A real-time detection method and system for infrared thermal imaging defects of wood composite fiberboard

Abstract

The present application relates to the technical field of image processing, and discloses a kind of wood composite fiberboard infrared thermal imaging defect real-time detection method and system, wherein, method includes: constructing board detection session dataset;Perform speed constraint exposure control and non-uniformity correction;Extract board surface area and perform board coordinate mapping;Background temperature field, temperature residual field and self-reference residual are constructed;Extract candidate defect region set;Perform threshold segmentation, connected domain screening, persistence check and internal delamination defect determination;Output board internal delamination defect detection result.Compared with prior art mainly relying on static thermal map contrast or single-frame temperature difference threshold discrimination, especially under the condition that heat distribution is uneven after hot pressing and board is continuously conveyed, the technical problem that stable identification of internal delamination defect area cannot be realized.Due to the present application, the accuracy and real-time performance of internal delamination defect online detection are improved by correlating conveying displacement information with infrared thermal image sequence.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method and system for real-time detection of defects in wood composite fiberboard using infrared thermal imaging. Background Technology

[0002] Currently, there are still significant shortcomings in the online quality inspection of wood-based composite fiberboard after hot pressing. For example, in the continuous hot pressing stage, when the board is transported from the hot press outlet to the cooling section, the overall surface temperature is high. Furthermore, different areas are affected by the hot pressing temperature distribution, the uniformity of adhesive application, fiber density, and local heat dissipation conditions, often resulting in a complex and non-uniform residual heat distribution. Under these circumstances, if there are defects such as uneven adhesive distribution, insufficient curing, or localized bonding failure leading to delamination within the board, these defects typically only appear as weak local thermal response differences in infrared thermography, easily masked by the high-temperature background and temperature fluctuations, making accurate identification difficult with existing detection methods.

[0003] Furthermore, existing infrared detection solutions mostly employ single-frame thermal image analysis under static conditions, or rely solely on fixed temperature difference thresholds to segment and determine thermal anomaly areas. While these methods can achieve preliminary detection under conditions where the board is stationary, the background temperature is relatively stable, and the surface temperature distribution is relatively uniform, in the continuous online conveying of wood-based composite fiberboard, the board exhibits significant displacement changes and speed fluctuations. This easily leads to ghosting, blurring, and edge distortion during infrared imaging. Simultaneously, influenced by ambient temperature, equipment radiant heat, roller reflective heat, and board edge heat dissipation, strong background drift and localized false hot spots are superimposed on the infrared image. This makes it difficult to effectively distinguish internal delamination defects from temperature fluctuations in normal areas using methods relying solely on single-frame grayscale differences or fixed thresholds.

[0004] Therefore, there is an urgent need for a method that can continuously extract and determine the degumming defect area inside the board even when the board surface is under high temperature background after hot pressing, the board is continuously and dynamically transported and there is temperature drift and motion ambiguity, so as to improve the accuracy, real-time performance and stability of online quality inspection of wood composite fiberboard. Summary of the Invention

[0005] To address the aforementioned technical shortcomings, the present invention aims to propose a real-time infrared thermal imaging defect detection method for wood-based composite fiberboard. This method addresses the technical problem that existing technologies primarily rely on static thermal image comparison or single-frame temperature difference threshold discrimination, particularly under conditions of uneven residual heat distribution after hot pressing and continuous board transport, which makes it impossible to stably identify internal degumming defect areas.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The present invention provides a method for real-time detection of defects in wood composite fiberboard using infrared thermal imaging.

[0007] The real-time infrared thermal imaging defect detection method for wood-based composite fiberboard includes: Step S10: Obtain the original infrared thermal image sequence, timestamp sequence, and conveying displacement pulse sequence of the wood composite fiberboard in the continuous hot pressing station. Based on the original infrared thermal image sequence, timestamp sequence, and conveying displacement pulse sequence, use the board detection session construction mechanism to perform the online detection session establishment task and output the board detection session dataset. Step S20: Based on the plate inspection session dataset, a velocity-constrained exposure and radiometric correction mechanism is used to perform thermal image correction tasks, and the corrected thermal image temperature sequence is output; Step S30: Based on the corrected thermal image temperature sequence, the plate coordinate mapping mechanism is used to perform the temperature time series field reconstruction task, and the plate coordinate system temperature time series field is output; Step S40: Based on the temperature time series field of the plate coordinate system, a self-reference cooling feature fusion mechanism is used to perform the defect candidate region extraction task and output the candidate defect region set; Step S50: Based on the candidate defect region set, perform connected component screening, continuity verification, and internal debonding defect determination, and output the detection results of internal debonding defects in the board.

[0008] Preferably, step S10, which involves acquiring the original infrared thermal image sequence, timestamp sequence, and conveying displacement pulse sequence of the wood-based composite fiberboard in the continuous hot-pressing station, and then using a board inspection session construction mechanism to establish an online inspection session based on the original infrared thermal image sequence, timestamp sequence, and conveying displacement pulse sequence, and outputting the board inspection session dataset, specifically includes: Step S101: An infrared thermal imaging acquisition unit and a conveyor displacement acquisition unit are installed between the hot press outlet and the cooling section. Based on the infrared thermal imaging acquisition unit and the conveyor displacement acquisition unit, raw infrared thermal image sequences are continuously acquired. Timestamp sequence and the transmission displacement pulse sequence ;in, This indicates the sequence number of the nth sampling frame during continuous acquisition. Step S102: Obtain the effective pixels on the board surface within the current acquisition area, and determine the percentage of effective pixels on the board surface within the current acquisition area. Establish entry criteria for the board material and the percentage of effective pixels on the board surface. satisfy: ;in, This indicates the number of pixels in the currently connected region of the front panel. Indicates the total number of pixels in the acquisition area; when At that time, establish the testing session identifier corresponding to the current board material. ; Step S103: Detect session identifier As an index, the original infrared thermal image sequence Timestamp sequence and the transmission displacement pulse sequence The data is stored in association to form a panel inspection session dataset.

[0009] Preferably, step S20, which involves performing a thermal image correction task based on the plate inspection session dataset using a velocity-constrained exposure and radiometric correction mechanism, and outputting a corrected thermal image temperature sequence, specifically includes: Step S201: Based on the transmission displacement pulse sequence Calculate the plate displacement sequence And based on the plate displacement sequence Calculate the conveying speed by the displacement change between adjacent sampling times. ;in, , This refers to the encoder single-pulse displacement calibration coefficient; Step S202: Obtain the pixel size in the transport direction and the upper limit of target fuzziness Combined with conveying speed A dynamic exposure control method based on Nyquist sampling constraints is used to perform velocity-constrained exposure control on the original infrared thermal image frames. Non-uniformity correction and bad pixel repair are then performed on the original infrared thermal image frames to obtain a corrected thermal image sequence. ; Step S203: Based on the temperature of the reflective reference region and the high emissivity reference region measured by the preset reference temperature acquisition unit, the thermal image sequence is calibrated. The Planck radiation inversion method was used to perform apparent temperature conversion, emissivity estimation, and reflectance temperature compensation to obtain the corrected thermal image temperature sequence. .

[0010] Preferably, in step S202, speed-constrained exposure control is performed on the original infrared thermal image frame, and non-uniformity correction and bad pixel repair are performed on the original infrared thermal image frame to obtain a corrected thermal image sequence. The steps are as follows: Step S2021: Based on the conveying speed Calculate the maximum allowable exposure time ;

[0011] and actual exposure time Set to:

[0012] in, This is the default exposure duration; Step S2022: After completing exposure control, perform non-uniformity correction on the original infrared thermal image frame using a combination of Wiener enhancement and column mean offset correction. The correction expression is:

[0013] Where x represents the column coordinate in the thermal image frame; y represents the row coordinate in the thermal image frame; and n represents the sampling frame number. Indicates the first The first frame Column pixel mean Indicates the first The global pixel mean of a frame; Step S2023: and according to the conveying speed Actual exposure time and the pixel size in the transport direction Calculate fuzzy length :

[0014] when At that time, the Richardson-Lucy one-dimensional deconvolution compensation method was used to perform deblurring along the conveying direction to obtain a deblurred temperature image; Step S2024: Use the Kalman recursive repair method to repair bad pixels in the deblurred temperature image to obtain the corrected thermal image sequence. .

[0015] Preferably, step S30, which involves performing a temperature time-series reconstruction task based on the corrected thermal image temperature sequence using a plate coordinate mapping mechanism and outputting the temperature time-series field of the plate coordinate system, specifically includes: Step S301: Extract the mask of the plate surface area based on the corrected thermal image temperature sequence And based on the mask of the plate area Temperature data within the current plate area is extracted using a combination of Otsu threshold segmentation and connected component filtering. Step S302: Combine with plate displacement and the pixel size in the delivery direction Map each pixel within the current frame's board surface area to the board coordinate system. The mapping relationship is as follows: ; in, This represents the pixel coordinates of a pixel in the current frame along the transport direction. This represents the vertical physical coordinates of the board material corresponding to the pixel; and the pixel coordinates of the pixel in the previous frame perpendicular to the conveying direction. According to pixel size Mapped to the horizontal physical coordinates of the board ,satisfy: ; in, This represents the pixel coordinates of a pixel in the current frame that are perpendicular to the transport direction. Indicates the horizontal pixel size; Step S303: Using the longitudinal physical coordinates of the board material and the horizontal physical coordinates of the board As an index, the temperature maps of consecutive moments within the current previous frame's board surface area are resampled to construct a temperature time series field in the board coordinate system. Where t represents the sampling time or the time index after resampling corresponding to the coordinates of the plate.

[0016] Preferably, step S40, which involves using a self-reference cooling feature fusion mechanism based on the temperature time-series field of the plate coordinate system to perform the defect candidate region extraction task and output a set of candidate defect regions, specifically includes: Step S401: Based on the temperature field and time sequence field of the plate coordinate system Constructing the background temperature field And calculate the temperature residual field. ,in:

[0017]

[0018] in, The constant term coefficients representing the background temperature field. Represents the horizontal physical coordinates of the board. The corresponding coefficient of the first term, Represents the longitudinal physical coordinates of the board. The corresponding coefficient of the first term, Represents the horizontal physical coordinates of the board. The corresponding quadratic coefficient, Represents the longitudinal physical coordinates of the board. The corresponding quadratic coefficient; Step S402: Based on the temperature residual field Calculate local reference values ​​within a preset neighborhood around the current coordinate point. And construct self-referenced residuals ; Step S403: Based on the self-reference residual Calculate the abnormal response intensity and perform threshold segmentation based on the preset segmentation threshold to output a set of candidate defect regions.

[0019] Preferably, local reference value and self-reference residual The formula is expressed as: ; ; in, This represents the median operation. Represents neighboring points within a preset neighborhood. At any moment The corresponding temperature residual, Indicates the coordinates of the current board material. A pre-defined spatial neighborhood constructed around the center. Indicates the preset spatial neighborhood Any neighborhood coordinate point within.

[0020] This invention also provides a real-time infrared thermal imaging defect detection system for wood-based composite fiberboard, comprising: The session construction module is used to acquire the original infrared thermal image sequence, timestamp sequence, and conveying displacement pulse sequence of wood composite fiberboard in the continuous hot pressing station. Based on the original infrared thermal image sequence, timestamp sequence, and conveying displacement pulse sequence, the board detection session construction mechanism is used to execute the online detection session establishment task and output the board detection session dataset. The thermal image correction module is used to perform thermal image correction tasks based on the plate inspection session dataset using a velocity-constrained exposure and radiation correction mechanism, and outputs a corrected thermal image temperature sequence. The temporal field reconstruction module is used to perform temperature temporal field reconstruction based on the temperature sequence of the corrected thermal image using a plate coordinate mapping mechanism, and outputs the temperature temporal field of the plate coordinate system. The candidate region extraction module is used to perform the defect candidate region extraction task based on the temperature time series field of the plate coordinate system and the self-reference cooling feature fusion mechanism, and outputs a set of candidate defect regions. The defect determination module is used to perform connected component filtering, continuous verification, and internal delamination defect determination based on the candidate defect region set, and outputs the detection results of internal delamination defects in the board.

[0021] The present invention also provides a real-time infrared thermal imaging defect detection device for wood composite fiberboard, comprising: a memory, a processor, and a real-time infrared thermal imaging defect detection program for wood composite fiberboard stored in the memory and executable on the processor. When the real-time infrared thermal imaging defect detection program for wood composite fiberboard is executed by the processor, a real-time infrared thermal imaging defect detection method for wood composite fiberboard is implemented.

[0022] The present invention also provides a computer program product, including a real-time detection program for infrared thermal imaging defects in wood composite fiberboard, wherein the real-time detection program for infrared thermal imaging defects in wood composite fiberboard is executed by a processor to implement the real-time detection method for infrared thermal imaging defects in wood composite fiberboard.

[0023] The beneficial effects of this invention are as follows: By synchronously linking the transport displacement pulse sequence with the original infrared thermal image sequence, and combining it with speed-constrained exposure control, non-uniformity correction, de-blurring compensation, and bad pixel repair processing, this invention can stably obtain a corrected thermal image temperature sequence that can be used for analysis under the conditions of high overall platen surface temperature after hot pressing, continuous transport of the platen, and motion blur, thereby improving the accuracy of extracting the thermal response of internal degumming defects and reducing the false detection rate and missed detection rate caused by high temperature background interference, imaging ghosting, and sensor noise.

[0024] This invention constructs a temperature time-series field of the board coordinate system by performing board coordinate mapping on the corrected thermal image temperature sequence, and further combines background temperature field construction, temperature residual calculation, self-reference residual analysis, candidate defect region extraction and continuous verification to achieve continuous identification and real-time judgment of degumming defect regions inside wood composite fiberboard. This improves the stability and consistency of defect identification results in dynamic online detection scenarios and enhances the ability to distinguish internal defects under conditions of uneven local residual heat, edge heat dissipation differences and temperature drift. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a flowchart illustrating the first embodiment of a real-time infrared thermal imaging defect detection method for wood-based composite fiberboard according to the present invention.

[0027] Figure 2 This is a schematic diagram of the equipment for real-time detection of defects in wood-based composite fiberboard using infrared thermal imaging, according to the present invention. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] Example 1: As Figure 1 The diagram shown is a flowchart of the first embodiment of the real-time infrared thermal imaging defect detection method for wood composite fiberboard of the present invention, which presents the first embodiment of the real-time infrared thermal imaging defect detection method for wood composite fiberboard of the present invention.

[0030] In the first embodiment, the real-time infrared thermal imaging defect detection method for wood-based composite fiberboard includes: Step S10: Obtain the original infrared thermal image sequence, timestamp sequence, and conveying displacement pulse sequence of the wood composite fiberboard in the continuous hot pressing station. Based on the original infrared thermal image sequence, timestamp sequence, and conveying displacement pulse sequence, use the board detection session construction mechanism to perform the online detection session establishment task and output the board detection session dataset. It should be noted that the original infrared thermal image sequence refers to the sequence of plate surface temperature distribution data continuously acquired by the infrared thermal image acquisition unit at a preset sampling frequency in the continuous hot pressing plate exit station. It includes a two-dimensional temperature matrix or radiation response matrix corresponding to each sampling frame; the timestamp sequence refers to the time identifier sequence corresponding to each sampling frame, used to characterize the acquisition time of each frame of infrared thermal image; the conveying displacement pulse sequence refers to the cumulative sequence of pulse signals generated by the encoder linked with the conveying roller or conveyor belt, used to reflect the displacement change process of the plate in the conveying direction; the plate detection session construction mechanism refers to the processing method of segmenting and managing continuously acquired data based on the appearance and disappearance of the effective area of ​​the plate surface. It includes plate entry judgment, detection session identification establishment, data collection, and session end judgment, etc., used to uniformly associate the infrared thermal image data, time information, and displacement information of the same plate.

[0031] It is understandable that by synchronously acquiring and uniformly associating the original infrared thermal image sequence, timestamp sequence, and transport displacement pulse sequence in this step, a stable data organization structure can be established during the continuous dynamic transport of the sheet material. This ensures that each subsequent thermal image frame not only has temporal attributes but also corresponding spatial location information. Simultaneously, by effectively segmenting and aggregating data from different sheet materials through the sheet material detection session construction mechanism, the problem of data overlap between multiple sheet materials can be avoided. This provides a continuous and consistent data foundation for subsequent temperature time-series analysis based on the sheet material coordinate system, improving the stability and reliability of subsequent defect identification processing. For example, by synchronously acquiring and uniformly associating the original infrared thermal image sequence, timestamp sequence, and transport displacement pulse sequence in this step, a stable data organization structure can be established during the continuous dynamic transport of the sheet material. This ensures that each subsequent thermal image frame not only has temporal attributes but also corresponding spatial location information. Simultaneously, by effectively segmenting and aggregating data from different sheet materials through the sheet material detection session construction mechanism, the problem of data overlap between multiple sheet materials can be avoided. This provides a continuous and consistent data foundation for subsequent temperature time-series analysis based on the sheet material coordinate system, improving the stability and reliability of subsequent defect identification processing.

[0032] Step S20: Based on the plate inspection session dataset, a velocity-constrained exposure and radiometric correction mechanism is used to perform thermal image correction tasks, and the corrected thermal image temperature sequence is output; It should be noted that the board inspection session dataset refers to the continuous thermal image data set after the timestamp sequence and the transport displacement pulse sequence have been associated in step S10. It includes the original infrared thermal image sequence, timestamp information, and the corresponding displacement information. Velocity-constrained exposure refers to dynamically limiting the exposure time of the infrared thermal image acquisition unit based on the instantaneous transport speed of the board calculated from the transport displacement pulse sequence, so as to control the length of the trailing image caused by the movement of the board during the imaging process. Radiation correction refers to the process of temperature inversion and error compensation of the radiation response in the original thermal image based on the infrared radiation imaging model. It includes non-uniformity correction, apparent temperature conversion, emissivity estimation, and reflection temperature compensation, thereby converting the original infrared response value into temperature distribution data with physical meaning. The corrected thermal image temperature sequence refers to the continuous temperature field data sequence obtained after the above processing, which is used for subsequent board coordinate mapping and time series analysis.

[0033] Understandably, by introducing exposure control based on conveying speed in this step, the infrared thermal imaging acquisition process can be adaptively adjusted according to the movement state of the material, thereby effectively limiting the range of motion blur in a single frame image and ensuring that the spatial morphology of the defect area in the image does not significantly expand or distort. At the same time, by performing radiometric correction processing on the original thermal image, temperature measurement deviations caused by inconsistent detector responses, environmental radiation interference, and differences in the emission characteristics of the material surface can be eliminated, making temperature data at different times and locations comparable. The synergistic effect of these two processes improves both the spatial resolution and temperature accuracy of the output corrected thermal image temperature sequence, providing a reliable data foundation for subsequent defect feature extraction.

[0034] It should be understood that, compared to the traditional method of using fixed exposure parameters for infrared imaging and directly performing threshold analysis on the original thermal image, this embodiment incorporates the conveying speed into the exposure control constraint, establishing a correlation between the exposure time and the movement state of the board, thereby avoiding image ghosting problems caused by excessive exposure under high-speed conveying conditions. At the same time, compared to methods that only perform simple grayscale normalization or single temperature correction, this embodiment comprehensively performs non-uniformity correction, emissivity estimation, and reflection temperature compensation, enabling the thermal image data to more accurately reflect the true temperature distribution of the board. Thus, even under high-temperature backgrounds and uneven temperature distribution conditions, it can still maintain sensitivity to weak thermal anomalies, significantly improving the detectability of internal delamination defects.

[0035] Step S30: Based on the corrected thermal image temperature sequence, the plate coordinate mapping mechanism is used to perform the temperature time series field reconstruction task, and the plate coordinate system temperature time series field is output; It should be noted that the corrected thermal image temperature sequence refers to the continuous temperature field data sequence obtained after speed-constrained exposure control, non-uniformity correction, temperature inversion, and radiation compensation in step S20; the plate coordinate mapping mechanism refers to the processing method of correspondingly converting the pixel positions in the infrared thermal image with the actual transport displacement information of the plate, which includes plate surface area extraction, image coordinate to plate physical coordinate conversion, and continuous temperature data resampling; the temperature temporal field reconstruction task refers to reorganizing the continuously acquired multi-frame plate surface temperature data into a temporal temperature expression form under a unified coordinate system according to the actual physical position of the plate, thereby forming temperature field data that can characterize the temperature change law of the same physical position of the plate over time; the plate coordinate system temperature temporal field refers to the temperature data set constructed with the longitudinal and transverse physical coordinates of the plate as plane indices and the time dimension as the sequence index, which is used to characterize the temperature change state of different physical positions of the plate during continuous detection.

[0036] It should be understood that, compared to traditional techniques that directly perform frame-by-frame anomaly analysis on the original thermal images or simply compare adjacent frames using differential methods, this embodiment does not use fixed pixels in the camera's field of view as the analysis object. Instead, it first uses a plate coordinate mapping mechanism to merge the temperature data corresponding to the actual position of the plate in each frame image into a unified physical coordinate system, and then reconstructs the temperature temporal field in this unified physical coordinate system. The main problem with traditional methods is that when the plate is continuously transported, the image position of the same defect area changes constantly at different times. If the image pixels are still used as the analysis benchmark, it is easy to mistake temperature changes at different physical locations for temperature changes at the same location, leading to distortion of the thermal response curve. This error will be further amplified, especially when there are fluctuations in transport speed, uneven sampling intervals between frames, or inconsistent heat dissipation in the plate edge area. This embodiment uses the actual displacement of the board material to physically correct the pixel coordinates, so that the temperature data collected at different sampling times can be recombined around the same position of the board material. This not only eliminates the position mismatch problem caused by continuous movement, but also makes the temperature change curve obtained later more consistent with the actual cooling process of the board material, thereby significantly improving the accuracy and stability of the thermal response extraction of internal degumming defects.

[0037] For example, compared to traditional techniques that perform frame-by-frame anomaly analysis directly on the original thermal images or simply compare adjacent frames using differential methods, this embodiment does not use fixed pixels in the camera's field of view as the analysis object. Instead, it first uses a plate coordinate mapping mechanism to merge the temperature data corresponding to the actual position of the plate in each frame image into a unified physical coordinate system, and then reconstructs the temperature time series field in this unified physical coordinate system. The main problem with traditional methods is that when the plate is continuously transported, the image position corresponding to the same defect area changes constantly at different times. If the image pixels are still used as the analysis benchmark, it is easy to mistake temperature changes at different physical locations for temperature changes at the same location, resulting in distortion of the thermal response curve. This error will be further amplified, especially when there are fluctuations in transport speed, uneven sampling intervals between frames, or inconsistent heat dissipation in the plate edge area. This embodiment uses the actual displacement of the board material to physically correct the pixel coordinates, so that the temperature data collected at different sampling times can be recombined around the same position of the board material. This not only eliminates the position mismatch problem caused by continuous movement, but also makes the temperature change curve obtained later more consistent with the actual cooling process of the board material, thereby significantly improving the accuracy and stability of the thermal response extraction of internal degumming defects.

[0038] Step S40: Based on the temperature time series field of the plate coordinate system, a self-reference cooling feature fusion mechanism is used to perform the defect candidate region extraction task and output the candidate defect region set; It should be noted that the temperature time series field of the board coordinate system refers to the continuous temperature data set after the image coordinates and the physical coordinates of the board have been unified in step S30. It can characterize the temperature response process of the same physical location of the board over time. The self-reference cooling feature fusion mechanism refers to the processing method of first constructing a background temperature field for the temperature time series data of each physical location in the board coordinate system, then calculating the temperature residual, self-reference residual, and abnormal response intensity based on the cooling process, and combining and analyzing multiple features used to characterize local thermal anomalies. The defect candidate region extraction task refers to identifying regions with abnormal cooling behavior from the temperature time series field of the board coordinate system and using these regions as the inspection areas for subsequent internal delamination determination. The candidate defect region set refers to one or more sets of local abnormal regions obtained after background removal, self-reference comparison, and abnormal response segmentation. It includes the spatial location range of each candidate region and the corresponding abnormal response information. Furthermore, the "self-reference" in this step refers to using the temperature residual distribution of the surrounding neighborhood of the location to be analyzed in the current plate material as a reference benchmark without relying on external standard plates or pre-calling fixed template curves, in order to measure whether the cooling behavior of that location deviates from its neighboring normal area. The "cooling feature fusion" in this step refers to not judging solely based on the temperature difference at a certain moment, but comprehensively considering the degree of deviation of the local temperature relative to the background, the residual difference relative to the neighborhood, and the retention of abnormal responses during continuous cooling, thereby improving the stability of the candidate defect area extraction results.

[0039] Understandably, by constructing the background temperature field and extracting residuals from the temperature time series field of the board coordinate system in this step, the influence of large-scale temperature changes caused by uneven overall residual heat distribution, differences in heat dissipation at the board edges, and longitudinal and transverse temperature gradients can be preemptively eliminated, allowing subsequent analysis to focus more on local abnormal thermal responses. Furthermore, by introducing self-referenced residual analysis, the temperature change behavior of the current board coordinate point can be compared with the temperature change behavior of its surrounding neighborhood, thus establishing anomaly judgment based on local references within the same board, batch, and operating conditions, rather than on a uniform fixed threshold or ideal temperature distribution assumption. Since the internal debonded area usually exhibits a different heat diffusion and heat dissipation path than the surrounding normally bonded area after hot pressing, its cooling process often shows persistent local temperature differences, significant differences in decay rates, or prolonged abnormal response durations. Therefore, this step, by analyzing the temperature response of the same physical location on the board over continuous time and fusing the obtained abnormal features, can more effectively extract areas with defect indication significance from the complex background. In other words, this step does not simply look for “hotter” or “colder” regions at a single moment, but rather starts from the continuous cooling process to identify locations that exhibit abnormal thermal response patterns relative to the surrounding areas, thereby making the set of candidate defect regions closer to the true distribution of internal degummed regions.

[0040] For example, in a real-world inspection scenario after continuous hot pressing, a wood-based composite fiberboard may still have a relatively high surface temperature when entering the inspection area. The central area of ​​the board is heated relatively evenly, while the edges, due to greater contact with the outside environment, dissipate heat faster, and a temperature gradient gradually decreases from the hot pressing outlet to the cooling section along the conveying direction. In this case, if global threshold segmentation is performed based solely on a single frame of thermal image, the faster heat dissipation area at the board edge may be misjudged as an abnormal area, while an internal delamination area in the center of the board may be missed due to its less pronounced temperature difference. This step addresses this by first constructing a background temperature field in the board's coordinate system temperature time series field to characterize the overall temperature change trend of the board under the current operating conditions; then subtracting the background temperature field from the actual temperature response at each location yields the temperature residual, thereby weakening the heat dissipation effect at the board edge and the overall temperature gradient; further, comparing the residuals of the target location with its surrounding neighborhood, if a location exhibits a stronger or more persistent abnormal deviation relative to its neighborhood across multiple consecutive sampling times, then that location and its adjacent areas are extracted as candidate defect areas. Taking a specific scenario as an example, the temperature response of the normal area of ​​the board material shows a steady decline during continuous sampling. However, due to an abnormal local heat conduction path, a certain internal delamination location consistently exhibits a temperature residual difference deviating from its neighboring area within the same cooling time, maintaining a high abnormal response intensity at multiple consecutive moments. In this case, although the absolute temperature difference at this location may not be significant in a single frame of the thermal image, it can still be stably extracted as a candidate defect region after background removal and self-reference comparison in this step. Therefore, this step can more accurately separate regions with truly abnormal cooling behavior from the overall board temperature field under complex residual heat backgrounds and dynamic transport conditions, providing highly reliable candidate region input for subsequent internal delamination defect determination.

[0041] Step S50: Based on the candidate defect region set, perform connected component screening, continuity verification, and internal debonding defect determination, and output the detection results of internal debonding defects in the board.

[0042] It should be noted that the candidate defect region set refers to one or more regions to be inspected obtained in step S40 based on the temperature time series field of the board coordinate system and after background temperature field removal, self-reference residual construction, and abnormal response segmentation. It includes the board coordinate range, abnormal response intensity distribution, and temperature response information over continuous time corresponding to each candidate region. Connectivity filtering refers to the process of aggregating adjacent abnormal pixels or adjacent abnormal coordinate points in the candidate defect region set and removing discrete small regions that obviously do not conform to the internal debonding morphology characteristics based on region area, region aspect ratio, boundary integrity, or spatial continuity. Persistence verification refers to further checking whether the candidate regions retained after connectivity filtering continuously exhibit abnormal thermal response under multiple consecutive sampling times or multiple consecutive resampling time indices to distinguish between real internal debonding regions and short-term thermal noise disturbance regions. Internal debonding defect determination refers to determining whether the candidate region corresponds to the internal bonding failure region of the board based on the spatial connectivity characteristics and temporal persistence characteristics of the candidate region, and outputting the defect location, defect range, and defect determination result. Furthermore, the "detection results of internal debonding defects in the board" in this step can include not only the conclusion that there are internal debonding defects, but also the characterization content such as the longitudinal position, transverse position, area, outer frame range, duration of abnormal response, and confidence status of the board corresponding to the defect, so as to realize alarm output, marking and sorting, or production quality traceability in the future.

[0043] It should be understood that, compared with the traditional method of directly outputting defect results based on area thresholds or temperature difference thresholds after completing heat map segmentation, this embodiment adds a joint judgment process of connected component screening and continuity verification in this step, which can significantly improve the problems of high false detection rate and difficulty in stable identification of weak defects in the traditional method under continuous hot pressing and unloading conditions. Traditional methods typically extract abnormal regions from a single frame of image and then determine defects based on the area, temperature peak, or grayscale contrast of that frame. The main limitations of this approach are: First, local anomalies in a single frame do not necessarily correspond to actual internal delamination areas. Especially when the overall board surface is hot, residual heat is unevenly distributed, or there is reflected heat interference from roller conveyors or equipment components, normal areas may experience sudden temperature changes at a certain moment, leading to misjudgment as defects. Second, traditional methods often lack continuous analysis of abnormal regions during the subsequent cooling process, making it difficult to effectively distinguish between "persistent internal structural anomalies" and "short-term environmental disturbances or measurement noise." Third, traditional methods rely heavily on single-threshold judgments. If the threshold is set too high, it easily misses shallow delamination or weakly defined abnormal regions; if the threshold is set too low, it easily retains small noise points that have no defect significance. The improvement in this embodiment is that, based on the candidate regions obtained in step S40, a defect conclusion is not immediately drawn. Instead, the spatial aggregation and temporal continuity of the candidate regions are further verified. In other words, this embodiment does not simply answer "whether there is an anomaly at a certain moment," but rather "whether the anomaly forms a region with spatial integrity, and whether this region persists during continuous cooling." In applications involving continuous hot pressing, internal delamination areas typically exhibit relatively stable thermal response deviations across multiple consecutive sampling times due to changes in local heat conduction paths. Therefore, this embodiment, through continuous verification, can more accurately identify such genuine defect areas. Simultaneously, false anomalies caused by localized thermal reflection, sensor random noise, or short-term temperature fluctuations on the board surface, often lacking stability over continuous time, can be effectively eliminated during the continuous verification phase. Thus, compared to traditional techniques, this step improves the robustness of internal delamination defect detection, reduces the probability of false positives, and enhances the ability to retain weakly anomalous regions without relying on overly stringent single-frame threshold settings.

[0044] For example, in actual production, after a piece of wood-based composite fiberboard enters the detection area from the hot press outlet, the overall temperature of the board surface is high, and multiple abnormal responses may appear simultaneously in the infrared thermography. Some of these abnormal responses originate from localized temperature differences caused by enhanced heat dissipation at the board edge, others from instantaneous reflected heat interference from metal components above the conveyor rollers, and still others may correspond to actual delamination areas within the board. After extracting candidate defect areas in step S40, assume three candidate areas are obtained: the first candidate area is located at the board edge, with a small area and scattered dots; the second candidate area is located in the middle of the board surface, with a moderate area and continuous space; the third candidate area is close to the projection area of ​​the high-temperature components of the equipment, and although the temperature difference is significant at a certain moment, it quickly disappears in subsequent moments. If the traditional method of directly outputting after single-frame segmentation is used, all three areas may be considered defect areas, leading to false alarms. In this embodiment, the three candidate regions are first screened for connectivity. The first candidate region is eliminated because its area is too small and it lacks spatial continuity. Then, a persistence check is performed on the remaining regions. The third candidate region is eliminated because it only exhibits anomalies at a single moment and does not show continuous deviations in subsequent times, thus it is judged as short-term interference. The second candidate region is ultimately determined to be an internal debonding defect region because it maintains a stable abnormal response across multiple sampling moments and forms a spatially coherent region. For example, in another production scenario, a shallow debonding region may not show a significant temperature difference in a single-frame thermal image. If only the instantaneous temperature difference threshold is used, it may not meet the traditional judgment criteria. However, this region consistently maintains a temperature residual deviation relative to the surrounding area during continuous cooling, exhibiting a continuous and stable abnormal persistence characteristic. Therefore, in the persistence check stage of this embodiment, it can still be retained and ultimately determined to be an internal debonding defect.

[0045] Example 2: Furthermore, the present invention provides a real-time infrared thermal imaging defect detection system for wood-based composite fiberboard, employing a real-time infrared thermal imaging defect detection method for wood-based composite fiberboard as described in the above embodiments, which can solve the technical problem of real-time infrared thermal imaging defect detection in wood-based composite fiberboard. The beneficial effects of the real-time infrared thermal imaging defect detection system for wood-based composite fiberboard provided by the present invention are the same as those of the real-time infrared thermal imaging defect detection method for wood-based composite fiberboard provided in the above embodiments, and other technical features of the real-time infrared thermal imaging defect detection system for wood-based composite fiberboard are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0046] Example 3: This invention provides a real-time infrared thermal imaging defect detection device for wood-based composite fiberboard. Please refer to... Figure 2A real-time infrared thermal imaging defect detection device for wood composite fiberboard includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enable the at least one processor to perform the real-time infrared thermal imaging defect detection method for wood composite fiberboard described in Embodiment 1 above. The real-time infrared thermal imaging defect detection device for wood composite fiberboard in this embodiment of the invention may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), and vehicle terminals (e.g., vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. This real-time infrared thermal imaging defect detection device for wood composite fiberboard is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the invention. A real-time infrared thermal imaging defect detection device for wood-based composite fiberboard may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory 1002 or a program loaded from a storage device 1003 into a random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of the real-time infrared thermal imaging defect detection device for wood-based composite fiberboard. The processing unit 1001, the read-only memory 1002, and the random access memory 1004 are interconnected via a bus 1005. An I / O interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touch screen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. The communication device 1009 allows a real-time infrared thermal imaging defect detection device for wood-based composite fiberboard to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows a real-time infrared thermal imaging defect detection device for wood-based composite fiberboard with various systems, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0047] Example 4: This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the real-time infrared thermal imaging defect detection method for wood-based composite fiberboard as described above. The computer program product provided by this invention can solve the technical problem of real-time infrared thermal imaging defect detection in wood-based composite fiberboard. Compared with the prior art, the beneficial effects of the computer program product provided by this invention are the same as those of the real-time infrared thermal imaging defect detection method for wood-based composite fiberboard provided in the above embodiments, and will not be repeated here.

[0048] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this invention.

[0049] It should be understood that the various parts disclosed in this invention can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0050] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for real-time detection of defects in wood-based composite fiberboard using infrared thermal imaging, characterized in that, The methods include: Step S10: Obtain the original infrared thermal image sequence, timestamp sequence, and conveying displacement pulse sequence of the wood composite fiberboard in the continuous hot pressing station. Based on the original infrared thermal image sequence, timestamp sequence, and conveying displacement pulse sequence, use the board detection session construction mechanism to perform the online detection session establishment task and output the board detection session dataset. Step S20: Based on the plate inspection session dataset, a velocity-constrained exposure and radiometric correction mechanism is used to perform thermal image correction tasks, and the corrected thermal image temperature sequence is output; Step S30: Based on the corrected thermal image temperature sequence, the plate coordinate mapping mechanism is used to perform the temperature time series field reconstruction task, and the plate coordinate system temperature time series field is output; Step S40: Based on the temperature time series field of the plate coordinate system, a self-reference cooling feature fusion mechanism is used to perform the defect candidate region extraction task and output the candidate defect region set; Step S50: Based on the candidate defect region set, perform connected component screening, continuity verification, and internal debonding defect determination, and output the detection results of internal debonding defects in the board.

2. The method for real-time detection of defects in wood-based composite fiberboard using infrared thermal imaging as described in claim 1, characterized in that, Step S10 involves acquiring the original infrared thermal image sequence, timestamp sequence, and conveying displacement pulse sequence of the wood-based composite fiberboard in the continuous hot-pressing station. Based on the original infrared thermal image sequence, timestamp sequence, and conveying displacement pulse sequence, an online detection session establishment task is performed using a board detection session construction mechanism to output the board detection session dataset. Specifically, this includes: Step S101: An infrared thermal imaging acquisition unit and a conveyor displacement acquisition unit are installed between the hot press outlet and the cooling section. Based on the infrared thermal imaging acquisition unit and the conveyor displacement acquisition unit, raw infrared thermal image sequences are continuously acquired. Timestamp sequence and the transmission displacement pulse sequence ;in, This indicates the sequence number of the nth sampling frame during continuous acquisition. Step S102: Obtain the effective pixels on the board surface within the current acquisition area, and determine the percentage of effective pixels on the board surface within the current acquisition area. Establish entry criteria for the board material and the percentage of effective pixels on the board surface. satisfy: ;in, This indicates the number of pixels in the currently connected region of the front panel. Indicates the total number of pixels in the acquisition area; when At that time, establish the testing session identifier corresponding to the current board material. ; Step S103: Detect session identifier As an index, the original infrared thermal image sequence Timestamp sequence and the transmission displacement pulse sequence The data is stored in association to form a panel inspection session dataset.

3. The method for real-time detection of defects in wood-based composite fiberboard using infrared thermal imaging as described in claim 1, characterized in that, Step S20, which involves performing a thermal image correction task based on the plate inspection session dataset using a velocity-constrained exposure and radiometric correction mechanism, and outputting a corrected thermal image temperature sequence, specifically includes: Step S201: Based on the transmission displacement pulse sequence Calculate the plate displacement sequence And based on the plate displacement sequence Calculate the conveying speed by the displacement change between adjacent sampling times. ;in, , This refers to the encoder single-pulse displacement calibration coefficient; Step S202: Obtain the pixel size in the transport direction and the upper limit of target fuzziness Combined with conveying speed A dynamic exposure control method based on Nyquist sampling constraints is used to perform velocity-constrained exposure control on the original infrared thermal image frames. Non-uniformity correction and bad pixel repair are then performed on the original infrared thermal image frames to obtain a corrected thermal image sequence. ; Step S203: Based on the temperature of the reflective reference region and the high emissivity reference region measured by the preset reference temperature acquisition unit, the thermal image sequence is calibrated. The Planck radiation inversion method was used to perform apparent temperature conversion, emissivity estimation, and reflectance temperature compensation to obtain the corrected thermal image temperature sequence. .

4. The method for real-time detection of defects in wood-based composite fiberboard using infrared thermal imaging as described in claim 3, characterized in that, In step S202, speed-constrained exposure control is performed on the original infrared thermal image frame, and non-uniformity correction and bad pixel repair are performed on the original infrared thermal image frame to obtain a corrected thermal image sequence. The steps are as follows: Step S2021: Based on the conveying speed Calculate the maximum allowable exposure time ; and actual exposure time Set to: in, This is the default exposure duration; Step S2022: After completing exposure control, perform non-uniformity correction on the original infrared thermal image frame using a combination of Wiener enhancement and column mean offset correction. The correction expression is: Where x represents the column coordinate in the thermal image frame; y represents the row coordinate in the thermal image frame; and n represents the sampling frame number. Indicates the first The first frame Column pixel mean Indicates the first The global pixel mean of a frame; Step S2023: and according to the conveying speed Actual exposure time and the pixel size in the transport direction Calculate fuzzy length : when At that time, the Richardson-Lucy one-dimensional deconvolution compensation method was used to perform deblurring along the conveying direction to obtain a deblurred temperature image; Step S2024: Use the Kalman recursive repair method to repair bad pixels in the deblurred temperature image to obtain the corrected thermal image sequence. .

5. The method for real-time detection of defects in wood-based composite fiberboard using infrared thermal imaging as described in claim 3, characterized in that, Step S30, which involves performing a temperature time-series reconstruction task based on the corrected thermal image temperature sequence using a plate coordinate mapping mechanism and outputting the temperature time-series field in the plate coordinate system, specifically includes: Step S301: Extract the mask of the plate surface area based on the corrected thermal image temperature sequence And based on the mask of the plate area Temperature data within the current plate area is extracted using a combination of Otsu threshold segmentation and connected component filtering. Step S302: Combine with plate displacement and the pixel size in the delivery direction Map each pixel within the current frame's board surface area to the board coordinate system. The mapping relationship is as follows: ； in, This represents the pixel coordinates of a pixel in the current frame along the transport direction. This represents the vertical physical coordinates of the board material corresponding to the pixel; and the pixel coordinates of the pixel in the previous frame perpendicular to the conveying direction. According to pixel size Mapped to the horizontal physical coordinates of the board ,satisfy: ； in, This represents the pixel coordinates of a pixel in the current frame that are perpendicular to the transport direction. Indicates the horizontal pixel size; Step S303: Using the longitudinal physical coordinates of the board material and the horizontal physical coordinates of the board As an index, the temperature maps of consecutive moments within the current previous frame's board surface area are resampled to construct a temperature time series field in the board coordinate system. Where t represents the sampling time or the time index after resampling corresponding to the plate coordinates.

6. The method for real-time detection of defects in wood-based composite fiberboard using infrared thermal imaging as described in claim 5, characterized in that, Step S40, which involves extracting candidate defect regions using a self-referenced cooling feature fusion mechanism based on the temperature time-series field of the plate coordinate system and outputting a set of candidate defect regions, specifically includes: Step S401: Based on the temperature field and time sequence field of the plate coordinate system Constructing the background temperature field And calculate the temperature residual field. ,in: in, The constant term coefficients representing the background temperature field. Represents the horizontal physical coordinates of the board. The corresponding coefficient of the first term, Represents the longitudinal physical coordinates of the board. The corresponding coefficient of the first term, Represents the horizontal physical coordinates of the board. The corresponding quadratic coefficient, Represents the longitudinal physical coordinates of the board. The corresponding quadratic coefficient; Step S402: Based on the temperature residual field Calculate local reference values ​​within a preset neighborhood around the current coordinate point. And construct self-referenced residuals ; Step S403: Based on the self-reference residual Calculate the abnormal response intensity and perform threshold segmentation based on the preset segmentation threshold to output a set of candidate defect regions.

7. The method for real-time detection of defects in wood-based composite fiberboard using infrared thermal imaging as described in claim 6, characterized in that, Local reference value and self-reference residual The formula is expressed as: ； ； in, This represents the median operation. Represents neighboring points within a preset neighborhood. At any moment The corresponding temperature residual, Indicates the coordinates of the current board material. A pre-defined spatial neighborhood constructed around the center. Represents the preset spatial neighborhood Any neighborhood coordinate point within.

8. A real-time infrared thermal imaging defect detection system for wood-based composite fiberboard, applied to the real-time infrared thermal imaging defect detection method for wood-based composite fiberboard according to any one of claims 1 to 7, characterized in that, The real-time infrared thermal imaging defect detection system for wood-based composite fiberboard includes: The session construction module is used to acquire the original infrared thermal image sequence, timestamp sequence, and conveying displacement pulse sequence of wood composite fiberboard in the continuous hot pressing station. Based on the original infrared thermal image sequence, timestamp sequence, and conveying displacement pulse sequence, the board detection session construction mechanism is used to execute the online detection session establishment task and output the board detection session dataset. The thermal image correction module is used to perform thermal image correction tasks based on the plate inspection session dataset using a velocity-constrained exposure and radiation correction mechanism, and outputs a corrected thermal image temperature sequence. The temporal field reconstruction module is used to perform temperature temporal field reconstruction based on the temperature sequence of the corrected thermal image using a plate coordinate mapping mechanism, and outputs the temperature temporal field of the plate coordinate system. The candidate region extraction module is used to perform the defect candidate region extraction task based on the temperature time series field of the plate coordinate system and the self-reference cooling feature fusion mechanism, and outputs a set of candidate defect regions. The defect determination module is used to perform connected component filtering, continuous verification, and internal delamination defect determination based on the candidate defect region set, and outputs the detection results of internal delamination defects in the board.

9. A real-time infrared thermal imaging defect detection device for wood-based composite fiberboard, characterized in that, The real-time infrared thermal imaging defect detection device for wood composite fiberboard includes: a memory, a processor, and a real-time infrared thermal imaging defect detection program for wood composite fiberboard stored in the memory and executable on the processor. When the real-time infrared thermal imaging defect detection program for wood composite fiberboard is executed by the processor, it implements a real-time infrared thermal imaging defect detection method for wood composite fiberboard according to any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a real-time infrared thermal imaging defect detection program for wood composite fiberboard, which, when executed by a processor, implements a real-time infrared thermal imaging defect detection method for wood composite fiberboard according to any one of claims 1 to 7.

Images