Foam strip pasting detection system for sheet metal parts

Abstract

The invention relates to the technical field of industrial automatic detection, and discloses a foam strip pasting detection system for sheet metal parts, which comprises a first-stage detection module, a second-stage detection module, a third-stage detection module, a fourth-stage detection module and a fifth-stage detection module, and is characterized in that the first-stage detection module utilizes a visible light image to quickly judge whether a foam strip pasting missing defect exists or not by comparing with a standard template; and if not, accurately generating the dynamic regions of interest of all the foam strips according to the real-time image. And then, the second-stage detection module analyzes a transient thermal image acquired by the thermal imaging detection module after transient thermal excitation only in the dynamic region of interest, and judges the pasting quality by identifying local temperature abnormality caused by false pasting defects such as bubbles or edge upwarp and the like. By means of two-stage series connection and multi-mode detection, the two core defects of missing adhesion and false adhesion can be accurately detected at the same time, full automation of detection is achieved, and the production efficiency and the product quality are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of industrial automation testing technology, specifically to a foam strip adhesion detection system for sheet metal parts. Background Technology

[0002] In modern manufacturing, especially in the automotive industry, foam strips are typically attached to specific locations on sheet metal parts to achieve sealing, vibration reduction, and noise reduction of the vehicle body. The quality of foam strip attachment directly affects the airtightness, watertightness, and NVH (noise, vibration, and harshness) performance of the entire vehicle. Therefore, reliable online testing of the attachment quality of each foam strip is a crucial step in ensuring product quality.

[0003] Currently, there are still many shortcomings in the methods for inspecting the quality of foam strip bonding on production lines. Traditional inspection methods mainly rely on manual visual inspection, which is not only labor-intensive and inefficient, but also highly susceptible to the influence of operator subjectivity and fatigue, leading to frequent missed detections and misjudgments. To overcome the drawbacks of manual inspection, some automated inspection solutions have been proposed, such as using conventional machine vision systems. However, these systems can usually only determine the presence or correct position of the foam strip through image comparison or contour analysis, meaning they can only effectively identify missed bonding defects. For situations where the foam strip exists but there are internal air bubbles or edge lifting between it and the sheet metal substrate, these defects are not obvious in visible light images, rendering conventional machine vision systems ineffective.

[0004] To detect such false bonding defects, some technical solutions have attempted to incorporate thermal imaging technology. However, in practical applications, these solutions often face new challenges. Firstly, if passive thermal imaging is used, the detection results heavily rely on the natural temperature difference between the sheet metal and the environment. In a temperature-controlled production workshop, this temperature difference is often very weak and unstable, easily affected by uncontrollable factors such as airflow and seasonal changes, resulting in weak detection signals and poor reliability. Secondly, some thermal imaging detection systems require global analysis and processing of a large image area containing the sheet metal to ensure accuracy. This processing method is not only algorithmically complex but also computationally burdensome due to the need to process a large amount of background pixel information unrelated to the foam strip, leading to long detection times and making it difficult to match the requirements of high-speed production cycles. Therefore, existing technologies lack a comprehensive solution that can simultaneously, efficiently, and reliably detect both incomplete bonding and false bonding defects. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a foam strip adhesion detection system for sheet metal parts, which solves the problem of accurately, efficiently, reliably, and automatically detecting missed adhesion defects and false adhesion defects of foam strips on sheet metal parts.

[0006] To achieve the above objectives, the present invention provides a foam strip adhesion detection system for sheet metal parts, comprising: Conveyor belt, used to transport sheet metal parts to be inspected; The visible light image acquisition module is used to acquire real-time visible light images of the sheet metal parts moving on the conveyor belt; the thermal imaging detection module is used to acquire transient thermal images of the sheet metal parts. The first-level detection module is connected to the visible light image acquisition module and is used to determine whether there is a missing adhesion defect based on the real-time visible light image, and generate a dynamic region of interest when there is no missing adhesion defect. The second-level detection module is connected to the thermal imaging detection module and the first-level detection module, and is used to analyze the transient thermal image within the dynamic interest region to determine whether there is a false pasting defect. The integrated decision-making and control module, connected to the first-level detection module and the second-level detection module, is used to generate a final judgment instruction based on the judgment results of the missing adhesion defect and the false adhesion defect; The unloading device, connected to the integrated decision and control module, is used to receive the final judgment instruction and remove the sheet metal parts with defects.

[0007] In one specific embodiment, the visible light image acquisition module includes: an industrial camera, disposed above the conveyor belt, for capturing real-time visible light images of the sheet metal part; and a light source assembly, arranged in conjunction with the industrial camera to provide supplementary lighting for the sheet metal part, for providing illumination when the industrial camera captures the real-time visible light image.

[0008] In one specific embodiment, the first-level detection module includes: a missing adhesive detection subunit, used to compare the real-time visible light image with a pre-stored visible light standard template to determine whether there is a missing adhesive defect of missing foam strips; and a region of interest generation subunit, used to segment the contours of all foam strips in the real-time visible light image to generate the dynamic region of interest when the missing adhesive detection subunit determines that there is no missing adhesive defect.

[0009] In one specific embodiment, the second-level detection module is specifically used to: analyze the temperature distribution characteristics of the transient thermal image within the dynamic interest region to identify false adhesion defects caused by edge lifting or internal air bubbles of the foam strip.

[0010] Preferably, when analyzing the temperature distribution characteristics of the transient thermal image, the second-level detection module is specifically used for: Calculate the temperature standard deviation of each foam strip area within the dynamic interest area, and determine the existence of the false pasting defect when the temperature standard deviation is greater than a preset temperature standard deviation threshold; Abnormal temperature patches with temperature values ​​higher than a preset abnormal temperature threshold are detected within the dynamic interest area. When the area of ​​the abnormal temperature patch is greater than a preset minimum defect area threshold, the false pasting defect is determined to exist.

[0011] Specifically, for any independent foam strip region r in the dynamic interest region k The second-level detection module calculates its average temperature μ. k and temperature standard deviation σ k The calculation formula is as follows: Where, μ k σ represents the arithmetic mean temperature of the k-th foam strip region; k r represents the temperature standard deviation of the k-th foam strip region; k Represents the set of pixels in the k-th bubble strip region; |r k | Represents region r k The total number of pixels within the range; ∑ is the summation symbol, and the subscript (x,y)∈r k This indicates that for region r k Iterate through each pixel within T; real (x,y) represents the actual temperature value of the pixel at coordinates (x,y); (·) 2 This represents the squaring operation.

[0012] When σ k If the temperature standard deviation exceeds the preset threshold τσ, a false pasting defect is determined to exist.

[0013] The second-level detection module in region r k Internally identify all temperatures above the abnormal temperature threshold T thresh,k The pixels constitute the abnormal patch set A. k : A k ={(x,y)∈r k |T real (x,y)>T thresh,k}; When set A k Area | A k | Greater than the preset minimum defect area threshold τ A At that time, it was determined that there was a false pasting defect.

[0014] In one specific embodiment, the integrated decision-making and control module includes: The decision logic subunit is used to perform a logical OR operation on the defect judgment results output by the first-level detection module and the second-level detection module; A control signal generation subunit is used to generate a control signal to drive the unloading device to operate when the operation result of the decision logic subunit is true.

[0015] In one specific embodiment, the thermal imaging detection module includes: A transient thermal excitation unit is disposed above the conveyor belt and is used to apply thermal excitation to the sheet metal part; an infrared thermal imaging camera is disposed above the conveyor belt and is used to acquire the transient thermal image after the thermal excitation is applied.

[0016] Preferably, the instantaneous thermal excitation unit is located downstream of the visible light image acquisition module; the infrared thermal imaging camera is located downstream of the instantaneous thermal excitation unit.

[0017] In one specific embodiment, the instantaneous thermal excitation unit is specifically one of a flash lamp array, a hot air jet device, or a cold air jet device.

[0018] In one specific embodiment, the unloading device includes: a slide rail, a slide arm, a push plate, and a pneumatic actuator. The slide rail is installed on the outer side of the conveyor belt, the slide wall is slidably connected to the middle of the slide rail, the pneumatic actuator is fixedly connected to the outer side of the slide rail, the output end of the pneumatic actuator is connected to the slide arm, and the push plate is installed on the side of the slide arm away from the pneumatic actuator, for pushing the sheet metal part off the conveyor belt.

[0019] This invention provides a foam strip adhesion detection system for sheet metal parts. It has the following advantages: 1. This invention constructs a two-stage cascaded detection process by setting up a visible light image acquisition module and a first-stage detection module, and a thermal imaging detection module and a second-stage detection module. The first-stage detection utilizes visible light images to quickly and accurately identify defects such as missing or unattached foam strips. The second-stage detection utilizes active thermal excitation and transient thermal image analysis to accurately identify false adhesion defects caused by edge lifting or internal air bubbles, which are difficult to detect with conventional vision. By combining the two detection modes, it can cover two key defects with different physical properties, achieving comprehensive defect detection and improving the final detection accuracy.

[0020] 2. After confirming the absence of any missing or adhered defects, the first-level detection module of this invention generates a dynamic region of interest (ROI) based on the real-time acquired visible light image for subsequent thermal imaging analysis. The second-level detection module only needs to analyze the temperature distribution characteristics within this ROI, without requiring global processing of the transient thermal image of the entire sheet metal part. This processing method focuses on the target area, eliminating interference from a large amount of irrelevant background information, reducing the computational load of image processing, shortening the time required for a single detection, and improving the overall detection efficiency of the system.

[0021] 3. The thermal imaging detection module of this invention includes an instantaneous thermal excitation unit. By applying active and instantaneous thermal excitation to the sheet metal part, a clearly observable temperature difference is artificially established between the well-attached area and the defective area with gaps. Through active thermal imaging detection, the detection results mainly depend on the excitation energy and the thermophysical properties of the material, effectively overcoming the shortcomings of passive thermal imaging methods that are easily affected by uncontrollable factors such as ambient temperature and the initial temperature of the sheet metal part. This ensures the stability and consistency of the detection signal, making the entire system more reliable and adaptable to different environments. Attached Figure Description

[0022] Figure 1 This is a perspective view of the present invention; Figure 2 This is a schematic diagram of the side structure of the conveyor belt of the present invention; Figure 3 This is a schematic diagram of the light source component structure of the present invention; Figure 4 This is a schematic diagram of the instantaneous thermal excitation unit structure of the present invention; Figure 5 This is a schematic diagram of the infrared thermal imaging camera structure of the present invention; Figure 6 This is a schematic diagram of the unloading device of the present invention; Figure 7 This is a schematic diagram of the foam strip adhesion detection system module of the present invention.

[0023] The components include: 1. Conveyor belt; 2. Industrial camera; 3. Light source assembly; 4. Instantaneous thermal excitation unit; 5. Infrared thermal imaging camera; 6. Sliding arm; 7. Push plate; 8. Slide rail; 9. Processor; and 10. Pneumatic actuator. Detailed Implementation

[0024] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] See attached document Figure 1 - Appendix Figure 7 This invention provides a foam strip adhesion detection system for sheet metal parts. The system includes: a conveyor belt 1, a visible light image acquisition module 100, a thermal imaging detection module 200, a first-level detection module 300, a second-level detection module 400, a comprehensive decision and control module 500, and an unloading device.

[0026] Conveyor belt 1 is used to transport sheet metal parts to be inspected at a preset speed. Visible light image acquisition module 100, thermal imaging detection module 200, and unloading device are arranged sequentially along the conveying direction of conveyor belt 1. The sheet metal parts to be inspected pass through the working areas of each module in sequence under the drive of conveyor belt 1.

[0027] A visible light image acquisition module 100 is located in the initial working section of the conveyor belt 1 and is used to acquire real-time visible light images of the sheet metal parts moving below it. This real-time visible light image is then transmitted to the first-stage detection module 300.

[0028] The thermal imaging detection module 200 is physically located downstream of the visible light image acquisition module 100. This module is used to apply thermal excitation to the sheet metal part and acquire transient thermal images of its surface. These transient thermal images are then transmitted to the second-stage detection module 400.

[0029] The first-level detection module 300 is connected to the visible light image acquisition module 100. This module receives real-time visible light images and determines whether there is a defect where the entire foam strip is missing. If no defect is found, the module further segments and generates dynamic regions of interest for all areas containing the foam strips based on the same real-time visible light image.

[0030] The second-level detection module 400 is data-connected to the thermal imaging detection module 200 and the first-level detection module 300. This module receives transient thermal images and dynamic regions of interest generated by the first-level detection module 300. This module analyzes the transient thermal images only within the dynamic regions of interest to determine whether there are false adhesion defects caused by the edge of the foam strip lifting or the presence of air bubbles inside.

[0031] In one specific implementation, the second-level detection module 400 determines the presence of false pasting defects by calculating the temperature distribution characteristics within the dynamic region of interest. The specific analysis method may include one of the following two: The first method is to calculate the temperature standard deviation. For any independent foam strip region r in the dynamic region of interest... k Calculate its average temperature μ k With temperature standard deviation σ k .

[0032] in: μ k This represents the arithmetic mean temperature of the k-th foam strip region; σ k This represents the temperature standard deviation of the k-th foam strip region; r k This represents the set of pixels in the k-th bubble strip region; |r k | Represents region r k The total number of pixels within; ∑ is the summation symbol, and the subscript (x,y)∈r k This indicates that for region r k Iterate through each pixel within the array; T real (x,y) represents the actual temperature value of the pixel at coordinates (x,y); (·) 2 This represents the squaring operation.

[0033] When the calculated temperature standard deviation σ k If the temperature exceeds a preset standard deviation threshold, the area is determined to have a false pasting defect.

[0034] The second method involves detecting anomalous temperature patches. Within a dynamic region of interest, all pixels with temperatures exceeding a preset anomalous temperature threshold are detected; these pixels constitute anomalous temperature patches. When the area of ​​an anomalous temperature patch exceeds a preset minimum defect area threshold, a false pasting defect is identified.

[0035] The integrated decision-making and control module 500 is data-connected to the first-level detection module 300 and the second-level detection module 400. This module receives defect judgment results from the two detection modules. When the judgment result of either detection module is that a defect exists, this module generates and sends an unloading control command.

[0036] The unloading device is signal-connected to the integrated decision and control module 500. Upon receiving an unloading control command, this device executes mechanical actions to remove defective sheet metal parts from conveyor belt 1, thereby achieving automatic sorting of non-conforming products. Conforming sheet metal parts continue to be conveyed along conveyor belt 1 to the next workstation. The functions of the first-level detection module 300, the second-level detection module 400, and the integrated decision and control module 500 can be implemented by one or more processors 9 executing pre-stored computer programs.

[0037] See attached document Figure 3 and attached Figure 7In one specific embodiment of the present invention, the visible light image acquisition module 100 has the following structure and operation. The function of this module is to acquire high-quality, distortion-free real-time visible light images of the surface of the sheet metal part to be inspected, providing raw data for the subsequent first-stage detection module 300.

[0038] The visible light image acquisition module 100 includes an industrial camera 2 and a light source assembly 3. The industrial camera 2 is a high-resolution area array industrial camera, fixedly mounted above the conveyor belt 1, with its lens optical axis perpendicular to the surface of the sheet metal part transported by the conveyor belt 1. This vertical mounting ensures that the acquired image has minimal perspective distortion and accurately reflects the size and position of the foam strip. The industrial camera 2 is fixed by a rigid bracket to eliminate the impact of vibrations in the production environment on image quality.

[0039] To ensure image acquisition is performed at the same location on the sheet metal part each time, the system also includes a trigger sensor located upstream of the visible light image acquisition module 100. This trigger sensor, such as a photoelectric sensor, detects the leading edge of the sheet metal part. When the leading edge of the sheet metal part passes the trigger sensor, the sensor generates a trigger signal. This trigger signal is sent to the industrial camera 2 after a settable time delay, thereby precisely controlling the camera to perform a single image acquisition action when the sheet metal part is directly below the lens. The length of this time delay is calibrated based on the speed of the conveyor belt 1 and the distance between the trigger sensor and the industrial camera 2.

[0040] The light source assembly 3 is deployed in conjunction with the industrial camera 2. Its function is to provide sufficiently intense and uniformly distributed illumination to the surface of the sheet metal part at the moment of exposure by the industrial camera 2. In one specific embodiment, the light source assembly 3 is a large-area white LED flat panel diffuse light source, which is located around or coaxially with the industrial camera 2. This high-angle diffuse lighting method allows light to evenly illuminate the surface of the sheet metal part from multiple angles, minimizing the high-gloss reflection generated by the metal surface of the sheet metal part, while eliminating shadows generated by structural components, thereby forming a clear grayscale contrast between the foam strip and the metal substrate in the final image.

[0041] The operating mode of the light source assembly 3 is strictly synchronized with the acquisition action of the industrial camera 2. The light source assembly 3 operates in strobe mode, meaning it only illuminates for a brief period during camera exposure when it receives the same trigger signal as the industrial camera 2. This synchronized strobe operation not only effectively "freezes" moving sheet metal parts, avoiding motion blur, but also achieves a better imaging signal-to-noise ratio with instantaneous high brightness and extends the lifespan of the LED light source. The digital images acquired by the industrial camera 2 are transmitted in real-time to the first-level detection module 300 for processing via a data interface.

[0042] See attached document Figure 4 Appendix Figure 5 and attached Figure 7 In one specific embodiment of the present invention, a thermal imaging detection module 200 is arranged downstream of the visible light image acquisition module 100 along the conveyor belt 1. The function of this module is to first apply a controlled, instantaneous thermal excitation to a sheet metal part passing beneath it, and then, at a precise time point after the excitation, acquire a transient thermal image of the sheet metal part's surface. This transient thermal image contains temperature distribution information reflecting the adhesion quality between the foam strip and the metal substrate.

[0043] The thermal imaging detection module 200 includes an instantaneous thermal excitation unit 4 and an infrared thermal imaging camera 5. The instantaneous thermal excitation unit 4 and the infrared thermal imaging camera 5 are arranged sequentially along the conveyor belt 1. The function of the instantaneous thermal excitation unit 4 is to uniformly inject or remove heat into the foam strip area on the surface of the sheet metal part in a very short time to establish a measurable temperature difference. When the foam strip is well bonded to the sheet metal substrate, heat can be quickly conducted between the two; conversely, if there are false bonding defects such as edge lifting or internal air bubbles, the air layer at the defect will form a thermal barrier layer, hindering the normal conduction of heat, thereby forming a local temperature anomaly area on the surface of the sheet metal part.

[0044] In one specific embodiment, the instantaneous thermal excitation unit 4 can be a flash lamp array. This array consists of one or more high-power xenon flash lamps, capable of releasing a high-energy light pulse within milliseconds. The energy of this light pulse is absorbed by the surface of the foam strip and converted into heat energy, achieving rapid heating.

[0045] In another specific embodiment, the instantaneous thermal excitation unit 4 can be a hot air jetting device. This device includes a set of nozzles capable of jetting hot air at a constant temperature and flow rate to non-contactly heat the surface of the sheet metal part. In this manner, the intensity and duration of the thermal excitation can be precisely adjusted by controlling the temperature, flow rate, and jetting duration of the hot air.

[0046] In another specific embodiment, the instantaneous thermal excitation unit 4 can also be a cold air jet device. This device rapidly cools the surface of sheet metal parts that are at ambient temperature or have been preheated by jetting low-temperature gas (e.g., cold airflow generated by a vortex tube). Under this excitation method, areas with good adhesion cool down faster due to smooth heat conduction, while areas with defects cool down slower due to obstructed heat conduction, appearing as relative "hot spots" on the thermal image.

[0047] An infrared thermal imaging camera 5, positioned downstream of the instantaneous thermal excitation unit 4, is used to acquire the infrared radiation emitted from the surface of the sheet metal part and convert it into a transient thermal image composed of a temperature value matrix. Specifically, this camera is a long-wave infrared (LWIR) camera, whose spectral response range matches the peak wavelength of thermal radiation from objects at room temperature. The camera's trigger acquisition action is precisely synchronized with the excitation action of the instantaneous thermal excitation unit 4.

[0048] Specifically, after the leading edge of the sheet metal part triggers the upstream trigger sensor, a synchronous controller, based on the speed of conveyor belt 1 and the distance between units, first triggers the instantaneous thermal excitation unit 4 to perform an excitation action when the sheet metal part moves to the excitation position. Then, after a preset, extremely short time delay (e.g., tens to hundreds of milliseconds), when the sheet metal part moves directly below the infrared thermal imaging camera 5, the camera is triggered to complete an image acquisition. This time delay is crucial; it ensures that when the camera acquires the image, the surface temperature difference caused by the defect has fully formed but has not yet become blurred due to the lateral diffusion of heat to the surrounding area. The acquired transient thermal image (i.e., temperature matrix data) is transmitted in real time to the second-level detection module 400 via the data interface for subsequent defect analysis.

[0049] See attached document Figure 1 Attached Figure 2 and attached Figure 7 In one specific embodiment of the present invention, the first-level detection module 300 is configured and operates as follows. This module is typically implemented by an industrial computer or embedded processor 9, which receives real-time visible light images transmitted by the visible light image acquisition module 100 and processes them to perform two core functions: determining whether there are any missing or bonded defects, and generating dynamic regions of interest when there are no missing or bonded defects.

[0050] When a real-time visible light image is input into the first-level detection module 300, the missing adhesive detection subunit within this module starts working. This subunit compares the input real-time image with a visible light standard template pre-stored in the system memory. This visible light standard template is a reference image of a known qualified sheet metal part acquired under the same lighting and shooting conditions; it defines the required quantity, position, and contour of all foam strips.

[0051] In one implementation, the comparison process is achieved through image registration and image subtraction algorithms. First, the system uses preset marker points on the sheet metal part or its inherent contour features to perform coordinate transformation on the real-time image, aligning it with the coordinate system of the standard template.

[0052] After alignment, the grayscale matrix of the real-time image is subtracted from the grayscale matrix of the standard template to generate a difference image. In areas where the foam strip is missing, a high-intensity pixel region will appear at the corresponding position in the difference image due to the grayscale difference between the background and the foam strip. By setting a brightness threshold for this difference image, the presence of a missing foam strip can be detected. If such a high-intensity region is detected, a missing foam strip defect is determined to exist, and a defect signal is generated and sent to the integrated decision and control module 500.

[0053] If the missing adhesive detection subunit does not detect any missing adhesive defects, the control flow switches to the Region of Interest (ROI) generation subunit within the module. This subunit's function is to precisely segment the actual contours of all foam strips using the current real-time visible light image, which has been confirmed to be free of missing adhesive. This process ensures that subsequent analysis is based on the real-time, actual position of the sheet metal part on conveyor belt 1, rather than the theoretical position of the template; therefore, the generated region is "dynamic."

[0054] In one specific implementation, the generation of regions of interest (ROIs) is achieved through the following steps: First, an adaptive thresholding algorithm is applied to the real-time image to separate the foam strip regions with significant grayscale differences from the metallic background, forming a binary image. Then, morphological operations, such as erosion followed by dilation (opening), are performed on the binary image to eliminate noise points and burrs caused by uneven illumination or minor reflections, resulting in clear foam strip region blocks. Finally, a contour discovery algorithm is applied to the processed binary image to generate a closed polygon or its circumscribed rectangle describing the boundary of each independent connected region (i.e., each foam strip).

[0055] The coordinates of these polygons or rectangles constitute the dynamic region of interest provided to the second-level detection module 400. The first-level detection module 300 sends this set of coordinate data to the second-level detection module 400 via an internal data bus to define the range for subsequent thermal imaging analysis. Afterward, the first-level detection module 300 completes the current detection cycle and waits to process the image of the next sheet metal part.

[0056] See attached document Figure 1 Attached Figure 2 and attached Figure 7 In one specific embodiment of the present invention, the second-level detection module 400 is configured and operates as follows. This module is typically implemented by an industrial computer or embedded processor 9, and its data input terminal is connected to the thermal imaging detection module 200 and the first-level detection module 300. The function of this module is to analyze the transient thermal image acquired by the thermal imaging detection module 200 within the dynamic region of interest provided by the first-level detection module 300 to determine whether a false pasting defect exists.

[0057] The physics behind this detection lies in the fact that the thermal conductivity between the foam strip and the sheet metal substrate directly reflects the bonding quality. In well-bonded areas, the two are in close contact, and the heat generated by instantaneous thermal excitation can be rapidly conducted between them. However, in areas with pseudo-bonding defects such as edge lifting or internal air bubbles, a tiny air layer exists between the foam strip and the substrate. Since the thermal conductivity of air is much lower than that of the foam strip and metal materials, this air layer constitutes a thermal insulation layer, significantly hindering normal heat conduction. Therefore, after instantaneous thermal excitation, the surface temperature change rate in the defective area will be significantly slower than in the intact area, thus forming a localized, measurable temperature anomaly on the transient thermal image.

[0058] After the second-level detection module 400 receives the transient thermal image and the coordinate data of the dynamic region of interest, it only processes the pixel temperature values ​​within that region of interest. In one specific implementation, this module employs an analysis method based on the temperature standard deviation, and its execution steps are as follows: For each independent dynamic region of interest r sent by the first-level detection module 300 k The module first extracts the temperature value T of all pixels in the region. real (x, y). Then, the arithmetic mean temperature μ of this region is calculated using the following formula. k and temperature standard deviation σ k : The calculated σ k It is compared with a preset temperature standard deviation threshold τσ. This threshold τσ is determined based on statistical analysis of experimental test data from a large number of known qualified and defective samples. If σ k >τ σ If so, it is determined that there is a false pasting defect in that area.

[0059] In another specific implementation, this module employs an analysis method based on anomalous temperature patches. For the dynamic region of interest r... k The module first determines an abnormal temperature threshold T. thresh,k This threshold can be a fixed empirical value, or it can be dynamically calculated to improve adaptability, for example, T. thresh,k It can be defined as the average temperature μ of this region. k Add a reference standard deviation that is a multiple of n, i.e.: T threshk =μ k +n·σ ref ; Where; σ ref It is a reference temperature standard deviation measured from a known qualified sample.

[0060] After determining the threshold, the module will apply the following formula to region r. k All pixels with temperatures exceeding this threshold constitute an abnormal pixel set AK. AK={(x,y)|(x,y)∈r k ∧T real (x,y)>T thresh,k}; Where: ∧ represents the logical OR operation.

[0061] This formula means that the set AK consists of all pixel coordinates (x, y) that simultaneously satisfy two conditions; this pixel belongs to the region of interest r. k Furthermore, its actual temperature value is greater than the abnormal temperature threshold of the region.

[0062] The module then applies a connected component labeling algorithm to the anomalous pixel set AK. This algorithm groups all spatially adjacent pixels in set AK (e.g., connected by four or eight neighbors) into one or more independent subsets. Each such subset is defined as an anomalous temperature patch.

[0063] The module then calculates the area of ​​each abnormal temperature patch, which is the total number of pixels contained in that patch (subset). Finally, the module compares the area of ​​each patch with a preset minimum defect area threshold τ. A Compare the threshold τ. A Its function is to filter out tiny temperature spots that are not indicative of defects and are caused by image noise or non-critical minor flaws. This threshold τ A The value is determined based on the minimum acceptable defect size in actual production. If region r k The area of ​​any abnormal temperature patch is greater than τ. A If so, it is determined that there is a false pasting defect in that area.

[0064] Regardless of the analysis method used, once the second-level detection module 400 determines the presence of a false pasting defect, it will immediately generate a defect signal and send it to the integrated decision and control module 500. If no defect is found within a detection cycle, no signal will be sent, or a signal indicating acceptance will be sent.

[0065] See attached document Figure 7 In one specific embodiment of the present invention, the integrated decision-making and control module 500 is configured and operates as follows. The physical entity of this module can be a programmable logic controller (PLC) or a processing unit integrated into an industrial computer. Its function is to receive the judgment results from the first-level detection module 300 and the second-level detection module 400, and generate and send control commands for driving the unloading device based on these results.

[0066] The data input end of the comprehensive decision-making and control module 500 is connected to the signal output ends of the first-level detection module 300 and the second-level detection module 400. For the same sheet metal part to be tested conveyed along the conveyor belt 1, this module will receive two independent determination signals: one from the first-level detection module 300, used to indicate whether there is a missing adhesion defect; the other from the second-level detection module 400, used to indicate whether there is a false adhesion defect. These two signals can be digital high / low level signals, or boolean flag bits in data packets transmitted through the industrial Ethernet protocol.

[0067] The decision logic sub-unit inside the module is responsible for merging and processing the received signals. This sub-unit performs the following logical judgment for each sheet metal part to be tested: when the determination signal of the first-level detection module 300 indicates a defect, or when the determination signal of the second-level detection module 400 indicates a defect, then the final detection result of this sheet metal part is determined to be defective. Only when the determination signals of both detection modules indicate qualified, the final detection result of this sheet metal part is determined to be qualified.

[0068] After the final detection result is determined to be defective, the control signal generation sub-unit inside the module starts to work. The function of this sub-unit is to convert this logical "defect" determination into a physical electrical signal with precise timing. To achieve precise control, this module obtains the running speed of the conveyor belt 1 from the system and pre-stores the physical distance between the detection completion position and the working position of the unloading device.

[0069] Based on the above speed and distance parameters, the control signal generation sub-unit calculates a time delay. At the moment when the defect determination is made, start timing this time delay. When the timing ends, it indicates that the defective sheet metal part has precisely moved to the front of the unloading device. At this time, the control signal generation sub-unit generates a valid unloading control instruction on one of its digital output ports. This instruction can specifically be a DC voltage signal that jumps from the default 0V low level to 24V high level and maintains this high level for a preset pulse width. This pulse width is set to be sufficient to ensure that the actuator (such as a solenoid valve) in the unloading device can complete a full driving action.

[0070] After sending a unloading control instruction once, this digital output port returns to the default state of 0V low level, waiting to process the detection results of the next sheet metal part. If a sheet metal part is determined to be qualified, the control signal generation sub-unit does not produce any action, and its digital output port maintains the default low level state.

[0071] Refer to Appendix Figure 6 and Appendix Figure 7In one specific embodiment of the present invention, the structure and operation of the unloading device are as follows. The device is an electromechanical actuator whose function is to perform physical actions after receiving the unloading control command from the integrated decision and control module 500 to remove the sheet metal parts that have been determined to have defects from the main conveyor belt 1.

[0072] The unloading device is located on one side of the conveyor belt 1, at the end of the entire inspection process. In one specific embodiment, the device includes a pneumatic actuator 10, a sliding arm 6, a pusher plate 7, and a slide rail 8.

[0073] The pneumatic actuator 10 includes a linear reciprocating cylinder, a solenoid valve connected to the cylinder, and a pipeline for supplying compressed air. The cylinder is horizontally mounted, with its piston rod extending perpendicular to the conveying direction of the conveyor belt 1. The electrical signal input terminal of the solenoid valve is connected to the control command output terminal of the integrated decision and control module 500. The function of the solenoid valve is to convert the input electrical signal (i.e., the unloading control command) into a switching action of the air path inside the cylinder, thereby controlling the extension and retraction of the piston rod.

[0074] The pusher plate 7 is a plate-shaped structural component with sufficient rigidity and contact area, which is fixedly installed at the end of the slide arm 6 away from the cylinder piston rod. In the initial state, the pushing surface of the pusher plate 7 is located outside the effective conveying width of the conveyor belt 1, and the pushing surface is parallel to the conveying direction of the conveyor belt 1.

[0075] The device operates as follows: In the initial state, the solenoid valve is not energized, and the piston rod of the cylinder is in the fully retracted position. When the integrated decision and control module 500 issues a valid unloading control command (e.g., a 24V high-level signal), this signal is applied to the solenoid valve, energizing its coil. After being energized, the solenoid valve switches its valve core position, introducing compressed air from an external air source into the rodless chamber of the cylinder, driving the piston rod to extend rapidly.

[0076] The extension of the piston rod causes the sliding arm 6 to slide in the middle of the slide rail 8, which in turn causes the slide rail 8 to move the push plate 7 laterally, applying a thrust to the side of the defective sheet metal part that is just in front of it. This thrust pushes the sheet metal part off the conveyor belt 1, causing it to leave the main conveying path and fall into a preset defective product collection area (e.g., a collection box or a separate defective product conveyor belt 1).

[0077] When the pulse of the unloading control command issued by the integrated decision and control module 500 ends (e.g., the high-level signal disappears), the solenoid valve coil is de-energized, and its valve core returns to its initial position under the action of the return spring. At this time, the gas in the rodless chamber of the cylinder is discharged, and the piston rod retracts under the action of the air pressure on the other side or its built-in return spring. The retraction of the piston rod drives the push plate 7 back to its initial waiting position, completing a complete unloading action and preparing for the next possible action.

Claims

1. A foam strip adhesion detection system for sheet metal parts, characterized in that, include: Conveyor belt (1) is used to transport sheet metal parts to be inspected; Visible light image acquisition module, used to acquire real-time visible light images of sheet metal parts moving on the conveyor belt (1); A thermal imaging detection module is used to acquire transient thermal images of the sheet metal part; Processor (9), which is mounted on the outside of the conveyor belt (1) and configured as follows: The first-level detection module determines whether there is a missing adhesion defect based on the real-time visible light image, and generates a dynamic region of interest when there is no missing adhesion defect. The second-level detection module is used to analyze the transient thermal image within the dynamic interest region to determine whether there is a false pasting defect. The integrated decision-making and control module is used to generate the final judgment instruction based on the judgment results of missing adhesion defects and false adhesion defects; The unloading device is used to receive the final judgment instruction and remove the sheet metal parts with defects.

2. The foam strip adhesion detection system for sheet metal parts according to claim 1, characterized in that, The visible light image acquisition module includes: An industrial camera (2) is positioned above the conveyor belt (1) and is used to capture real-time visible light images of the sheet metal parts. The light source assembly (3) is located above the conveyor belt (1) and is used to cooperate with the industrial camera (2) to provide supplementary lighting for the sheet metal parts and to provide illumination when the industrial camera (2) captures the real-time visible light image.

3. The foam strip adhesion detection system for sheet metal parts according to claim 1, characterized in that, The first-level detection module includes: The missing adhesive detection subunit is used to compare the real-time visible light image with a pre-stored visible light standard template to determine whether there is a missing foam strip defect. The region of interest generation subunit is used to segment the outlines of all foam strips in the real-time visible light image and generate the dynamic region of interest when the leak detection subunit determines that there is no leak defect.

4. The foam strip adhesion detection system for sheet metal parts according to claim 1, characterized in that, The second-level detection module is specifically used for: Within the dynamic region of interest, the temperature distribution characteristics of the transient thermal image are analyzed to identify false adhesion defects caused by edge lifting of the foam strip and internal air bubbles.

5. The foam strip adhesion detection system for sheet metal parts according to claim 4, characterized in that, The analysis of the temperature distribution characteristics of the transient thermal image includes: Calculate the temperature standard deviation of each foam strip area within the dynamic interest area, and determine the existence of the false pasting defect when the temperature standard deviation is greater than a preset temperature standard deviation threshold; The abnormal temperature patch with a temperature value higher than a preset abnormal temperature threshold is detected within the dynamic interest area, and the false pasting defect is determined to exist when the area of ​​the abnormal temperature patch is greater than a preset minimum defect area threshold.

6. The foam strip adhesion detection system for sheet metal parts according to claim 1, characterized in that, The integrated decision-making and control module includes: The decision logic subunit is used to perform a logical OR operation on the defect judgment results output by the first-level detection module and the second-level detection module; A control signal generation subunit is used to generate a control signal to drive the unloading device to operate when the operation result of the decision logic subunit is true.

7. The foam strip adhesion detection system for sheet metal parts according to claim 1, characterized in that, The thermal imaging detection module includes: Instantaneous thermal excitation unit (4) is disposed above the conveyor belt (1) and is used to apply thermal excitation to the sheet metal part; An infrared thermal imaging camera (5) is positioned above the conveyor belt (1) to acquire the transient thermal image after thermal excitation is applied.

8. A foam strip adhesion detection system for sheet metal parts according to claim 7, characterized in that, The instantaneous thermal excitation unit (4) is specifically: One of a flash array, a hot air jet device, or a cold air jet device.

9. A foam strip adhesion detection system for sheet metal parts according to claim 1, characterized in that, The unloading device includes a slide rail (8), a slide arm (6), a push plate (7), and a pneumatic actuator (10). The slide rail (8) is installed on the outside of the conveyor belt (1). The slide wall (6) is slidably connected to the middle of the slide rail (8). The pneumatic actuator (10) is fixedly connected to the outside of the slide rail (8). The output end of the pneumatic actuator (10) is connected to the slide arm (6). The push plate (7) is installed on the side of the slide arm (6) away from the pneumatic actuator (10) and is used to push the sheet metal part off the conveyor belt (1).

10. A foam strip adhesion detection system for sheet metal parts according to claim 7, characterized in that, The instantaneous thermal excitation unit (4) is located downstream of the visible light image acquisition module, and the infrared thermal imaging camera (5) is located downstream of the instantaneous thermal excitation unit (4).

Images