Hollow glass dew point detection method based on image dynamic discrimination

Abstract

The invention discloses a hollow glass dew point detection method based on image dynamic discrimination, and the method is characterized in that the method comprises the following steps: 1, placing a to-be-detected hollow glass at a detection station, and enabling a detection region of the to-be-detected hollow glass to be attached to a cold conduction surface; 2, controlling the temperature of the cold conduction surface to be reduced to a target detection temperature zone according to a preset cooling program; in the cooling and constant temperature process, continuous image acquisition is carried out on the detection area, and an image frame sequence is obtained; a third step of extracting candidate condensation or frosting targets based on the image frame sequence, and performing dynamic comparison on the candidate condensation or frosting targets at different moments; fourthly, when it is detected that the newly-appearing dew formation or frosting target meets the preset judgment condition, it is judged that the dew point phenomenon happens to the hollow glass, and the corresponding dew formation temperature is recorded. The method has the advantage that the accuracy and repeatability of dew point detection results can be improved.

Description

Technical Field

[0001] This invention belongs to the field of insulated glass performance testing methods, specifically relating to an image-based dynamic discrimination method for insulated glass dew point detection. Background Technology

[0002] Insulating glass is widely used in building doors, windows, and curtain walls due to its excellent heat and sound insulation properties. The sealing performance of insulating glass directly affects its service life and energy-saving effect. Dew point is one of the important indicators for measuring the sealing quality and the dryness of the cavity. If the water vapor content inside the insulating glass cavity is high, condensation or frost is likely to occur in low-temperature environments, affecting not only transparency and aesthetics but also potentially reducing its thermal insulation and energy-saving performance.

[0003] Currently, the dew point performance of insulated glass is typically tested according to relevant standards. The basic principle involves locally cooling the insulated glass under specific environmental conditions and observing whether condensation or frost forms inside the glass. If condensation or frost occurs, the insulated glass is placed until it is completely free of condensation or frost, and then the cooling temperature is increased by 5°C each time until no more condensation or frost appears. This temperature is then determined as the dew point temperature of the insulated glass. In practical applications, this type of testing method mainly involves gradually increasing the testing temperature from low to high and using manual observation or simple image recording to complete the dew point test.

[0004] However, existing methods for detecting the dew point of insulating glass still have the following shortcomings in actual testing: 1. Existing methods for testing the dew point of insulated glass begin at a low temperature. If condensation or frost appears, the temperature is gradually increased until no more frost or condensation occurs, at which point the temperature is determined as the condensation temperature of the insulated glass. This type of testing method is time-consuming, labor-intensive, and has low accuracy.

[0005] 2. During the dew point testing of insulated glass, the dew point testing device needs to be removed to manually observe whether condensation or frost occurs inside the insulated glass. It is impossible to observe and record the condensation or frost process in real time, and the traceability of the test results is difficult to guarantee.

[0006] 3. Traditional detection methods usually rely on manual visual judgment to determine the occurrence of condensation or frost. The detection results are easily affected by factors such as the operator's experience, observation angle, and lighting conditions, which are highly subjective and make it difficult to guarantee the consistency and repeatability of the detection results.

[0007] 4. During the cooling detection of insulating glass, the operating environment of the detection device is complex. The refrigeration unit may vibrate during operation, and there may be dust, stains or inherent defects on the surface of the insulating glass. These factors can easily form interference features in the image, leading to misjudgment or missed judgment by automatic recognition methods based on single-frame images or simple threshold judgment.

[0008] 5. Existing automated detection solutions often focus on the analysis of single image features, without fully considering the evolution characteristics of condensation or frost phenomena over time. This makes it difficult to accurately distinguish between initial condensation and background interference, and also makes it difficult to provide stable and reliable dew point determination results when condensation first appears.

[0009] Therefore, how to improve the detection efficiency and accuracy in the dew point detection process of insulating glass, and how to achieve automatic, objective, and stable identification and visual recording of condensation or frost phenomena under complex working conditions such as low temperature and vibration, so as to improve the accuracy and repeatability of dew point detection results, remains a technical problem that urgently needs to be solved in this field.

[0010] Based on this, the applicant considers designing an image-based dynamic discrimination method for dew point detection of insulating glass that can improve the accuracy and repeatability of dew point detection results. Summary of the Invention

[0011] In view of the shortcomings of the prior art, the technical problem to be solved by the present invention is: How to design an image-based dynamic discrimination method for dew point detection of insulating glass that can improve the accuracy and repeatability of dew point detection results?

[0012] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: A method for detecting dew point of insulating glass based on image dynamic discrimination, characterized by comprising the following steps: Step 1: Place the insulating glass to be tested at the testing station, ensuring that the testing area is in contact with the cooling surface; The second step is to control the temperature of the cooling surface to decrease to the target detection temperature zone according to a preset cooling program; during the cooling and constant temperature process, continuous image acquisition is performed on the detection area to obtain an image frame sequence; The third step is to extract candidate dew or frost targets based on the image frame sequence, and to dynamically compare the candidate dew or frost targets at different times. Step 4: When a newly detected condensation or frost target meets the preset judgment conditions, it is determined that the insulated glass has a dew point phenomenon, and the corresponding condensation temperature is recorded.

[0013] Compared with existing technologies, the image-based dynamic discrimination method for detecting dew point in insulating glass has the following advantages: 1. Continuous and precise cooling from high to low temperatures improves detection efficiency and accuracy. This invention not only controls the temperature of the cooling surface to gradually increase from low to high temperature and maintain a constant temperature for dew point standard method detection, but also controls the temperature of the cooling surface to continuously decrease from high to low temperature for high-precision dew point detection, greatly improving the efficiency and accuracy of dew point detection for insulating glass, and making it more suitable for on-site testing.

[0014] 2. Determine condensation or frost based on dynamic change characteristics to avoid misjudgment in a single frame. Unlike existing technologies that determine condensation or frost based on single images or instantaneous features, this invention continuously and dynamically compares candidate water mist targets at different times to identify the number of newly appearing water mist targets, and uses this as the basis for determining condensation or frost. By fully utilizing the temporal generation characteristics of water mist, it significantly reduces the risk of misjudgment caused by instantaneous noise, changes in lighting, or occasional interference.

[0015] 3. Test results are traceable, facilitating quality assessment and verification. Upon determining that a dew point phenomenon has occurred, the present invention can automatically trigger the recording and storage of image or video data, and associate the data with the detection temperature and detection time for storage, which facilitates subsequent review of the detection process and results, improves the traceability and credibility of the detection conclusions, and better ensures the accuracy and repeatability of the dew point detection results. Attached Figure Description

[0016] Figure 1 This is a flowchart of the image-based dynamic discrimination method for detecting dew point in insulating glass according to the present invention; Figure 2 This is a screenshot of the user interface of the detection terminal tablet using the method of the present invention before detection; Figure 3a This is a single-frame image of the detection area (without dew points) acquired using the method of the present invention. Figure 3b A single frame image of the detection area acquired using the method of this invention (acquisition time point later than...) Figure 3a ); Figure 3c A single frame image of the detection area acquired using the method of this invention (acquisition time point later than...) Figure 3b (Dew has already appeared). Figure 4 This is a screenshot of the user interface of the detection terminal tablet using the method of the present invention when a preset dew point number is detected; Figure 5 The image shows the intelligent dew point detection device for insulating glass and the actual insulating glass used in the method of this invention. Figure 6This is a three-dimensional structural diagram of the intelligent dew point detection device for insulating glass used in the method of the present invention. Figure 7 This is a three-dimensional structural diagram of the intelligent dew point detection device for insulating glass used in the method of the present invention. Figure 8 This is a photograph of the control and processing unit of the intelligent dew point detection device for insulating glass used in the method of this invention.

[0017] The diagram is marked as follows: 10. Insulating glass; 20 Refrigeration unit; 21 Cooling surface; 22 Brass block; 23 Guide rail; 24 Slider; 25 Lead screw; 30 Image acquisition unit: 31 Suction cup bracket, 32 Camera, 33 Support rod; 40 Support and positioning mechanism: 41 Slot block, 42 ​​Threaded push rod. Detailed Implementation

[0018] The present invention will now be described in further detail with reference to the accompanying drawings.

[0019] like Figure 1 As shown, the image-based dynamic discrimination method for detecting the dew point of insulating glass includes the following steps: Step 1: Place the insulating glass to be tested at the testing station, ensuring that the testing area is in contact with the cooling surface; The second step is to control the temperature of the cooling surface to decrease to the target detection temperature zone according to a preset cooling program; during the cooling and constant temperature process, continuous image acquisition is performed on the detection area to obtain an image frame sequence; The third step is to extract candidate dew or frost targets based on the image frame sequence, and to dynamically compare the candidate dew or frost targets at different times. Step 4: When a newly detected condensation or frost target meets the preset judgment conditions, it is determined that the insulated glass has a dew point phenomenon, and the corresponding condensation temperature is recorded.

[0020] In the initial stage of the cooling process in the second step, a background acquisition step is set up. This step records the image features that are continuously present in the detection area and excludes the image features as background interference in the subsequent detection process.

[0021] This invention introduces a background acquisition step (also known as a background acquisition time period) before dew point determination. Targets that persist within the detection area in the initial stage are recorded and treated as background interference targets. In subsequent detection processes, only newly appearing targets that do not match the aforementioned background interference targets are identified as water mist targets. This method effectively avoids misidentifying dust, scratches, or inherent impurities on the glass surface as dew point water mist, fundamentally improving the reliability of the detection results.

[0022] In the third step, the cross-union ratio (CUI) between candidate condensation or frost targets at different time points is calculated. When the CUI is less than a preset threshold, the corresponding candidate condensation or frost target is determined as a newly appearing condensation or frost target.

[0023] The Intersection over Union (IoU) ratio is defined as: IoU = The numerator (intersection) is the area of ​​overlap between the two regions; the denominator (union) is the total area occupied by the two regions together.

[0024] In this embodiment, in order to distinguish whether the condensation or frost targets detected at different time points are the same condensation or frost phenomenon, spatial overlap analysis is performed on the candidate condensation or frost target areas at different times.

[0025] Specifically, the cross-union ratio (CUI) between the current candidate condensation or frost target area and the historically recorded condensation or frost target areas is calculated. The CUI is the ratio of the overlapping area of ​​the two target areas to their union area. When the CUI is less than a preset threshold, the current candidate condensation or frost target is determined to be a newly appearing condensation or frost target. For example, when IoU ≈ 0.6-0.9, it is considered basically overlapping; when IoU ≈ 0.3-0.5, it is partially overlapping; and when IoU < 0.2, it is almost non-overlapping, and thus a new water droplet is determined to have appeared. In implementation, the preset threshold is preferably 0.5–0.6.

[0026] In summary, this preferred solution, by introducing an intersection-union ratio (IUU) mechanism, can effectively improve the stability of water mist identification. In the preferred embodiment, this invention calculates the IUU among candidate water mist targets to match and distinguish targets at different time points. Only when the IUU is less than a preset threshold is it determined to be a newly appearing water mist target. Therefore, the IUU-based new target determination mechanism can accurately distinguish between "newly generated water mist" and "existing targets whose positions remain basically unchanged," avoiding duplicate counting and making the dew point determination results more stable and reliable.

[0027] In practice, a candidate condensation or frost target is only considered a valid condensation or frost target for dew point determination if its area falls within a preset area range. Similarly, a detection area for dew point detection is defined within the image frame sequence, and condensation detection processing is performed only on image data within this detection area.

[0028] The condensation detection is performed periodically at a preset time interval, which is 0.5-2 seconds.

[0029] In this embodiment, condensation detection is not performed on every frame of the image, but rather periodically at preset time intervals. The preset time interval is preferably 0.5-2 seconds, more preferably 1 second, to reduce computational load and improve system stability while maintaining initial condensation detection sensitivity.

[0030] In an optional implementation, the preset time interval can be adaptively adjusted according to the temperature stage: a smaller time interval is used during the background data acquisition stage, and the time interval is appropriately reduced during the stage approaching the dew point temperature range to improve the accuracy of dew point determination. That is, a staged adaptive cycle approach is preferred, specifically divided into the following stages: 1. Background acquisition phase (first 8–12 seconds): 0.2–0.5 seconds / time (more frequent, to ensure reliable background modeling); 2. Stable detection phase: 1 second / test; 3. Approaching the dew point temperature range (e.g., when the temperature enters 0–5℃ or approaches the target temperature ±2℃): 0.5s / cycle; 4. After triggering: Revert to 1–2 seconds per trigger or simply retain evidence.

[0031] As can be seen from the above, in addition to continuous imaging and detection, the method of the present invention can easily adopt a periodic detection strategy, rather than performing water mist detection on every image frame. While ensuring detection sensitivity, it effectively reduces the computational load of image processing and avoids system lag or crashes during long-term operation. This allows the method of the present invention to run stably in actual detection equipment for extended periods, demonstrating good engineering applicability.

[0032] When the number of newly formed condensation or frost targets reaches the first preset threshold, the dew point determination is triggered; when the number of newly formed condensation or frost targets reaches the second preset threshold, the detection process is terminated.

[0033] In implementation, the first preset threshold is preferably 8-15, more preferably 10; the second preset threshold is preferably 40-90, more preferably 80 (e.g., Figure 4 As shown, the second preset threshold is 85.

[0034] By setting a first preset threshold and a second preset threshold to classify and control the number of newly appearing condensation or frost targets, this method can achieve stable and reliable judgment and operation control during dew point detection. Specifically, when the number of newly appearing condensation or frost targets reaches the first preset threshold, dew point determination is triggered, thereby avoiding misjudgment due to occasional condensation or momentary interference, and improving the accuracy and consistency of dew point determination; when the number of newly appearing condensation or frost targets reaches the second preset threshold, the detection process is automatically terminated, which can effectively prevent meaningless repeated calculations caused by a large amount of condensation, reduce the system's computational load, and avoid system lag or abnormal operation.

[0035] Based on the above-mentioned graded threshold control method, the dew point detection has both high reliability in judgment and stability and practicality in system operation, making it suitable for long-term stable application in actual testing equipment.

[0036] In the fourth step: before and after the dew point determination, video data of the detection area is automatically recorded, and the video data is stored in association with the detection temperature and detection time.

[0037] In this invention, "before and after dew point determination" refers to a time interval surrounding the trigger moment of dew point determination, rather than a single instant. Specifically, "before dew point determination" refers to the time period from the start of cooling or detection until the dew point determination conditions are met during the detection process; "after dew point determination" refers to the time period after the dew point determination conditions are met and the dew point determination is completed.

[0038] In practice, the video data recording should at least cover the detection process before the dew point determination trigger time, and may further cover a portion of the time after the dew point determination trigger time as needed, for the purpose of verifying and analyzing the dew point occurrence process. It should be understood that the specific duration of the "before and after" time interval can be set according to actual detection needs and does not constitute a limitation on the scope of protection of this invention.

[0039] This preferred solution automatically records video data of the detection area during dew point detection and associates and stores this video data with the corresponding detection temperature and detection time, ensuring complete and continuous video recording of the dew point determination process. This achieves the following advantages: 1. The video recording covers the cooling process before the dew point occurs and the trigger time for dew point determination, which can provide intuitive and traceable evidence to support the dew point detection results, facilitating subsequent review and analysis; 2. Automatically terminate video recording after detecting that the dew point determination condition is met. This avoids meaningless continuous recording during periods of heavy condensation, reducing data storage pressure and system operating burden.

[0040] Therefore, by adopting the above methods, while ensuring the reliability of dew point detection results, the automation level of the detection process and the system operating efficiency are improved, making dew point detection more suitable for the long-term stable use of actual testing equipment.

[0041] Preferably, the above-mentioned insulated glass dew point detection also includes an image recognition and determination algorithm, specifically: I. Overview of Image Recognition and Judgment Algorithms This algorithm can solve the following technical problems and achieve corresponding beneficial technical effects: 1. Using a cooling surface (preferably made of brass) as a background can enhance the contrast of water droplets / mist, but it can also cause reflections and texture interference. 2. Slight image displacement caused by glass or equipment vibration can easily lead to misjudgment of single-frame thresholds; 3. Dust and stains are "static targets" and need to be automatically eliminated in the initial stage of detection; 4. Initial condensation usually manifests as localized bright spots, increased local contrast, and changes in edge density. The "newly appearing dew point" should be determined using the time dimension.

[0042] II. Objectives and Input / Output of Image Recognition and Judgment Algorithms 1. Objective During the cooling and temperature maintenance process, the camera continuously acquires images to automatically identify the moment when initial condensation (dew point) occurs in the cavity of the insulating glass unit, and outputs: dew point determination result (whether condensation occurs) and dew point temperature. (The glass surface temperature or cooling surface temperature corresponding to the trigger of condensation determination) and condensation evidence (keyframe, condensation area, condensation or frost target quantity curve).

[0043] 2. The algorithm input includes: a sequence of consecutive image frames. (Frame rate 15–30fps, 1080p / 2k), temperature sequence (From cooling control / temperature sensor, cycle 0.5–1s) and optional inputs: contact status / stability flag (no input if none).

[0044] 3. The algorithm output is: dew point occurrence flag. Dew point temperature and the target quantity of condensation or frost. Confidence level of condensation .

[0045] III. Algorithm Implementation Steps for Image Recognition and Judgment 1. Step S1: Adaptive ROI Determination (Detection Area Localization) After the first frame or each refocus, the image is coarsely segmented to locate the hollow glass detection area and construct the Region of Interest (ROI): the effective glass area is located through edge detection and contour filtering, and the ROI is taken as the center of the glass area. The rectangular region is selected, avoiding borders, clamps, or strongly reflective edges, to obtain the final ROI image. (Note: ROI limitation can significantly reduce false positives and improve real-time performance.)

[0046] 2. Step S2: Image stabilization and brightness normalization (anti-shake and anti-illumination fluctuation) To combat high-frequency jitter and brightness variations, two types of processing are applied to the ROI: (1) Inter-frame micro-alignment (anti-shake): Calculation and Phase-related displacement ,right Translation compensation is obtained .

[0047] (2) Brightness normalization (anti-glare): For Perform Claeshe or piecewise linear normalization to obtain (Note: Jitter compensation and brightness normalization can significantly reduce "pseudo-condensation or frosting (reflection / jitter edges)").

[0048] 3. Step S3: Dynamic background modeling and static interference removal (dust / stain removal) Set the initial background capture window (e.g., 8–12 seconds). Within the window, candidate targets are extracted for each frame and accumulated into a "static background library": for Calculate the candidate target set (See step S4) Targets that appear continuously or overlap significantly within the window will be added to the background library. In subsequent detection phases, any candidate target that is related to... The objective satisfies: intersection-union ratio (e.g., 0.3) or distance < And the area difference < If it is not, it is determined to be background interference and removed (the advantage is that background interference is defined by "persistence combined with overlap", which is more robust than a simple threshold).

[0049] 4. Step S4: Extraction of dew or frost candidates (multi-feature fusion to generate a "dew or frost probability map") For each frame Calculate the three types of features and fuse them into a probability map of condensation or frost. : Feature 1: Localized high-brightness speckle pattern (sensitive to water droplet reflection): Calculating the high-brightness mask .

[0050] Feature 2: Local contrast abrupt change feature (sensitive to water mist texture): Calculate local variance / local entropy ,Pick High response area.

[0051] Feature 3: Edge density variation characteristics (sensitive to the edges of newly formed water droplets): Calculate gradient magnitude or Laplacian Select the high response region.

[0052] The fusion method can adopt a weighted approach: in Weights (can be determined in calibration, default values ​​are acceptable) ).

[0053] Subsequently Binarization and morphological closing operations yield a set of candidate connected components. .

[0054] The advantage of the fusion method in this scheme is that condensation or frost is not a single edge or a single brightness feature. This algorithm uses "high brightness, contrast and edge" fusion to significantly improve the initial accuracy of condensation or frost recognition.

[0055] 5. Step S5: Candidate selection and target modeling (size constraints + shape constraints) For each connected component Calculate: Area Roundness ( For perimeter, aspect ratio, and brightness contrast, etc., retain the following: , (Target set excluding fine scratches / edges) .

[0056] 6. Step S6: Confirmation of newly appearing condensation or frost (timing consistency + IoU newness detection + multi-frame confirmation) Maintaining the set of historical goals (Including location, area, and time of occurrence). For the current target... Calculate its maximum intersection-union ratio with historical targets. ,like (e.g., 0.2–0.3), is judged as a "new target candidate". To avoid noise flicker, the target is required to be continuously Only when it appears repeatedly within a frame (e.g., 3–5 frames) is it confirmed as a new dew or frost target and added to the list. To obtain the number of new condensations or frosts that appear at each moment. .

[0057] The advantage of this step is that it avoids false detections due to reflections, flickering, and jitter by using "IoU newness detection combined with multi-frame confirmation".

[0058] 7. Step S7: Dew point determination (quantity threshold, confidence curve consistency with temperature) Constructing the confidence formula for condensation or frost: in: This represents the average value of the condensation probability plot within the ROI. The threshold for the amount of condensation (e.g., 10). A value of 0.6–0.8 is acceptable.

[0059] Configure "Dual Threshold Hysteresis Decision": when And continue If it occurs within 2–5 seconds, then dew point is determined to have occurred. when And continue If so, the dew point status will be deactivated (optional); The dew point temperature is determined as follows: .

[0060] The advantage of this step is that it does not mean "the presence of water droplets in a single frame constitutes the dew point", but rather "the confidence level is continuously satisfied"; the temperature time sequence is introduced into the judgment, so that the detection conforms to the national standard engineering definition of "the temperature at which condensation or frost forms is the dew point".

[0061] 8. Step S8: Evidence Preservation and Report Data Generation Automatically saves data before and after dew point detection is triggered, preferably before triggering. (e.g., 20s) and after triggering Key video clips (e.g., 20s); keyframes (trigger frames, maximum condensation or frost frames); , Temperature curve It is used for traceability and verification.

[0062] IV. The beneficial technical effects of the above-mentioned insulated glass dew point detection, which also includes image recognition and judgment algorithms, are: 1. Anti-jitter: Through inter-frame micro-alignment and multi-frame confirmation, it significantly reduces misjudgments caused by high-frequency jitter of the device.

[0063] 2. Anti-background interference: The background acquisition window establishes a static interference library to automatically eliminate fixed defects such as dust or stains.

[0064] 3. Sensitive to initial condensation or frost: Multi-feature fusion is better suited for recognizing the composite appearance of "highlight, texture and edge" of initial water droplets or mist.

[0065] 4. Results are repeatable: Through confidence hysteresis judgment and temperature consistency verification, transient noise triggering is avoided, and the test results are stable.

[0066] 5. Complete chain of evidence: Automatically saves images and curves before and after the trigger, facilitating the generation of test reports and quality traceability.

[0067] V. Parameter Recommendations for Implementation Background capture window : 8–12s; New target IoU threshold : 0.2–0.3; Background removal IoU threshold : 0.3–0.5; Multi-frame confirmation 3–5 frames; Condensation or frost trigger threshold :10 (can be set after calibration); Confidence threshold: , (Example, can be calibrated).

[0068] The image-based dynamic discrimination method for detecting dew point in insulating glass in this technical solution preferably employs the following intelligent dew point detection device for insulating glass.

[0069] like Figures 5 to 7 As shown, a smart dew point detection device for insulating glass includes: The cooling unit 20 is used to provide controllable low-temperature cooling to the detection area of ​​the insulating glass 10; The cooling surface 21 is disposed at the cooling end of the refrigeration unit 20 and is used to conduct the cold energy generated by the refrigeration unit 20 to the insulating glass and to fit tightly against the surface of the insulating glass. Image acquisition unit 30 is used to continuously acquire images of the detection area; The control and processing unit is used to control the cooling process of the cooling unit 20 and to determine whether the insulated glass has dew point based on the output of the image acquisition unit 30.

[0070] During implementation, the preferred control and processing unit is a tablet terminal (such as...). Figure 8 (as shown) or computer.

[0071] This technical solution's intelligent dew point detection device for insulating glass uses the publicly disclosed "an integral free piston Stirling refrigerator device" (publication number CN116753636A) as the cold source basis to achieve stable and controllable deep cooling of the insulating glass surface temperature. Simultaneously, through image acquisition and dynamic analysis, it continuously monitors and objectively determines the formation process of water vapor on the glass surface, transforming the dew point detection process from traditional manual observation to an automated and repeatable process. This effectively meets the requirements for dew point performance testing of insulating glass in GB / T11944-2012. The refrigeration unit 20 is used to stably output cooling capacity within the detection temperature range of room temperature to -70℃.

[0072] The existing technical solution of "An Integrated Free Piston Stirling Refrigeration Unit" (Announcement No. CN116753636A) includes: an exhaust structure, a power piston structure, and a vibration damping component; the power piston structure includes a power leaf spring, a power bearing, and a power piston arranged sequentially along the axial direction, wherein the power leaf spring and the power bearing are fixedly connected; the exhaust structure includes a first piston, an exhaust leaf spring, and a second piston arranged sequentially along the axial direction, wherein the exhaust leaf spring is connected to the first piston and the second piston through a first fastener, so that when the first piston and the second piston reciprocate under the action of the air pressure difference, the exhaust leaf spring radially supports the first piston and the second piston; this device eliminates the linkage mechanism in the conventional exhaust structure, significantly reduces the exhaust movement damping, and provides hundreds of watts of cooling capacity in the temperature range of -196℃ to -80℃. Since it is existing technology, it will not be described in detail here.

[0073] Compared with existing technologies, the advantages of the intelligent dew point detection device for insulating glass of the present invention are: 1. A controllable low-temperature cooling capacity is provided to the detection area of ​​the insulating glass through the cooling unit 20, and the cooling capacity is efficiently and uniformly transferred to the surface of the insulating glass by means of the cooling surface 21 set at the cooling end of the cooling unit 20, thereby stably constructing the dew point formation conditions without damaging the insulating glass structure. 2. The cold-conducting surface 21 is in close contact with the surface of the insulating glass, which effectively reduces the thermal resistance in the process of cold conduction, improves the local cooling efficiency, and makes the temperature change in the detection area more controllable, which is conducive to improving the consistency and repeatability of dew point detection. 3. The image acquisition unit 30 continuously acquires images of the detection area, which can capture the changes in the condensation or frost state of the insulating glass surface during the cooling process in real time, providing an intuitive and quantifiable image data basis for dew point determination; 4. The control and processing unit coordinates and controls the cooling process of the cooling unit 20, and combines the image information acquired by the image acquisition unit 30 to intelligently determine whether the insulated glass has dew point, thereby realizing the automation and intelligence of the dew point detection process, reducing human judgment errors, and improving detection accuracy and efficiency.

[0074] like Figure 6 and 7 As shown, the insulated glass dew point intelligent detection device also includes a support and positioning mechanism 40, which is used to keep the insulated glass in a fixed state during the detection process; the support and positioning mechanism 40 includes a pair of vertically arranged slot blocks 41, the slot openings of the slots are horizontally aligned; each slot block has at least two threaded holes spaced from top to bottom on the side of the slot, and a threaded top rod 42 for locking the glass is threadedly connected in each threaded hole.

[0075] In practice, it is preferable to provide an elastic rubber pad at the end of the threaded push rod used to press against the glass.

[0076] The beneficial technical effects of this preferred solution are: 1. By setting a pair of slot blocks vertically arranged with their slots facing each other horizontally in the device, the insulating glass can be clamped between the two slot blocks by insertion, thereby achieving rapid positioning and initial support of the insulating glass from a structural point of view, which is convenient for loading and unloading and applicable to insulating glass of different specifications (thickness).

[0077] 2. The slot block has at least two threaded holes spaced from top to bottom on the side of the slot, and a threaded push rod is provided in each threaded hole, so that the insulating glass can be selectively locked at different height positions, thereby adapting to insulating glass of different sizes or different detection position requirements.

[0078] 3. Point-locking of the insulating glass with threaded top rods can ensure clamping stability while avoiding large-area compression of the glass surface, which helps to reduce glass deformation or detection errors caused by clamping stress.

[0079] 4. The support and positioning mechanism 40 has a simple structure and flexible adjustment, which can effectively limit the shaking, displacement or posture change of the insulating glass during the dew point detection process, thereby providing a reliable structural basis for the stable bonding of the cooling surface 21 and the insulating glass and the accurate imaging of the image acquisition unit 30, and improving the stability and repeatability of the dew point detection results.

[0080] like Figure 6 and 7 As shown, the image acquisition unit 30 is detachably mounted on the surface of the insulating glass via a suction cup bracket 31. The suction cup bracket is used to stably attach the image acquisition unit 30 to the insulating glass during the detection process and to adjust and maintain the relative positional relationship between the image acquisition unit 30 and the detection area.

[0081] The advantages of this preferred solution are: 1. By using a suction cup bracket to detachably install the image acquisition unit 30 onto the surface of the insulating glass, the image acquisition unit 30 can be quickly installed and fixed without damaging the insulating glass structure. It is easy to operate and suitable for insulating glass of different sizes and installation positions. 2. During the detection process, the suction cup bracket keeps the image acquisition unit 30 stably attached to the surface of the hollow glass, which can effectively limit the shaking or displacement of the camera and ensure the stability of the relative positional relationship between the image acquisition unit 30 and the detection area, thereby improving the imaging consistency during continuous image acquisition.

[0082] like Figure 6 and 7As shown, the image acquisition unit 30 includes a camera 32, and a ring-shaped LED light is arranged circumferentially at the lens end of the camera 32 to provide uniform illumination to the detection area of ​​the insulating glass.

[0083] The advantages of the above preferred solutions are: 1. A ring-shaped LED light is set around the camera lens to provide close-range and uniform illumination to the detection area, which can effectively reduce the impact of ambient light changes on image acquisition results and improve the brightness uniformity and imaging clarity of the detection area; 2. When condensation or frost occurs on the surface of the insulating glass, the ring LED lighting and the camera work together to highlight the brightness changes, reflection changes, or texture changes in the condensation or frost area, thereby providing a more stable and reliable image data basis for the control and processing unit to determine the dew point based on image information. 3. Through the above structural combination, the dependence of the dew point detection process on ambient light is reduced, and the image acquisition conditions are more controllable, which helps to improve the accuracy and repeatability of the dew point detection results.

[0084] like Figure 6 and 7 As shown, the cooling surface 21 is a metal surface, and the cooling surface 21 constitutes the imaging background of the image acquisition unit 30.

[0085] Therefore, the advantages of this preferred solution are: 1. By setting the cooling surface 21 as a metal surface, it has good thermal conductivity, stable surface morphology, and uniform reflection characteristics, which is conducive to forming a stable optical background during the detection process; 2. The cooling surface 21 forms the imaging background of the image acquisition unit 30 during the detection process, so that the detection area of ​​the insulating glass has relatively uniform and controllable background conditions during imaging, thereby reducing the interference of complex environmental background on image acquisition and dew point identification. 3. When condensation or frost occurs on the surface of the insulating glass, the brightness, reflection or texture changes of the condensation or frost area are more obvious against the background of the metal cooling surface 21, which is conducive to the image acquisition unit 30 accurately capturing the condensation or frost features. 4. By integrating the cooling function and the imaging background function into the same cooling surface 21 structure, the setting of additional background plates or auxiliary structures is reduced, making the overall structure of the device more compact, which is conducive to improving the system integration and detection stability. In summary, the above-mentioned optimized structure, combined with the control and processing unit's analysis of image information, helps to improve the accuracy and repeatability of dew point detection and judgment, and reduces the risk of misjudgment caused by background changes.

[0086] like Figure 6 and Figure 7 As shown, preferably, the cooling end of the cooling unit 20 is provided with a brass block 22, and the outer end face of the brass block 22 facing away from the cooling end is a pure plane and constitutes the cooling surface 21.

[0087] Because the cooling surface 21 of the brass block 22 is a machined cut surface, its surface naturally possesses color and micro-texture characteristics that distinguish it from glass surfaces and other metal materials. This cut surface typically exhibits a stable metallic color and a fine distribution of machining textures. Therefore, this preferred solution offers the following advantages: 1. The brass cut surface has significantly different color and reflective properties compared to the surface of insulated glass. During the imaging process of the image acquisition unit 30, it can form a stable and easily distinguishable visual background, which is beneficial to improving the contrast between the detection area and the background. 2. The fine processing texture formed on the brass cut surface can be represented as a regular and stable texture distribution in the image acquired by the image acquisition unit 30. When condensation or frost occurs on the surface of the hollow glass, the scattering and reflection characteristics of light in the condensation or frost area change, and it is easier to produce identifiable brightness changes or texture blurring features relative to the texture background. 3. By utilizing the inherent color and texture characteristics of the brass cut surface, the image acquisition unit 30 can obtain an imaging background with stable visual characteristics without the need to set up an artificial background or a complex calibration structure, thereby providing a clearer and more reliable image basis for subsequent dew point identification. 4. The aforementioned imaging background characteristics help the control and processing unit to improve the stability and accuracy of dew point identification by analyzing changes in brightness, texture clarity, or reflection when determining dew point, and reduce interference caused by changes in ambient light or the transparency of the glass itself. 5. By combining the cooling function of the brass block 22 with the visual characteristics naturally formed by its cut surface, the cooling surface 21 achieves a synergistic effect between its thermal function and image recognition auxiliary function, further improving the overall detection reliability and practicality of the insulated glass dew point detection device.

[0088] The surface of the cooling surface 21 is also covered with a replaceable flexible cooling adhesive layer (not shown in the figure), which can be any one of a thin thermally conductive silicone pad, thermally conductive gel, or phase change thermally conductive film.

[0089] Because the cooling unit 20 is quite large, and because the high-frequency vibration of the cooling unit 20 during operation is quite noticeable, a flexible heat-conducting layer is provided on the surface of the cooling surface 21 to compensate for the microscopic unevenness of the glass surface and stabilize the heat-conducting contact. The surface of the cooling surface 21 has surface characteristics that enhance the imaging contrast of targets with condensation or frost.

[0090] In practice, the surface of the cooling surface 21 of the replaceable flexible cooling bonding layer preferably has microtexture, which is a plurality of concentric annular grooves with a protrusion of less than 0.5 mm or a groove depth of less than 0.5 mm.

[0091] like Figures 5 to 7 As shown, preferably, the image acquisition unit 30 includes an autofocus mechanism and a manual focus mechanism. The image acquisition unit 30 is mounted on a horizontally arranged support rod 33 via a bracket and can be adjusted in position along the length of the support rod. The manual focus mechanism is used to change the distance between the image acquisition unit 30 and the insulating glass detection area by adjusting the movement position of the image acquisition unit 30 along the support rod, thereby achieving manual focusing. The autofocus mechanism is used to automatically adjust the focal length of the image acquisition unit 30 during image acquisition.

[0092] The advantages of the above preferred solutions are: 1. By simultaneously setting a manual focusing mechanism and an automatic focusing mechanism in the image acquisition unit 30, the image acquisition unit 30 can achieve coarse focusing through structural adjustment and automatic fine focusing during the image acquisition process, thereby adapting to the inspection needs of insulated glass with different installation positions and different specifications. 2. The image acquisition unit 30 is mounted on a horizontally arranged support rod via a bracket, and its position can be adjusted along the length of the support rod, so that the distance between the image acquisition unit 30 and the detection area of ​​the insulating glass can be flexibly adjusted, making it easy to quickly obtain a suitable imaging range before detection; 3. By manually adjusting the position of the image acquisition unit 30 along the support rod direction using the focusing mechanism, initial focusing on the detection area can be achieved, which is beneficial to quickly obtain a clear image before the detection begins and reduce debugging time; 4. The autofocus mechanism automatically adjusts the focal length of the image acquisition unit 30 during image acquisition, which can compensate for imaging deviations caused by environmental vibration, installation errors or minor changes in the glass surface, and improve the imaging stability during continuous image acquisition. The combination of manual and automatic focusing helps improve the clarity and consistency of the image in the detection area, providing a reliable data foundation for subsequent dew point determination based on image information, thereby improving the accuracy and reliability of dew point detection for insulating glass.

[0093] like Figures 5 to 7 As shown, the refrigeration unit 20 is fixedly installed on the upper surface of the guide rail 23 at the top of the bracket, and the lower surface of the guide rail 23 is slidably connected to the slider 24. It also includes a lead screw drive mechanism, which includes a lead screw 25 and a nut that is threadedly engaged with the lead screw. The nut is fixedly connected to the slider 24. By driving the lead screw to rotate, the slider 24 moves along the guide rail 23, thereby causing the cooling end of the cooling unit 20 to move closer to or away from the insulating glass.

[0094] In this way, by fixing the refrigeration unit 20 as a whole to the upper surface of the guide rail 23 at the top of the bracket, and setting a slider 24 that slides with it on the lower surface of the guide rail 23, the refrigeration unit 20 is constrained by the guide rail 23 during movement, and the direction of movement is clear, which helps to ensure the stability of the position adjustment of the refrigeration end of the refrigeration unit 20. By setting a screw drive mechanism consisting of a screw and a nut, and fixing the nut to the slider 24, the rotational movement of the screw can be reliably converted into the linear movement of the slider 24 along the direction of the guide rail 23, thereby realizing the precise advance and retreat adjustment of the refrigeration end of the refrigeration unit 20 relative to the insulating glass. Using the screw drive method to adjust the position of the refrigeration unit 20, the adjustment process is smooth and controllable, which can effectively avoid the impact of rapid displacement of the refrigeration unit 20 on the insulating glass and improve the safety of the testing process.

[0095] As described above, the optimized structure allows the cooling end of the cooling unit 20 to gradually approach or move away from the insulating glass as needed during the testing process. This improves the controllability and repeatability of the cooling bonding process, thereby enhancing the stability and reliability of the insulating glass dew point test results.

[0096] From the above and in conjunction with the appendix Figure 5-7 As can be seen, compared with the existing testing standards that typically involve testing the insulating glass in a flat position, the intelligent dew point detection device for insulating glass of the present invention adopts a vertical support for the insulating glass and an adjustable proximity between the cooling unit 20 and the image acquisition unit 30 in its structural arrangement. This eliminates the reliance on a flat placement for the testing process and simultaneously achieves the following technical effects: 1. This device can perform dew point detection when the insulating glass is in a state close to the actual installation state, which is closer to the stress state and working posture of the insulating glass in the actual use environment of building doors, windows, curtain walls, etc., which helps to improve the reference value of the test results for actual use performance; 2. Through the vertically arranged support structure and the adjustable installation method of the cooling unit 20 and the image acquisition unit 30, the overall structure of the device is compact, occupies little space, and is easy to transport and deploy on site. 3. This device can complete the testing without disassembling and transferring the insulating glass to a dedicated laboratory environment. It can be carried to the site of existing buildings to conduct in-situ or near-in-situ testing on installed or used insulating glass, which significantly improves the flexibility and applicability of the testing. 4. The aforementioned portable testing methods facilitate the rapid implementation of dew point testing in scenarios such as building quality spot checks, existing building performance assessments, or maintenance inspections, reducing manpower, time, and site costs during the testing process.

[0097] The above are merely preferred embodiments of the present invention. It should be noted that any modifications and improvements made by those skilled in the art without departing from the present technical solution should also be considered to fall within the scope of protection claimed in this claim.

Claims

1. A method for detecting dew point of insulating glass based on dynamic image discrimination, characterized in that, Includes the following steps: Step 1: Place the insulating glass to be tested at the testing station, ensuring that the testing area is in contact with the cooling surface; The second step is to control the temperature of the cooling surface to decrease to the target detection temperature zone according to a preset cooling program; during the cooling and constant temperature process, continuous image acquisition is performed on the detection area to obtain an image frame sequence; The third step is to extract candidate dew or frost targets based on the image frame sequence, and to dynamically compare the candidate dew or frost targets at different times. Step 4: When a newly detected condensation or frost target meets the preset judgment conditions, it is determined that the insulated glass has a dew point phenomenon, and the corresponding condensation temperature is recorded.

2. The method of claim 1, wherein: In the initial stage of the cooling process in the second step, a background acquisition step is set up. This step records the image features that are continuously present in the detection area and excludes the image features as background interference in the subsequent detection process.

3. The method of claim 1, wherein: In the third step, the cross-union ratio (CUI) between candidate condensation or frost targets at different time points is calculated. When the CUI is less than a preset threshold, the corresponding candidate condensation or frost target is determined as a newly appearing condensation or frost target.

4. The method of claim 1, wherein: The condensation detection is performed periodically at a preset time interval, which is 0.5-2 seconds.

5. The method of claim 1, wherein: When the number of newly formed condensation or frost targets reaches the first preset threshold, the dew point determination is triggered; when the number of newly formed condensation or frost targets reaches the second preset threshold, the detection process is terminated.

6. The method of claim 1, wherein: In the fourth step: before and after the dew point determination, video data of the detection area is automatically recorded, and the video data is stored in association with the detection temperature and detection time.

7. The method of claim 1, wherein: This is achieved using an intelligent dew point detection device for insulated glass.

8. The method of claim 7, wherein: The intelligent dew point detection device for insulating glass includes a refrigeration unit, a cooling surface, an image acquisition unit, and a control and processing unit.

Images