ​​the target container, excluding the influence range of the highest signal intensity point L_max. This ensures that the real background noise that exists simultaneously with the leakage signal is evaluated.

[0086] By traversing all pixel points (or voxel points) in the determined noise assessment area and reading their sound pressure level values, the average value of these sound pressure level values ​​can be calculated to obtain the background noise intensity corresponding to the noise assessment area.

[0087] Considering that the unit of measurement for the sound pressure level is decibel, it is physically inaccurate to directly calculate the average value of the decibel value because the decibel value is a logarithmic unit. Therefore, after obtaining the sound pressure level values ​​of all pixels in the noise assessment area, the present invention performs the following operations to obtain the background noise intensity: Each pixel in the noise evaluation area i The sound pressure level value SPL_i is converted back to the linear scale sound intensity (or sound power value) I_i, and the conversion formula is: I_i=10^(SPL_i / 10).

[0088] By calculating the arithmetic mean of all these linear sound intensities I_i, we can get the average sound intensity I_noise_avg. Then convert this average sound intensity I_noise_avg back to decibel units to get the final background noise intensity N. The conversion formula is: N=SPL_noise=10 log10(I_noise_avg).

[0089] The signal-to-noise ratio (SNR) is the direct subtraction of the signal strength S and the noise strength N on a decibel scale: SNR(dB)=SN=SPL_max-SPL_noise.

[0090] Here, SPL_noise refers to the background noise intensity.

[0091] Finally, the calculated SNR value is compared with a preset signal-to-noise ratio threshold T_snr (eg, 5 dB), and the comparison result can be used as a basis for determining the validity of the highest signal strength point.

[0092] The following describes in detail how to define the contour area of ​​a target container in the container airtightness detection method provided by the present invention through specific embodiments.

[0093] As an optional embodiment, the first method for determining the contour area of ​​the target container in the image of the product to be tested provided by the present invention is a method for determining the contour area based on gradient edge detection and edge connection technology, such as Figure 4 As shown, it mainly includes but is not limited to the following steps: Step 111 : Use a first-order differential operator to perform convolution on the image of the product to be tested in both horizontal and vertical directions to obtain the gradient magnitude and gradient direction of each pixel on the image of the product to be tested.

[0094] A first-order differential operator (such as the Sobel operator) is used to perform convolution operations on the horizontal and vertical directions of the image of the product to be tested. This operation can calculate the horizontal and vertical gradient values ​​of each pixel in the image of the product to be tested.

[0095] Based on these gradient values, the Pythagorean theorem is then used to calculate the gradient magnitude (edge ​​strength) for each pixel. This is the vector sum of the horizontal and vertical gradients. The gradient direction (edge ​​direction) is also calculated for each pixel, typically the angle between the gradient vector and the horizontal axis. The gradient magnitude reflects the strength of the edge at that point, while the gradient direction indicates the direction of the edge, laying the foundation for subsequent edge refinement and connection.

[0096] In step 112, pixel points whose gradient amplitude is greater than a preset upper amplitude threshold are set as strong edge points, and pixel points whose gradient amplitude is less than or equal to the preset upper amplitude threshold but greater than a preset lower amplitude threshold are set as weak edge points.

[0097] The gradient magnitudes calculated in step 111 are classified by setting two magnitude thresholds: a preset high magnitude threshold and a preset low magnitude threshold. All pixels with magnitudes greater than the high magnitude threshold are classified as strong edge points, indicating points with very clear and reliable edge information. Pixels with magnitudes between the low and high magnitude thresholds are classified as weak edge points, indicating weak edge strength or potential edge points. Points below the low magnitude threshold are generally considered non-edge points and discarded. This classification method helps accurately capture contour edges and prevents the omission of important edge information.

[0098] Step 113 : determining, based on the gradient direction of each weak edge point, some weak edge points connected to any strong edge point among all the weak edge points.

[0099] For each weak edge point, the gradient direction is used to determine whether its neighboring pixels are strong edge points. A weak edge point is retained only if it is spatially connected to at least one strong edge point; otherwise, it is discarded. This gradient-direction-based connectivity check allows for the retention of continuous edge pixels along the edge direction and effectively filters out isolated or spurious edge points caused by noise, thereby optimizing the coherence of the edge map.

[0100] Step 114 : sequentially connect the strong edge points and some of the weak edge points to construct a closed contour of the target container.

[0101] All strong edge points and their connected valid weak edge points are sequentially connected to form continuous edge lines. Topological analysis locates the start and end points, and image processing techniques (such as edge tracing algorithms) are used to further close these edge lines to construct a closed contour of the target container. This step ensures that the container contour is completely closed, providing a clear boundary for subsequent contour region definition.

[0102] Step 115 : Determine the contour area of ​​the target container based on the closed contour.

[0103] Based on the boundaries of the closed contour, the pixel area within the contour is identified as the contour of the target container. This contour area accurately locates the target container's position and range within the image of the product under test. This contour area information is subsequently used in conjunction with the acoustic atlas to achieve corresponding sound source signal spatial mapping and leak location.

[0104] The container airtightness testing method provided by this invention significantly enhances the accuracy and integrity of target container contour edge extraction through hierarchical edge detection and gradient-direction-based weak edge point screening, reducing edge breakage and misjudgment caused by noise, interference, and image blur. The resulting closed contour area provides precise and reliable spatial information support for subsequent acoustic signal analysis, improving the spatial matching accuracy of multiple leak point locations during airtightness testing and enhancing the overall stability and reliability of the detection system.

[0105] As another optional embodiment, the second method for determining the contour area of ​​the target container in the image of the product to be tested provided by the present invention is a method based on the calculation and binarization of the gradient amplitude, thereby extracting the closed contour and contour area of ​​the container, such as Figure 5 As shown, it mainly includes but is not limited to the following steps: Step 121: Determine the gradient amplitude map of the image of the product to be tested.

[0106] Image processing algorithms (such as the Sobel operator, Prewitt operator, or Roberts operator) are used to process the image of the product being tested and calculate the gradient magnitude of each pixel. Each pixel value in the gradient magnitude map reflects the edge strength at that location, highlighting the significant features of object boundaries in the image. By calculating the grayscale change rate in the horizontal and vertical directions of the image, a complete gradient magnitude map can be generated, which serves as the basis for subsequent edge detection.

[0107] Step 122: binarize the gradient amplitude map to obtain a binarized gradient amplitude map.

[0108] To clearly distinguish edge regions from non-edge regions, the resulting gradient magnitude map can be binarized. This binarization process sets an appropriate threshold, assigning a binary value of "1" (indicating an edge) to pixels with a gradient magnitude above the threshold, and "0" (indicating a non-edge) to pixels with a gradient magnitude below the threshold. This threshold can be dynamically adjusted based on ambient lighting conditions and image quality to ensure that edge features are fully preserved while minimizing the impact of background noise.

[0109] Step 123: Determine the closed contour of the target container from the binary gradient amplitude map.

[0110] Based on the binarized gradient magnitude map, the closed contour of the target container can be identified and extracted through connected domain analysis and morphological processing (such as dilation, erosion, and closing operations). This operation involves finding edge-connected regions in the image and filling in edge breakpoints to generate a closed curve. This closed contour accurately reflects the spatial shape of the target container's edge, providing a reliable basis for locating the target container's edge.

[0111] Step 124 : determining a contour area of ​​the target container based on the closed contour.

[0112] Finally, based on the closed contour obtained in step 123, the pixel set corresponding to the closed contour is used as the contour area of ​​the target container. This contour area represents the spatial projection range of the target container in the image of the product to be tested.

[0113] The container air tightness detection method provided by the present invention adopts gradient amplitude calculation and adaptive binarization technology, combined with morphological closed contour extraction, to effectively solve the problem of insufficient or broken image edge details, ensuring the integrity and accuracy of the closed contour of the target container. The method is simple to operate, has high computational efficiency, and strong adaptability. It can stably extract the contour area of ​​the target container under various environments, significantly improving the matching degree of signal spatial positioning and overall detection accuracy during the air tightness detection process.

[0114] As an optional embodiment, the above-mentioned steps of determining the contour area of ​​the target container based on the closed contour mainly include but are not limited to: Determining a target closed contour based on the contour area, contour level, and contour position of each of the closed contours; The area where the target closed contour is located is used as the contour area of ​​the target container.

[0115] The present invention obtains a set of multiple closed contours on the target container through an image processing process. For each closed contour, its contour area, contour hierarchy structure and position distribution in the image are calculated and analyzed.

[0116] Calculating the contour area refers to measuring the number of pixels contained within the closed contour. The area size reflects the spatial range represented by the contour. Usually, the contour area of ​​the target container must be within the expected size range to avoid selecting too small noise contours or too large background contours.

[0117] Contour level refers to the relationship between a contour and other contours, that is, whether the contour is an internal segment (such as a hole or multi-layer structure) of a previous contour or an external contour. Based on the characteristics of the target container structure, the most appropriate level (usually the outermost contour) is selected to ensure contour integrity.

[0118] The contour position is determined by analyzing the spatial coordinate distribution of the contour, combining it with the preset container position range or model matching results, screening the contour area that matches the overall structural position of the product to be tested, and excluding isolated or misplaced contours caused by environmental interference.

[0119] Based on the comprehensive judgment of the above three indicators, the present invention determines the target closed contour that best meets the characteristics of the target container, and then uses the area enclosed by the determined target closed contour as the contour area of ​​the target container, which serves as the spatial reference basis for subsequent airtightness detection and leak location.

[0120] As another optional embodiment, the third method for determining the contour area of ​​a target container in an image of a product to be tested provided by the present invention is a method for determining the contour area of ​​a target container based on a contour annotation model of deep learning, such as Figure 6 As shown, it mainly includes but is not limited to the following steps: Step 131: input the image of the product to be tested into a contour annotation model, and obtain a contour area probability map output by the contour annotation model.

[0121] The product image to be tested is input into a pre-trained contour annotation model. The contour annotation model is trained based on product image samples. Each product image sample is pre-labeled with a contour area probability map label. The model learns how to accurately infer the contour distribution probability from the input image.

[0122] After training, the contour annotation model outputs a contour region probability map with the same resolution as the input image. Each pixel value in the contour region probability map represents the probability that the pixel belongs to the container contour region, and the values ​​are typically in the range [0, 1]. The loss between the output probability map of the contour annotation model and the manually annotated contour region probability map labels is calculated. Common loss functions include Binary Cross-Entropy Loss or Dice Loss. Using an optimization algorithm such as Adam, the calculated loss is backpropagated and the weight parameters in the contour annotation model are updated to minimize the loss function. When the trained contour annotation model's performance meets a preset standard on the validation set (e.g., segmentation accuracy exceeding 99.5%), training can be stopped and the trained model weight file saved for subsequent online deployment.

[0123] Among them, the contour annotation model can use Mask R-CNN (a model used for instance segmentation) or other segmentation models such as the DeepLab series.

[0124] Step 132 : performing difference assignment on the pixel points in the contour region probability map that are greater than a preset probability threshold and less than or equal to the preset probability threshold to obtain a binary mask map.

[0125] The contour area probability map can be binarized by setting a preset probability threshold (e.g., 0.5). Specifically, pixels with a contour probability greater than the threshold are assigned a value of "1," indicating that the pixel is considered a valid part of the container's contour area. Pixels with a probability less than or equal to the threshold are assigned a value of "0," indicating a non-contour area. This differential assignment generates a clear binary mask image, effectively segmenting the contour from the background.

[0126] Step 133: Determine the contour area of ​​the target container based on the binary mask image.

[0127] Based on the resulting binary mask image, morphological processing (such as opening and closing operations) can be further used to optimize the mask boundary and remove noise points or small isolated areas. Ultimately, the continuous and closed area in the mask is determined as the contour area of ​​the target container, providing a precise spatial basis for subsequent airtightness testing and leak location.

[0128] The container airtightness testing method proposed in this paper utilizes a deep learning contour annotation model to automatically extract contour regions. Compared to traditional gradient-based edge detection methods, it offers greater adaptability and robustness, effectively addressing the impact of complex backgrounds, noise interference, and image quality variations. Probabilistic graph threshold segmentation and morphological optimization ensure the accuracy and completeness of the extracted contour regions, helping to improve the precision and reliability of target container positioning during airtightness testing.

[0129] One of the core objectives of this invention is to eliminate the impact of interfering sound sources on detection results, such as those generated by the target container's external environment (e.g., gas pipes, fixtures, wall reflections, etc.). To this end, this invention proposes a spatial filtering method. Essentially, this method creates a virtual mask, defined by the camera's visual system, that only allows sound source signals within the target container's outline to pass through, while shielding or ignoring all sound source signals outside this outline.

[0130] The following will describe in detail how the present invention specifically filters out the first sound source signal from the full-space acoustic signal collected by the microphone array in conjunction with relevant embodiments.

[0131] Figure 7 FIG. 1 is a flow chart of extracting the first sound source signal within the target container contour area provided by the present invention, such as Figure 7 As shown, it mainly includes but is not limited to the following steps: Step 211, determine the coordinate system transformation matrix between the microphone sound source coordinate system and the camera image coordinate system; the microphone sound source coordinate system is the spatial coordinate system where the microphone array used for container air tightness detection is located, and the camera image coordinate system is the coordinate system where the camera used to capture the image of the product to be tested is located.

[0132] The prerequisite for extracting the contour area of ​​the target container is to establish the unification of the visual coordinate system (i.e., the camera image coordinate system) and the acoustic coordinate system (i.e., the microphone sound source coordinate system). The key lies in determining the coordinate system transformation matrix connecting these two different spatial coordinate systems.

[0133] The space in which the microphone array resides is called the microphone sound source coordinate system, which is the physical three-dimensional coordinate system for acoustic signal acquisition. The space in which the camera resides is called the camera image coordinate system, which corresponds to the two-dimensional pixel coordinate system of the image of the product being measured. Through a calibration process, such as using a known reference object or a fabricated calibration object, the coordinate transformation matrix between the two systems (including the rotation matrix and translation vector) is calculated to achieve precise coordinate conversion from the sound source space point to the image pixel point.

[0134] Step 212: Generate a full-space acoustic spectrum covering the container airtightness detection space based on the acoustic signals collected by the microphone array.

[0135] Using the acoustic signals collected by the microphone array, a full-space acoustic map covering the detection space is reconstructed using sound source localization algorithms (such as delay and sum beamforming and acoustic imaging algorithms). This acoustic map represents the sound intensity distribution at each potential sound source location in the space surrounding the target container and forms the basis for the spatial distribution of the target container's leakage signal.

[0136] Step 213: traverse each potential sound source position point in the full-space acoustic spectrum and prepare to perform spatial mapping and judgment on each sound source point.

[0137] For each potential sound source location point traversed, the three-dimensional sound source coordinates in the microphone sound source coordinate system are transformed and projected into the camera image coordinate system using the determined coordinate system transformation matrix to obtain the corresponding two-dimensional image pixel coordinates, thereby achieving spatial correspondence between the sound source location point and the container area in the visual image.

[0138] Step 214 : Projecting the microphone sound source coordinates of each potential sound source position point in the microphone sound source coordinate system to the camera image coordinate system using the coordinate system transformation matrix to obtain corresponding image pixel coordinates.

[0139] Step 215 : Determine whether the image pixel coordinates corresponding to each potential sound source position point are located within the contour area of ​​the target container.

[0140] Determine whether the coordinates of each converted image pixel fall within the predetermined target container contour area. This judgment ensures that only potential sound source signals within the target container area are focused on, and noise sources or other irrelevant sound source signals outside the container are excluded.

[0141] Step 216: If the image pixel coordinates are within the contour area, retain the area corresponding to the potential sound source position point in the full-space acoustic map.

[0142] Only when the pixel coordinates corresponding to a potential sound source point are determined to be within the target container's contour area will the area corresponding to that potential sound source point in the full-space acoustic map be retained. Otherwise, the acoustic signal corresponding to that potential sound source point is discarded, preventing the detection of sound sources outside the target container from being mistakenly included in the detection range.

[0143] Step 217: Generate the first sound source signal based on all retained areas in the full-space acoustic spectrum.

[0144] Finally, based on all retained potential sound source locations and their corresponding sound intensity values, a first sound source signal within the target container's contour area is synthesized. This first sound source signal more accurately reflects the acoustic characteristics of the target container's interior and contour area, providing an accurate data foundation for subsequent airtightness testing and leak location.

[0145] The acoustic spectrum optimization screening based on coordinate system transformation and spatial mapping provided by the present invention achieves a high degree of fusion of the spatial information of the container image and the spatial distribution of the acoustic signal, significantly improves the extraction accuracy of effective leakage sound sources within the target container contour area, effectively eliminates the source of environmental noise, enhances the pertinence and reliability of airtightness detection, and improves the spatial matching and signal quality of the entire detection process.

[0146] As an optional embodiment, Figure 8 As shown, the present invention further provides a method for determining the coordinate system transformation matrix between the microphone sound source coordinate system and the camera image coordinate system, mainly including but not limited to: Step 221 : selecting a calibration object, on which a visual calibration pattern and an acoustic beacon are provided, and the physical position relationship of the acoustic beacon relative to the visual calibration pattern is predetermined.

[0147] When installing and deploying for the first time or replacing key components (camera, microphone array), a one-time calibration procedure must be performed to establish an accurate mapping relationship between the two different sensor coordinate systems.

[0148] First, a calibration object can be created. This dual-modal calibration object can be a flat calibration plate (such as a checkerboard or a circular visual calibration pattern) with multiple acoustic beacons (such as piezoelectric ceramic buzzers that emit ultrasonic waves that can be located by a microphone array) precisely mounted at specific locations on it (for example, the center or four corners). The spatial positional relationship (including relative distance and direction) between each acoustic beacon and the visual calibration pattern on the calibration object can be measured in advance and stored as fixed physical parameters.

[0149] Step 222: photograph the calibration object with the camera and identify the visual calibration pattern, and determine the position and posture of the visual calibration pattern in the camera image coordinate system.

[0150] The calibration object is placed within the typical working area of ​​the target container and photographed with a camera to obtain an image of the product under test containing the visual calibration pattern. Using computer vision technology, the corner points or feature points of the visual calibration pattern in the product under test image are identified, and the camera's intrinsic and extrinsic parameters are calculated based on these feature points.

[0151] Camera intrinsic calibration involves photographing a calibration plate from multiple angles and using the checkerboard or circular pattern on it to calculate the camera's internal parameters (including focal length, principal point coordinates) and distortion coefficients. This step corrects for lens distortion.

[0152] Camera extrinsic calibration refers to calculating the external parameters (rotation matrix R and translation vector T) of the camera coordinate system relative to the world coordinate system (which can be defined as the plane where the calibration plate is located) based on the position of the calibration plate in the camera's field of view.

[0153] Ultimately, the 3D pose (including position and orientation) of the visual calibration pattern in the camera image coordinate system can be determined. This process can be accomplished using traditional PnP (Perspective-n-Point) algorithms or deep learning-assisted visual localization algorithms.

[0154] Step 223: Collect the acoustic signal emitted by the acoustic beacon through the microphone array, and locate the coordinates of the acoustic beacon in the microphone sound source coordinate system.

[0155] At the same location, a microphone array collects the acoustic signals emitted by each acoustic beacon. Using acoustic localization algorithms (such as sound source localization, beamforming, and time difference methods), the precise three-dimensional spatial coordinates of the acoustic beacon in the microphone source coordinate system are calculated.

[0156] Step 224 : Calculate the coordinate system transformation matrix based on the pose of the visual calibration pattern in the camera image coordinate system, the coordinates of the acoustic beacon in the microphone sound source coordinate system, and the physical position relationship.

[0157] After the above steps, several sets of paired coordinate points (depending on the number of acoustic beacons) can be obtained: the physical position of the known acoustic beacon in the world coordinate system (x_world, y_world, z_world) and the positioning result of the acoustic beacon in the microphone array coordinate system (x_mic, y_mic, z_mic).

[0158] Through these pairs of coordinate points, the transformation matrix from the microphone array coordinate system to the world coordinate system can be solved. Since the camera's extrinsic parameters also establish their connection with the world coordinate system, we can eventually obtain a projective transformation relationship directly from the microphone sound source coordinate system to the camera image coordinate system.

[0159] The present invention achieves the accurate fusion of the acoustic coordinate system and the visual coordinate system by jointly using visual calibration patterns and acoustic beacons, combined with the spatial relationship of physical measurements. This dual-modal calibration method overcomes the shortcomings of single visual or acoustic calibration, significantly improves the accuracy and stability of the coordinate transformation matrix, and provides a solid spatial alignment foundation for subsequent container leakage positioning based on acoustic signals, effectively ensuring the accuracy of spatial mapping and acoustic analysis for airtightness detection.

[0160] As another optional embodiment, the present invention also provides a method based on manually specifying corresponding points to determine the coordinate transformation matrix between the microphone sound source coordinate system and the camera image coordinate system. This method is particularly suitable for achieving accurate mapping between the two coordinate systems manually when it is impossible or inconvenient to use an automatic calibration device. The specific steps include but are not limited to: Select several landmark points with clear physical features on the surface of the target container, such as screw holes, container corners, and symbol marks. These landmark points should be easy to visually identify and evenly distributed on the structure of the target container to ensure the accuracy of spatial mapping.

[0161] The operator manually measures and records the three-dimensional coordinates of these landmarks in three dimensions. These coordinates are determined based on the microphone source coordinate system, which is the spatial reference system relative to the microphone array. Laser ranging, mechanical measuring instruments, or other high-precision measuring equipment can be used to ensure the reliability of the coordinate positions.

[0162] Then, in the image of the product to be tested captured by the camera, the 2D pixel coordinates corresponding to the above landmark points are manually marked. The operator can complete the image coordinate collection by clicking the corresponding feature point on the image through the graphical user interface.

[0163] Once we have two sets of corresponding points (three-dimensional spatial coordinates and two-dimensional pixel coordinates), we can use mathematical methods to solve the perspective transformation matrix (also known as the projection transformation matrix) using the corresponding point sets. This perspective transformation matrix (which contains the camera's internal and external parameters and spatial mapping relationships to form a coordinate system transformation matrix) can be used to accurately project the coordinate system from the microphone sound source to the camera image coordinate system.

[0164] The method of manually specifying landmark points provided by the present invention does not rely on complex calibration objects or automated equipment, is simple and flexible to operate, and is suitable for rapid deployment in complex environments in actual detection sites. By accurately selecting and measuring multiple spatial landmark points, high-precision mapping of coordinate system transformation is guaranteed, thereby improving the spatial fusion effect of acoustic signals and visual images in airtightness detection, and enhancing the accuracy and practicality of detection.

[0165] Figure 9 This is a schematic diagram of the process of determining the weld position area within the contour area provided by the present invention, such as Figure 9 As shown, it mainly includes but is not limited to the following steps: Step 311: Acquire a process design model of the target container.

[0166] The process design model is typically a 3D CAD model or a digital twin model, containing the precise structural design and manufacturing process information of the target container. The present invention provides several possible implementation methods for obtaining the process design model of the target container.

[0167] In the production management of household appliances, a barcode or QR code containing the model information is affixed to a fixed location on each target container (such as the inner tank of a water heater). This can be accomplished using a specialized industrial barcode reader or a camera. While capturing an image of the product under test, a high-definition image of the barcode or QR code can be captured simultaneously. By invoking a sophisticated barcode recognition library (such as ZBar or ZXing) and decoding the barcode in the image of the product under test, the target container's model string (for example, "Model-A-Plus-2025") can be directly and accurately obtained.

[0168] Alternatively, optical character recognition (OCR) can be used to read the model number of the target container, especially when the target container has the model number text or numbers printed on it through inkjet coding, laser engraving, etc. First, the approximate area of ​​the model number text is located through template matching, and then an OCR engine (such as Tesseract OCR) is called to recognize the image of this area and extract the model number string.

[0169] Target container models can also be identified based on image feature matching of the container's appearance or deep learning classification. For example, one or more standard images of each container model are captured, and key feature points are extracted (using algorithms such as SIFT and SURF) to build a feature library. During the actual inspection process, the real-time image of the target container is matched against the key feature points of each model in the feature library. The model with the highest match is identified as the target container model.

[0170] Prior to this, a process design model for each vessel model is obtained, typically as a CAD (Computer-Aided Design) file, perhaps in the form of a 2D DXF file or a 3D STEP / IGES file. An offline processing program is invoked to parse these CAD files and specifically extract the geometric information (typically lines or curves) identified as welds. This extracted weld information is stored in a database in a standardized format.

[0171] After obtaining the model string of the target container in the above manner, the process design model corresponding to the target container can be obtained by accessing the above database and using the determined model string as a query index to search the database.

[0172] Step 312: Based on the process design model, a three-dimensional coordinate point sequence of all welds of the target container is obtained.

[0173] Based on the retrieved process design model of the target container, the predefined weld information can be called or the spatial position data of all welds can be extracted through an automatic recognition algorithm. The specific methods include: If the process design model contains weld annotations, including related structural features (such as weld paths, weld entities, etc.), its three-dimensional boundary points, center lines or weld area boundary point sequences can be directly read.

[0174] If the model does not have clear weld markings, the weld area can be automatically identified through surface recognition (such as boundary detection and feature line extraction) and the corresponding three-dimensional coordinate point sequence can be fitted.

[0175] Step 313: Project the three-dimensional coordinate point sequence onto the image of the product to be tested based on the pose transformation matrix between the model coordinate system of the process design model and the camera image coordinate system, and obtain the weld line segments of all welds of the target container on the image of the product to be tested.

[0176] The process design model of the target container exists in its own model coordinate system, while the captured image of the product under test corresponds to a different camera image coordinate system. Spatial correspondence between the two is achieved using a pre-calculated pose transformation matrix (consisting of a rotation matrix and a translation vector). Specifically, the following transformation is performed on each 3D weld coordinate point: Using a pose transformation matrix, the 3D coordinate points are converted from the model coordinate system to the camera coordinate system. Using the camera's intrinsic parameters (focal length, principal point offset, and distortion parameters), the 3D coordinate points are then mapped back to the camera image coordinate system via the camera projection model, yielding the projected position of the weld on the image of the product being tested. By continuously mapping all weld points, a sequence of 2D image line segments corresponding to each weld seam of the target container is generated.

[0177] Step 314: Generate a weld mask on the image of the product to be tested based on the weld line segment.

[0178] Based on the projected weld line segments, a binary weld mask can be generated in the image of the product to be tested. The specific solution involves expanding the width of the weld line segments. The mask width is typically set based on the actual weld width and image resolution, for example, by expanding the line segments outward by a certain pixel range to form a strip-like area. This width expansion is achieved using the image processing dilation operation to ensure that the weld mask area fully covers the actual possible weld location within a certain tolerance range. The mask image is a binary image, with pixels in the weld mask area set to "1" and the background set to "0."

[0179] Step 315: Determine the intersection of the weld mask and the contour area as the weld position area.

[0180] Finally, an image processing algorithm can be used to calculate the intersection of the weld mask and the previously determined target container's contour area. This can be achieved using a logical AND operation on binary images. The resulting intersection is the portion within the weld mask that falls within the target container's contour area. This intersection is then used to determine the final weld location, ensuring accurate weld positioning within the container structure. This provides a reliable spatial reference for subsequent weld-based acoustic testing and leak analysis.

[0181] The container airtightness testing method provided by this invention precisely extracts the three-dimensional coordinates of welds from a process design model and maps them to a two-dimensional image using accurate spatial transformation. Combined with image processing, this method generates a weld mask of appropriate width, enabling precise identification of the weld location range in actual product images. Furthermore, by intersecting the weld mask with the target container's contour, the identified weld area is ensured to fully conform to the container structure, effectively eliminating background and interference areas. This method significantly improves the accuracy and stability of weld area positioning, providing a reliable spatial basis for focused weld analysis and leak source identification during airtightness testing.

[0182] Figure 10 : is a flow chart of extracting a second sound source signal in a weld position area from a first sound source signal provided by the present invention, such as Figure 10 As shown, it mainly includes but is not limited to the following steps: Step 411: traverse each potential sound source position point in the acoustic spectrum corresponding to the first sound source signal.

[0183] First, the acoustic map corresponding to the first sound source signal is loaded. This acoustic map is a 3D spatial coordinate grid or point cloud structure, with each coordinate point (potential sound source location) carrying corresponding sound intensity information. A loop traversal algorithm is used, starting from the starting spatial point to the ending spatial point (for example, from the upper left corner to the lower right corner), to access and read the 3D coordinates and corresponding sound energy intensity of each potential sound source location point one by one.

[0184] The traversal can use a spatial index structure, such as an octree or a KD tree, to accelerate the search, ensuring traversal efficiency while supporting fast access to large-scale acoustic atlas data.

[0185] Step 412 : Projecting the microphone sound source coordinates of each potential sound source position point in the microphone sound source coordinate system to the camera image coordinate system using the coordinate system transformation matrix to obtain corresponding image pixel coordinates.

[0186] For each potential sound source position point traversed, the coordinate system transformation matrix obtained in advance through calibration or measurement can be called to convert it from the microphone sound source coordinates to the two-dimensional camera image coordinates: The rotation matrix and translation vector are used to transform the three-dimensional coordinates from the microphone sound source coordinate system to the camera image coordinate system, obtaining the converted three-dimensional point. The resulting three-dimensional point is then projected onto the two-dimensional image plane using the camera's intrinsic parameters (including focal length, principal point coordinates, and radial and tangential distortion coefficients) to obtain the accurate image pixel coordinates (x, y).

[0187] In order to reduce the influence of distortion, a camera distortion correction algorithm can also be used to perform distortion correction on the obtained image pixel coordinates to ensure the accuracy of the mapping.

[0188] Step 413: If the image pixel coordinates are located within the weld location area, retain the area corresponding to the potential sound source location point in the acoustic spectrum corresponding to the first sound source signal.

[0189] Using the pre-obtained weld mask (binary image), the regional judgment is performed on each projected pixel coordinate: according to the position of the image pixel coordinate in the weld mask, the corresponding pixel value is read. If the value is "1" (representing the weld mask area), it indicates that the potential sound source location point is inside the weld area.

[0190] Only points that meet this condition will retain their valid 3D sound source data to construct a valid sound source set for the weld area. Of course, a boundary tolerance can be set during the judgment process, for example, allowing a tolerance range of ± a few pixels near the weld mask boundary to account for uncertainty in the mask edge.

[0191] Step 414: Generate the second sound source signal based on all retained regions in the acoustic spectrum corresponding to the first sound source signal.

[0192] By weighting or accumulating the sound intensity data from the selected sound source set within the weld location area, a second sound source signal reflecting the weld location area can be generated. Sound intensity weighting can be combined with spatial neighborhood filtering, such as Gaussian filtering, to smooth the local sound intensity distribution and reduce the influence of random noise. Clustering algorithms (such as DBSCAN) can also be used to perform density analysis on selected sound source points to extract representative sound source distributions, further improving the spatial focus of the signal.

[0193] The container airtightness detection method provided by the present invention effectively eliminates irrelevant sound source data outside the weld location area through specific spatial traversal, precise coordinate transformation and strict screening based on weld masks, while retaining the real leakage sound source characteristics within the weld location area, providing a reliable and clear acoustic information basis for the airtightness detection system, and facilitating the precise implementation of subsequent leak identification and positioning algorithms.

[0194] As an optional embodiment, the container air tightness detection method provided by the present invention can be used to detect the inner tank of an electric water heater, or other sealed containers that require air tightness detection.

[0195] During actual testing, it was found that the ultrasonic signal generated by a simple gas leak has a low signal-to-noise ratio. While spraying water mist or soapy water can amplify the ultrasonic signal, the water vapor can affect the detection sensitivity of the microphone array unit. In light of this, the container airtightness testing method provided by the present invention employs spraying a rust-resistant, highly volatile solvent on the surface of the target container to enhance the signal-to-noise ratio of the ultrasonic signal generated by the leak point.

[0196] As an optional embodiment, the solvent is isopropyl alcohol, D-limonene, ethanol, glycerol or a mixture of multiple solvents.

[0197] Figure 11 Schematic diagram of the structure of the container air tightness detection device provided by the present invention, as shown in FIG. Figure 11 As shown, the present invention also provides a container airtightness detection device, which mainly includes but is not limited to: The container positioning unit 101 is used to determine the contour area of ​​the target container in the image of the product to be tested; A sound source signal preliminary screening unit 102 is used to extract the first sound source signal in the contour area; The weld positioning unit 103 is used to determine the weld position area within the contour area; A sound source signal re-screening unit 104 is configured to extract a second sound source signal in the weld position area from the first sound source signal; The leakage point locating unit 105 is configured to locate a possible leakage point on the target container based on the second sound source signal.

[0198] It should be noted that, when the container air tightness detection device provided by the present invention is executed, it can implement the container air tightness detection method provided by any of the above embodiments, which will not be described in detail here.

[0199] The container airtightness detection device provided by the present invention creatively integrates visual information and acoustic information in depth. Through two-stage progressive spatial filtering, the detected sound source signal is strictly limited to the weld location area where leakage is most likely to occur, thereby greatly eliminating interference from environmental noise, signal reflections and signals in non-critical areas, and achieving high-precision and high-reliability positioning of the container leakage point.

[0200] Figure 12 Schematic diagram of the structure of the electronic device provided by the present invention, such as Figure 12 As shown, the electronic device may include: a processor 1210, a communications interface 1220, a memory 1230, and a communications bus 1240. The processor 1210, the communications interface 1220, and the memory 1230 communicate with each other via the communications bus 1240. The processor 1210 may invoke logic instructions in the memory 1230 to execute a container airtightness detection method, which includes: determining a contour region of a target container in an image of a product to be tested; extracting a first sound source signal within the contour region; determining a weld location region within the contour region; extracting a second sound source signal within the weld location region from the first sound source signal; and locating a potential leak point on the target container based on the second sound source signal.

[0201] Furthermore, the logic instructions in the aforementioned memory 1230 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product, stored in a storage medium, includes instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0202] 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 container air tightness detection method provided by the above-mentioned embodiments, the method including: determining the contour area of ​​the target container in the image of the product to be tested; extracting a first sound source signal within the contour area; determining the weld position area within the contour area; extracting a second sound source signal within the weld position area from the first sound source signal; and locating possible leakage points on the target container based on the second sound source signal.

[0203] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, it is implemented to perform the container air tightness detection method provided in the above-mentioned embodiments, the method including: determining the contour area of ​​the target container in the image of the product to be tested; extracting a first sound source signal within the contour area; determining the weld position area within the contour area; extracting a second sound source signal within the weld position area from the first sound source signal; and locating possible leakage points on the target container based on the second sound source signal.

[0204] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0205] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0206] 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 various embodiments of the present invention.

Claims

1. A method for detecting air tightness of a container, characterized in that: include: Determine the contour area of ​​the target container in the image of the product to be tested; Extracting a first sound source signal within the contour area; determining a weld position region within the contour region; Extracting a second sound source signal in the weld position area from the first sound source signal; Based on the second sound source signal, a possible leakage point on the target container is located.

2. The container airtightness detection method according to claim 1, characterized in that: The locating a possible leakage point on the target container based on the second sound source signal includes: Obtaining an acoustic spectrum corresponding to the second sound source signal; Determining the point with the highest signal intensity on the acoustic spectrum; When it is determined that the highest signal intensity point is valid, determining a leakage point of the target container according to the position of the highest signal intensity point in the acoustic spectrum; Shielding the influence range of the point with the highest signal intensity in the acoustic spectrum; Iteratively executing the step of determining the highest signal strength point on the acoustic spectrum, shielding the influence range of the highest signal strength point in the acoustic spectrum, and determining that the highest signal strength point is invalid; Obtaining a set of leakage points of the target container determined during each iteration; The influence range of the highest signal strength point refers to all connected areas centered on the highest signal strength point where the signal strength is greater than a preset signal strength.

3. The container airtightness detection method according to claim 2, characterized in that: Whether the highest signal strength point is valid is determined by judging whether the signal-to-noise ratio of the highest signal strength point is greater than a preset signal-to-noise ratio threshold, and judging whether the influence range of the highest signal strength point is greater than a preset minimum area threshold.

4. The container airtightness detection method according to claim 3, characterized in that: The signal-to-noise ratio at the highest signal strength point is determined based on the following steps: Determine the signal strength at the highest signal strength point; Determining a noise assessment area, where the noise assessment area is other areas in the acoustic spectrum after shielding the influence range of the point with the highest signal intensity; determining a background noise intensity associated with the noise assessment area; The signal-to-noise ratio of the highest signal strength point is determined based on the signal strength of the highest signal strength point and the background noise strength.

5. The container airtightness detection method according to claim 1, characterized in that: The step of determining a contour area of ​​a target container in the image of the product to be tested includes: Using a first-order differential operator to convolve the image of the product to be tested in the horizontal and vertical directions to obtain the gradient amplitude and gradient direction of each pixel point on the image of the product to be tested; Pixels whose gradient amplitude is greater than a preset upper amplitude threshold are defined as strong edge points, and pixels whose gradient amplitude is less than or equal to the preset upper amplitude threshold but greater than a preset lower amplitude threshold are defined as weak edge points; Based on the gradient direction of each of the weak edge points, determining a portion of the weak edge points connected to any of the strong edge points among all the weak edge points; sequentially connecting the strong edge points and some of the weak edge points to construct a closed contour of the target container; Based on the closed contour, a contour area of ​​the target container is determined.

6. The method for detecting air tightness of a container according to claim 1, wherein: The step of determining the contour area of ​​the target container in the image of the product to be tested further includes: Determining a gradient magnitude map of the image of the product to be tested; performing binarization processing on the gradient amplitude map to obtain a binarized gradient amplitude map; determining a closed contour of the target container from the binary gradient amplitude map; Based on the closed contour, a contour area of ​​the target container is determined.

7. The container airtightness detection method according to claim 5 or 6, characterized in that: The determining of the contour area of ​​the target container based on the closed contour includes: Determining a target closed contour based on the contour area, contour level, and contour position of each of the closed contours; The area where the target closed contour is located is used as the contour area of ​​the target container.

8. The container airtightness detection method according to claim 1, characterized in that: The step of determining the contour area of ​​the target container in the image of the product to be tested further includes: Inputting the image of the product to be tested into a contour annotation model, and obtaining a contour area probability map output by the contour annotation model; Performing differential assignment on pixel points in the contour area probability map that are greater than a preset probability threshold and less than or equal to the preset probability threshold to obtain a binary mask map; determining a contour area of ​​the target container based on the binary mask image; The contour annotation model is obtained by training based on product image samples, and each of the product image samples is pre-annotated with a contour region probability map label.

9. The container airtightness detection method according to claim 1, characterized in that: The extracting the first sound source signal within the contour area includes: Determine a coordinate system transformation matrix between a microphone sound source coordinate system and a camera image coordinate system; the microphone sound source coordinate system is the spatial coordinate system of a microphone array used for container air tightness testing, and the camera image coordinate system is the coordinate system of a camera used to capture images of the product to be tested; Based on the acoustic signals collected by the microphone array, a full-space acoustic spectrum covering the airtightness detection space of the container is generated; Traversing each potential sound source position point in the full-space acoustic map; Using the coordinate system transformation matrix, projecting the microphone sound source coordinates of each potential sound source position point in the microphone sound source coordinate system to the camera image coordinate system to obtain corresponding image pixel coordinates; Determining whether the image pixel coordinates corresponding to each potential sound source position point are located within the contour area of ​​the target container; If the image pixel coordinates are within the contour area, retaining the area corresponding to the potential sound source position point in the full-space acoustic map; The first sound source signal is generated based on all retained areas in the full-space acoustic spectrum.

10. The container airtightness detection method according to claim 9, characterized in that: Determining the coordinate system transformation matrix between the microphone sound source coordinate system and the camera image coordinate system includes: Selecting a calibration object, wherein the calibration object is provided with a visual calibration pattern and an acoustic beacon, and the physical position relationship of the acoustic beacon relative to the visual calibration pattern is predetermined; photographing the calibration object with the camera and identifying the visual calibration pattern, and determining the position and pose of the visual calibration pattern in the camera image coordinate system; collecting the acoustic signal emitted by the acoustic beacon through the microphone array, and locating the coordinates of the acoustic beacon in the microphone sound source coordinate system; The coordinate system transformation matrix is ​​calculated based on the pose of the visual calibration pattern in the camera image coordinate system, the coordinates of the acoustic beacon in the microphone sound source coordinate system, and the physical position relationship.

11. The container airtightness detection method according to claim 9, characterized in that: Determining the weld position area within the contour area includes: Obtaining a process design model of the target container; Based on the process design model, a three-dimensional coordinate point sequence of all welds of the target container is obtained; Projecting the three-dimensional coordinate point sequence onto the image of the product to be tested based on a pose transformation matrix between the model coordinate system of the process design model and the camera image coordinate system, and obtaining weld line segments of all welds of the target container on the image of the product to be tested; generating a weld mask on the image of the product to be tested based on the weld line segment; An intersection of the weld mask and the contour area is determined as the weld position area.

12. The container airtightness detection method according to claim 11, characterized in that: The extracting the second sound source signal in the weld position area from the first sound source signal comprises: Traversing each potential sound source position point in the acoustic spectrum corresponding to the first sound source signal; Using the coordinate system transformation matrix, projecting the microphone sound source coordinates of each potential sound source position point in the microphone sound source coordinate system to the camera image coordinate system to obtain corresponding image pixel coordinates; If the image pixel coordinates are located within the weld location area, retaining the area corresponding to the potential sound source location point in the acoustic spectrum corresponding to the first sound source signal; The second sound source signal is generated based on all retained regions in the acoustic spectrum corresponding to the first sound source signal.

13. The container airtightness detection method according to claim 1, characterized in that: The target container is the inner tank of an electric water heater.

14. The container airtightness detection method according to claim 1, characterized in that: A solvent having rust-proof, strong tension and volatile properties is sprayed on the surface of the target container.

15. A container air tightness detection device, characterized in that: include: A container positioning unit, used to determine the contour area of ​​the target container in the image of the product to be tested; a sound source signal primary screening unit, configured to extract a first sound source signal within the contour area; a weld positioning unit, configured to determine a weld position region within the contour region; a sound source signal re-screening unit, configured to extract a second sound source signal in the weld position area from the first sound source signal; A leakage point locating unit is used to locate a possible leakage point on the target container based on the second sound source signal.

16. 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 container airtightness detection method according to any one of claims 1 to 14 is implemented.

17. 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 container airtightness detection method according to any one of claims 1 to 14 is implemented.

18. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the container airtightness detection method according to any one of claims 1 to 14 is implemented.

Images