Textile fabric wicking rate detection method and equipment based on thermal imaging

Abstract

The invention provides a textile fabric wicking rate detection method and equipment based on thermal imaging, and the detection method comprises the steps: fixing the installation posture of a camera part, calibrating the imaging geometry of a sample, building an image pixel coordinate system and a sample plane coordinate system based on the imaging geometry, a perspective mapping relation between image pixel coordinates and sample plane coordinates is determined; recording a sample dry position as a dry area, recording a sample wet position as a wet area, contacting the sample with liquid, collecting an image sequence of temperature distribution, identifying the dry area and the wet area through the image sequence, and separating dry and wet boundaries; and converting the dry and wet boundaries into boundary coordinates based on the mapping relationship, tracking the boundary coordinates in the image sequence in a time sequence, and calculating a quantitative index of the wicking rate of the sample. According to the detection method, a temperature distribution image sequence is collected, dry and wet boundaries are separated, and pixel-physical scale conversion is realized through a link of calibrating imaging geometry, establishing image pixel coordinates and sample plane coordinates, and establishing perspective mapping.

Description

Technical Field

[0001] This application relates to the field of fabric testing methods, and in particular to a method and device for detecting the wicking rate of textile fabrics based on thermal imaging. Background Technology

[0002] The water absorption and diffusion behavior of textile fabrics, especially clothing fabrics, is one of the important indicators for evaluating wearing comfort. In the industry, wicking height, diffusion distance, diffusion area, and their changes over time are usually used as quantitative parameters. Under laboratory conditions, samples are often subjected to droplet or end-wetting tests in a standard environment to record the diffusion of liquid water in the warp and weft directions or diagonally.

[0003] Current detection methods primarily rely on manual observation: after the sample comes into partial contact with liquid water, the operator visually identifies the boundary of the wet area and measures the diffusion distance with a ruler, or records the wicking height for strip-shaped samples. To enhance visibility, some methods add colored tracers to the water. In recent years, there have also been automatic reading methods using visible light cameras and computer image processing, which segment dry and wet areas based on grayscale / color differences and convert the lengths. Industry standards (e.g., FZ / T 01071) require wicking height readings of three parallel samples to achieve millimeter-level accuracy, and measurements to be completed at set time points or time intervals to obtain comparable results.

[0004] However, existing fabric wicking diffusion detection methods mostly rely on the naked eye or visible light cameras to identify wet areas based on color / brightness differences, which are easily affected by fabric color and texture, ambient lighting, and surface reflection. Manual interpretation and measurement are highly subjective, with poor accuracy and repeatability, and it is difficult to simultaneously read multiple samples at the same time in rapid diffusion scenarios. Ordinary image methods often ignore lens distortion, visual field trapezoids, and pixel-to-physical scale conversion, and lack complete calibration and mapping, resulting in systematic errors in quantitative results such as distance and area. Therefore, there is an urgent need for a detection method and equipment that can accurately measure the wicking position of textile fabrics to solve the above problems. Summary of the Invention

[0005] In view of this, it is necessary to provide a method and device for detecting the wicking rate of textile fabrics based on thermal imaging in order to solve the above problems.

[0006] The embodiments of this application provide a method for detecting the wicking rate of textile fabric based on thermal imaging, including fixing the installation posture of the camera, calibrating the imaging geometry of the sample, establishing an image pixel coordinate system and a sample plane coordinate system based on the imaging geometry, and establishing a perspective mapping relationship between the image pixel coordinates and the sample plane coordinates.

[0007] The dry areas of the sample are designated as dry zones, and the wet areas are designated as wet zones. The sample is brought into contact with the liquid, and an image sequence of temperature distribution is acquired. The dry zones and wet zones are identified through the image sequence, and the dry-wet boundary is separated.

[0008] Based on the mapping relationship, the dry and wet boundary is converted into boundary coordinates, the boundary coordinates in the image sequence are tracked in time, and the quantitative index of the wicking rate of the sample is calculated.

[0009] In at least one embodiment of this application, the step of "calibrating the imaging geometry of the sample" is further included before the step of:

[0010] The sample was screened and its standard geometric dimensions were calibrated. The sample and the camera were placed in a stable and uniform environment, the sample posture was fixed, and the ambient temperature was set to 20°C and the ambient humidity to 65%.

[0011] In at least one embodiment of this application, the step "calibrating the imaging geometry of the sample and establishing image pixel coordinates and sample plane coordinates based on the imaging geometry" specifically includes the following steps:

[0012] The imaging area of ​​the camera is obtained, the standard geometric size features are identified and the imaging geometry is calibrated, the origin and axis of the coordinate system are defined at the location of the sample near the water source, and the plane coordinates of the sample are established.

[0013] Acquire thermal imaging image frames within the imaging area, and establish image pixel coordinates based on the thermal imaging image frames.

[0014] In at least one embodiment of this application, the step "establishing the mapping relationship between the image pixel coordinates and the sample plane coordinates" specifically includes the following steps:

[0015] Identify at least four sets of non-collinear physical coordinate points of the sample plane coordinates, obtain pixel coordinates in the image, pair the pixel coordinates with the physical coordinates, establish pixel-physical correspondence points, and calculate the perspective mapping relationship between the sample plane coordinates and the corresponding coordinates of the image pixel coordinates based on the pixel-physical correspondence points.

[0016] In at least one embodiment of this application, the step of "pairing the pixel coordinates with the physical coordinates and establishing the pixel physical correspondence point" further includes the following step:

[0017] The distortion parameters in the imaging geometry are calculated based on the physical corresponding points of the pixels. The thermal imaging frame is remapped according to the distortion parameters to obtain the corrected pixel coordinates. The perspective mapping relationship is calculated based on the corrected pixel coordinates and the sample plane coordinates.

[0018] In at least one embodiment of this application, the imaging region includes a water source and a sample intended to diffuse.

[0019] In at least one embodiment of this application, the step "acquiring an image sequence of temperature distribution, identifying the dry area and the wet area through the image sequence, and separating the dry and wet boundaries" includes the following specific steps:

[0020] A preset temperature threshold is set, and a temperature reference value for the dry zone is obtained. Based on the temperature reference value and the temperature threshold, a separation temperature value is calculated to identify the wet zone and define the dry-wet boundary.

[0021] In at least one embodiment of this application, the step "converting the wet and dry boundary into boundary coordinates based on the mapping relationship" includes the following specific steps:

[0022] Based on the perspective mapping relationship, the pixel coordinates are mapped to the sample plane coordinates to form a boundary point set. The boundary point set is then moved to align with the origin and axis of the sample plane coordinates to output the boundary coordinates corresponding to each image frame.

[0023] In at least one embodiment of this application, the steps of claim 1 are repeated to select an average quantization index to establish an accurate wicking rate quantization index.

[0024] In at least one embodiment of this application, the detection device is applied to a method for detecting the wicking rate of textile fabrics based on thermal imaging as described in any one of claims 1-9.

[0025] The aforementioned method and device for detecting the wicking rate of textile fabrics based on thermal imaging acquires a sequence of temperature distribution images and uses the temperature of the dry area as a reference to separate the wet and dry boundaries. The identification is based on temperature difference rather than color / brightness, avoiding interference from fabric color, texture, and ambient light reflection. Through a process of "calibrating imaging geometry → establishing image pixel coordinates and sample plane coordinates → establishing perspective mapping," pixel-to-physical scale conversion is achieved, transforming the boundary from the pixel domain to boundary coordinates on the sample plane. This provides a unified physical standard for quantitative results such as distance and area, significantly reducing systematic errors caused by viewing angle and scale. By performing time-series tracking of the boundary coordinates under the sample plane coordinates, dynamic indicators such as wicking rate (e.g., boundary movement speed / direction) are directly output, reflecting the diffusion process more accurately than simply measuring distance / height. Attached Figure Description

[0026] Figure 1 This is a flowchart of a method for detecting the wicking rate of textile fabric based on thermal imaging, according to one embodiment of this application. Detailed Implementation

[0027] The embodiments of this application will now be described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0028] It should be noted that when a component is considered to be "connected" to another component, it can be directly connected to the other component or may also have an intervening component. When a component is considered to be "placed" on another component, it can be directly placed on the other component or may also have an intervening component. The terms "top," "bottom," "upper," "lower," "left," "right," "front," "back," and similar expressions used in this article are for illustrative purposes only.

[0029] The embodiments of this application provide a method and device for detecting the wicking rate of textile fabrics based on thermal imaging. The method includes fixing the installation posture of the camera, calibrating the imaging geometry of the sample, establishing an image pixel coordinate system and a sample plane coordinate system based on the imaging geometry, and establishing a perspective mapping relationship between the image pixel coordinates and the sample plane coordinates.

[0030] The dry areas of the sample are designated as dry zones, and the wet areas are designated as wet zones. The sample is brought into contact with the liquid, and an image sequence of temperature distribution is acquired. The dry zones and wet zones are identified through the image sequence, and the dry-wet boundary is separated.

[0031] Based on the mapping relationship, the dry and wet boundary is converted into boundary coordinates, the boundary coordinates in the image sequence are tracked in time, and the quantitative index of the wicking rate of the sample is calculated.

[0032] The aforementioned method and device for detecting the wicking rate of textile fabrics based on thermal imaging acquires a sequence of temperature distribution images and uses the temperature of the dry area as a reference to separate the wet and dry boundaries. Identification is based on temperature difference rather than color / brightness, avoiding interference from fabric color, texture, and ambient light reflection. Through a process of "calibrating imaging geometry → establishing image pixel coordinates and sample plane coordinates → establishing perspective mapping," pixel-to-physical scale conversion is achieved, transforming the boundary from the pixel domain to boundary coordinates on the sample plane. This provides a unified physical caliber for quantitative results such as distance and area, significantly reducing systematic errors caused by viewing angle and scale. By performing time-series tracking of the boundary coordinates under the sample plane coordinates, dynamic indicators such as wicking rate (e.g., boundary movement speed / direction) are directly output, reflecting the diffusion process more accurately than simply measuring distance / height.

[0033] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0034] Please see Figure 1The embodiments of this application provide a method S100 for detecting the wicking rate of textile fabrics based on thermal imaging, comprising:

[0035] S10 fixes the installation posture of the camera, calibrates the imaging geometry of the sample, establishes the image pixel coordinate system and the sample plane coordinate system based on the imaging geometry, and establishes the perspective mapping relationship between the image pixel coordinates and the sample plane coordinates.

[0036] The dry area of ​​the S20 sample is designated as the dry zone, and the wet area is designated as the wet zone. The sample is brought into contact with the liquid, and an image sequence of temperature distribution is acquired. The dry zone and the wet zone are identified through the image sequence, and the dry and wet boundaries are separated.

[0037] S30 converts the wet and dry boundary into boundary coordinates based on the mapping relationship, tracks the boundary coordinates in the image sequence in time, and calculates the quantitative index of the wicking rate of the sample.

[0038] Preferably, in this embodiment, it should be noted that the camera used is a combination of an infrared thermal imaging device and a regular lens, with a fixed installation posture to ensure its field of view covers the expected diffusion area of ​​the water source and the sample. Based on the calibrated imaging geometry, an image pixel coordinate system is established on the thermal imaging image frame, and a sample plane coordinate system is established on the sample surface. The perspective mapping relationship between the two is established to achieve quantitative conversion between the pixel domain and the physical plane. During the experiment, the dry area of ​​the sample is defined as the dry zone, and the area formed after liquid immersion is defined as the wet zone. After the sample comes into contact with the liquid, a temperature distribution image sequence is continuously acquired according to a predetermined procedure, and each frame of the image is used as a temperature matrix for subsequent analysis. In the pixel coordinate system, the known dry zone far from the water source is used as the temperature reference. The temperature difference between the pixel temperature and the dry zone reference is compared for each frame of the sequence to determine the pixel category and achieve dry / wet region separation. Then, the dry / wet boundary curve is extracted in the pixel domain.

[0039] Specifically, after the sample comes into contact with the water source, the liquid diffuses outward driven by capillary forces in the fiber pores and yarn channels. The combined effect of evaporation heat absorption on the surface of the wet area and latent heat exchange during the phase change of moisture within the fabric causes the apparent radiation temperature of the wet area to be lower than that of the dry area for a short period of time, exhibiting a time-dependent temperature gradient as diffusion progresses. Therefore, under stable and uniform environmental conditions, the temperature of the wet area can serve as an objective basis for segmentation. By comparing it with the temperature benchmark of the dry area, the segmentation temperature value is determined, thereby identifying dry / wet regions frame by frame and extracting the dry / wet boundary.

[0040] Furthermore, by invoking the aforementioned perspective mapping relationship, the pixel domain boundary is converted frame by frame into a set of boundary coordinates in the sample plane, uniformly falling under the same physical coordinate system for representation. To obtain a quantitative index of the wicking rate, this embodiment selects a measurement path or feature quantity (e.g., the shortest / longest distance from the boundary to the sample origin, or the projected distance along the main diffusion direction) within the sample plane coordinate system, and performs temporal tracking and differential calculation on the boundary coordinates at adjacent time points:

[0041]

[0042] Where s(ti) is the value of the feature quantity at time t, and v(ti) is the corresponding boundary movement velocity. After the sequence traversal is completed, the wicking rate quantization result with velocity as the core is output. If necessary, the trajectory of the boundary coordinates over time can be saved simultaneously for verification and tracking.

[0043] Furthermore, by using thermal imaging as the information source and dry area temperature as the benchmark for dry / wet discrimination, the identification is based on temperature difference rather than visible light color / brightness difference. This achieves consistent area separation on fabrics of different colors and textures, reducing the influence of visual attributes on the detection results and improving the objectivity and stability of the discrimination. By establishing an image pixel coordinate system and a sample plane coordinate system before the experiment begins, and establishing the perspective mapping relationship between the two, the boundary positions of the pixel domain can be mapped one-to-one to physical plane coordinates. Geometric quantities such as length and distance can be directly calculated under the same physical benchmark, avoiding the scale inconsistency problem caused by measuring only within the pixel grid.

[0044] Furthermore, by using a "sequence" of temperature distribution images instead of single-frame images, and uniformly mapping the boundary coordinates to the sample plane for time-series tracking, the change of boundary displacement over time can be calculated under the same time reference, thereby directly obtaining dynamic quantitative indicators such as wicking rate. This process reduces errors caused by inconsistent time point selection during manual readings, improving the repeatability and comparability of rate assessment. The above three elements—thermal imaging temperature difference information, pixel-to-plane perspective mapping, and time-series tracking of boundary coordinates—work synergistically within the same coordinate system, closing the "identification-conversion-measurement" link: first, the boundary is stably identified in the pixel domain, then mapped to the physical plane to complete geometric quantization, and finally, a rate output is formed on the time axis. From source to result, everything is achieved in a coordinate-based and calculable manner, reducing the randomness brought about by subjective human interpretation.

[0045] In summary, since the acquisition of boundary coordinates and the calculation of the rate rely entirely on the perspective mapping and image sequence established in this embodiment, without depending on additional color enhancement or chemical tracing methods, it is easier to interface with traditional indicators such as wicking height / diffusion distance in a standardized experimental environment, achieving consistent expression and verification of results. Through the above steps and effects, this embodiment completes the entire technical path of "fixing the camera—calibrating the imaging geometry—establishing a dual coordinate system and perspective mapping—acquiring temperature sequences and separating dry and wet conditions—mapping boundaries and time-tracking—outputting quantitative indicators of wicking rate" without introducing other additional conditions, ensuring that the measurement of wicking position and rate is objective, geometrically consistent, and temporally consistent.

[0046] In one specific embodiment, the step of "calibrating the imaging geometry of the sample" is further included before the step of:

[0047] The sample was screened and its standard geometric dimensions were calibrated. The sample and the camera were placed in a stable and uniform environment, the sample posture was fixed, and the ambient temperature was set to 20°C and the ambient humidity to 65%.

[0048] Preferably, in this embodiment, it should be noted that: before performing "calibrating the imaging geometry of the sample," pre-test preparation and environmental readiness are completed. First, the textile fabric sample to be tested is screened for appearance and flatness, and samples with obvious wrinkles, warping, damage, oil stains, or foreign matter are rejected; if necessary, the sample edges are regularized to avoid curling edges interfering with subsequent field of view and boundary identification. Subsequently, standard geometric dimensional features are calibrated in the area coplanar with the sample plane. These features are reference graphics with known dimensions and spacing (e.g., geometric marks with infrared visibility, equidistant dot matrix, or reference lines) to facilitate the subsequent establishment of pixel-physical scale correspondence.

[0049] Specifically, the sample and the camera are placed together in a stable and uniform environment, and the ambient temperature is set and maintained at 20°C and the relative humidity at 65% (which can be set according to actual customer needs). Direct airflow, strong heat sources, or strong reflective objects are avoided from entering the field of view. The placement posture of the sample is fixed (any predetermined posture among horizontal, vertical, or tilted postures), and the installation posture of the camera is fixed, so that the relative positions of the two remain unchanged throughout the entire test cycle. Under dry conditions without the application of liquid, after the temperature and humidity readings and the radiation state of the sample surface tend to stabilize, the imaging geometry calibration step is then performed.

[0050] Specifically, by setting standard geometric dimensions on the coplanar sample plane and ensuring the relative orientation of the sample and the camera remains fixed during subsequent calibration, the pixel-to-physical scale correspondence can be established on a stable spatial basis, reducing systematic errors caused by assembly displacement and ensuring the reusability of calibration results under the same assembly conditions. Pre-test static placement in an environment of 20℃ / 65%RH without strong convection or reflection reduces the impact of environmental radiation background and convective heat transfer fluctuations on thermal image readings, providing a low-noise baseline for subsequent dry / wet region discrimination based on temperature difference. Screening the sample and maintaining its flatness and wrinkle-free surface reduces projection errors caused by non-coplanarity and local warping, making subsequent geometric conversions based on the sample plane more consistent with the applicable conditions of the planar element model.

[0051] The pre-test screening, standard geometric dimension feature calibration, environmental settings, and attitude fixation in this embodiment constitute the necessary prerequisites for subsequent imaging geometric calibration and pixel-to-physical scale conversion, effectively reducing the uncertainty introduced by environmental and assembly factors, and ensuring that subsequent measurements are established on a stable and traceable baseline.

[0052] In one specific embodiment, the step "calibrating the imaging geometry of the sample and establishing image pixel coordinates and sample plane coordinates based on the imaging geometry" specifically includes the following steps:

[0053] The imaging area of ​​the camera is obtained, the standard geometric size features are identified and the imaging geometry is calibrated, the origin and axis of the coordinate system are defined at the location of the sample near the water source, and the plane coordinates of the sample are established.

[0054] Acquire thermal imaging image frames within the imaging area, and establish image pixel coordinates based on the thermal imaging image frames.

[0055] Preferably, in this embodiment, it should be noted that: firstly, the imaging area of ​​the camera is acquired, ensuring its field of view completely covers the location of the water source and the expected diffusion path. Then, standard geometric features are identified in the region coplanar with the sample plane. These features have known dimensions and spacing, facilitating subsequent geometric solutions. Based on these features, a calibration image is acquired, and imaging geometric parameters are calculated. These imaging geometric parameters include at least one or more of intrinsic parameters, distortion parameters, and relative pose, used to describe the spatial relationship between the camera and the sample plane. Based on the determined imaging geometry, a point near the water source on the sample surface is selected as the origin of the sample plane coordinates, and coordinate axes are defined according to the longitude / latitude direction or a reference edge. The sample plane coordinates are established accordingly and used as a unified benchmark for subsequent physical quantity measurements.

[0056] Furthermore, without changing the relative posture of the camera and the sample, thermal imaging image frames within the imaging area are acquired. An image pixel coordinate system is established based on these image frames, with the upper left corner of the image frame as the origin, the horizontal axis as the u-axis, and the vertical axis as the v-axis, with units in pixels. When fine positioning of boundaries or calibration features is required, pixel-level or sub-pixel-level feature point extraction can be performed within this coordinate system. Thus, this embodiment completes the definition and readiness of two coordinate systems: "sample plane coordinates" and "image pixel coordinates," providing the prerequisites for subsequent establishment and quantization calculations of coordinate relationships.

[0057] Specifically, by identifying standard geometric dimensions and calibrating the imaging geometry, the intrinsic imaging parameters of the camera and its external orientation relative to the sample plane were clarified, allowing subsequent geometric conversions to be based on a reliable spatial model. After determining the imaging geometry, an origin was set on the sample surface near the water source, and an axis was defined. The sample plane coordinates were directly aligned with the experimental physical process (wicking extending outward from the water source), allowing any subsequent boundary position to be intuitively measured relative to the zero point of the water source in the physical domain. A pixel coordinate system was established using thermal imaging image frames as a reference, unifying the data acquisition and positioning rules in the image domain. This ensured consistency and reproducibility in the positional representation of boundary points, calibration points, etc., in the digital image, providing a stable coordinate framework for feature reading and comparison on any subsequent frame.

[0058] Furthermore, two separate systems are established: the sample plane coordinate system and the image pixel coordinate system. This ensures that the physical domain and the image domain are independent and complete, facilitating the establishment of a clear input-output relationship between the two domains in subsequent steps and avoiding inconsistencies in geometric conversions caused by ambiguity in definitions. Under the premise that the relative orientation of the camera and the sample remains unchanged, the imaging geometry and the settings of the two coordinate systems can be reused under the same assembly conditions. This is beneficial for maintaining consistent geometric benchmarks in batch samples or repeated tests, improving the comparability and repeatability of results.

[0059] In one specific embodiment, the step "establishing the mapping relationship between the image pixel coordinates and the sample plane coordinates" specifically includes the following steps:

[0060] Identify at least four sets of non-collinear physical coordinate points of the sample plane coordinates, obtain pixel coordinates in the image, pair the pixel coordinates with the physical coordinates, establish pixel-physical correspondence points, and calculate the perspective mapping relationship between the sample plane coordinates and the corresponding coordinates of the image pixel coordinates based on the pixel-physical correspondence points.

[0061] Preferably, after establishing the sample plane coordinate system and the image pixel coordinate system, at least four sets of physical coordinate points located in the sample plane and not collinear are obtained. These are preferably given by standard geometric features of known dimensions (such as the center of a calibration matrix, checkerboard corner points, or intersections of reference lines), and are evenly distributed within the field of view to enhance numerical stability. Simultaneously, the corresponding pixel coordinates are extracted from the thermal imaging image frame, and sub-pixel fitting can be performed on the corner points / centers to reduce positioning errors. The above physical-pixel pair data are combined to form a set of pixel-physical corresponding points. Under the premise that the sample area is approximately coplanar, a planar perspective (homography) model is adopted.

[0062] Furthermore, based on direct linear transformation or least squares, robust estimation is used to obtain the homography matrix; subsequently, the inverse homography matrix of its inverse transformation is calculated for coordinate conversion in the pixel-to-sample plane direction. To verify the reliability of the mapping, independent check points or known lengths are selected to check reprojection error and length error; when the error meets a predetermined threshold, the homography matrix and inverse homography matrix, along with the corresponding coordinate system information, are permanently stored and reused when the relative posture of the image and the sample remains unchanged; if the error exceeds the limit, corresponding points can be added or the extraction quality can be checked and the solution recalculated. The above process ensures that the perspective mapping relationship of "image pixel coordinates - sample plane coordinates" is uniquely and stably established.

[0063] Specifically, homography mapping maps pixel-domain coordinates one-to-one to sample plane coordinates, achieving a unified benchmark for physical quantities such as millimeters / centimeters and offsetting the impact of perspective distortion caused by shooting angle and distance on geometric quantities such as length and area. Based on no fewer than four sets of non-collinear corresponding points, supplemented by sub-pixel fitting, normalization, and outlier removal, the homography matrix and inverse homography matrix are solidified and reused under the condition of unchanged relative pose, reducing the cost of repeated calibration; different test batches are conducted under the same mapping benchmark, improving the horizontal comparability and repeatability of the results.

[0064] Furthermore, the computational link is closed: this mapping precisely couples "image domain localization" with "physical domain measurement", providing a direct and traceable coordinate transformation channel for any subsequent geometric features that need to be expressed on the sample plane (such as distance and area calculation of boundary point sets), thereby avoiding systematic bias caused by direct pixel-scale measurement.

[0065] In one specific embodiment, the step of "pairing the pixel coordinates with the physical coordinates and establishing the pixel physical correspondence point" further includes the following step:

[0066] The distortion parameters in the imaging geometry are calculated based on the physical corresponding points of the pixels. The thermal imaging frame is remapped according to the distortion parameters to obtain the corrected pixel coordinates. The perspective mapping relationship is calculated based on the corrected pixel coordinates and the sample plane coordinates.

[0067] Preferably, in this embodiment, it should be noted that after completing the process of "pairing the pixel coordinates with the physical coordinates and establishing pixel-physical correspondence points", the set of corresponding points is used as calibration observation data to calculate the distortion parameters in the imaging geometry. Specifically, standard geometric calibration feature points (such as the center of a dot matrix, a checkerboard corner point, or the intersection of reference lines) located in the sample plane are selected, their pixel coordinates in the thermal imaging image are read, and they are paired with their known physical coordinates in the sample plane to form a set of corresponding points.

[0068] Furthermore, based on this set, a camera calibration algorithm is used to obtain the lens distortion parameters, which include at least radial distortion coefficients and / or tangential distortion coefficients. After obtaining the distortion parameters, pixel-by-pixel remapping processing is performed on the thermal imaging image frames to generate a distortion-corrected image, and the corresponding corrected pixel coordinates are obtained on the corrected image. Subsequently, without changing the relative orientation of the camera and the sample, the pixel positions of the calibration features are re-extracted from the corrected image and paired again with the existing physical coordinates of the sample plane. Based on this, the perspective mapping relationship between the image pixel coordinates and the sample plane coordinates is calculated.

[0069] Furthermore, to ensure reliability, this embodiment verifies the corrected mapping results, including conversion and comparison using independent known lengths, known spacings, or test point positions; when the error meets the preset threshold, the distortion parameters, remapping table, and perspective mapping relationship are solidified and stored, and reused in subsequent tests under the condition that the relative posture of the camera and the sample remains unchanged; if the verification fails, the corresponding points can be supplemented or replaced and recalculated until the requirements are met.

[0070] Specifically, distortion parameters are first obtained and distortion correction is performed, followed by perspective mapping. This effectively eliminates geometric deviations caused by lens distortion and viewing angle before mapping is established, thereby reducing systematic errors in physical quantities such as length and area. After distortion correction, the geometric relationship between the image center and edges is more consistent, and the perspective mapping exhibits better linearity and stability throughout the entire field of view, avoiding metric drift caused when boundary points are at the edge of the image. By verifying and solidifying the distortion parameters, remapping table, and perspective mapping relationship, they can be directly reused when the relative pose remains unchanged, reducing the workload of repeated calibration and ensuring consistency of geometric benchmarks across different batches of measurements.

[0071] In one specific embodiment, the imaging area includes the water source and the expected diffusion area of ​​the sample.

[0072] Specifically, the imaging area refers to the effective field of view of the camera during measurement. In this embodiment, the imaging area is planned as the expected diffusion area that simultaneously covers both the water source and the sample. The water source refers to the initial contact point between the liquid and the sample, such as the drop point or the inlet line; the expected diffusion area is determined based on the sample shape, placement posture (horizontal, vertical, or inclined), and previous experimental experience or pre-experiment results. It is the range that water may reach from the water source along the longitudinal / latitudinal or oblique direction, and an appropriate margin is reserved around this range to avoid boundary overflow.

[0073] Furthermore, to achieve the aforementioned coverage, it is preferable to first complete the framing and positioning in a dry state: determine the fixed reference position of the water source in the image (to facilitate subsequent measurements based on the water source), and then select the working distance and focal length combination of the camera accordingly, so that the entire diffusion path and its safety margin are within the same frame; when the sample is elongated and the diffusion direction is clear, the field of view can be expanded along the main diffusion axis. When the sample size is large or the diffusion path is long and the single field of view is insufficient, a single field of view coverage priority strategy can be adopted (selecting a long-distance framing that covers the entire area), or segmented close-up scanning can be implemented while maintaining the installation posture unchanged, but it must be ensured that each framing simultaneously includes the water source and its adjacent diffusion front.

[0074] Furthermore, the entire diffusion process occurs within the same imaging area, ensuring that the wet and dry boundaries are not lost even when the image extends beyond the frame, thus reducing geometric inconsistencies and temporal discontinuities caused by moving the viewfinder or stitching. The water source remains consistently within the frame, facilitating distance and path measurements using its location as a stable reference, reducing systematic errors caused by missing or drifting reference points. The diffusion front appears continuously within a unified field of view, allowing for complete acquisition of the temporal sequence of the boundary positions, minimizing data gaps caused by field-of-view switching, and improving the reliability and repeatability of dynamic indicators such as rate.

[0075] In one specific embodiment, the step "acquiring an image sequence of temperature distribution, identifying the dry area and the wet area through the image sequence, and separating the dry and wet boundaries" includes the following specific steps:

[0076] A preset temperature threshold is set, and a temperature reference value for the dry zone is obtained. Based on the temperature reference value and the temperature threshold, a separation temperature value is calculated to identify the wet zone and define the dry-wet boundary.

[0077] Specifically, in this embodiment, it should be noted that the implementation of "collecting image sequences of temperature distribution and identifying the dry and wet areas through the image sequences" employs a segmentation mechanism combining a preset temperature threshold and a dry area temperature reference value. Specifically:

[0078] Before the experiment begins, select a dry reference area (which may be one or more small areas) in the image pixel coordinate system that is far away from the water source and least affected by the liquid. Read the pixel temperature from this area for each frame and calculate a representative dry temperature baseline value (preferably using robust statistics, such as the median or truncated mean, and a small amount of time smoothing can be added to suppress instantaneous fluctuations).

[0079] The temperature threshold parameter is preset according to the sample type and environmental conditions. This parameter is used to define the "lowest resolvable temperature difference between the dry area and the wet area". In order to take into account different fabrics and diffusion rates, two configurations can be provided: a fixed dimension (temperature difference) or a relative dimension related to the temperature fluctuation of the dry area. However, in this embodiment, only one of the preset strategies is enabled.

[0080] For each frame in the sequence, the above-mentioned dry area temperature reference value is combined with the preset temperature threshold parameter to form the separation temperature value of the frame; this value serves as the threshold for pixel-level discrimination, used to divide pixels into "below the threshold" (candidate wet area) and "above the threshold" (candidate dry area) within the frame.

[0081] The temperature matrix of the entire frame is discriminated pixel by pixel based on the temperature values ​​of the separation to obtain the initial wet area mask and dry area mask; then connected component filtering (prioritizing the retention of the main connected components connected to or closest to the water source), morphological opening and closing operations and pinhole filling are performed to remove isolated noise and jagged artifacts, and obtain stable segmentation results of the wet and dry areas.

[0082] Extract the wet and dry boundary curves on the refined mask; to reduce inter-frame jitter, threshold hysteresis (using slightly different thresholds for entering and exiting the wet region) and maximum displacement constraint (limiting the maximum movement of the boundary in two adjacent frames) can be introduced to ensure that the boundary advances continuously and monotonically along the diffusion direction.

[0083] For each frame, the baseline temperature value of the dry area, the separating temperature value, and the boundary pixel coordinates are recorded, along with a timestamp, forming the frame-level record required for subsequent time-series analysis. This completes the identification and boundary definition of dry / wet regions in the pixel domain, providing direct input for subsequent coordinate mapping and rate calculation.

[0084] In summary, using temperature as the sole criterion avoids interference from differences in color, brightness, and texture under visible light conditions, ensuring consistent identification of fabrics with different colors and structures under the same discrimination rule. Each frame calculates a segmentation temperature value based on the current dry area temperature baseline, which can offset the overall shift caused by slow background changes and environmental perturbations, maintaining the stability of wet area identification. Through threshold hysteresis, connected component constraints, and maximum displacement constraints, inter-frame boundary jitter and misjudgments of "isolated wet points" are significantly suppressed, resulting in continuous and physically reasonable boundary trajectories, facilitating subsequent temporal tracking.

[0085] In one specific embodiment, the step "converting the wet and dry boundary into boundary coordinates based on the mapping relationship" includes the following specific steps:

[0086] Based on the perspective mapping relationship, the pixel coordinates are mapped to the sample plane coordinates to form a boundary point set. The boundary point set is then moved to align with the origin and axis of the sample plane coordinates to output the boundary coordinates corresponding to each image frame.

[0087] Specifically, in this embodiment, it should be noted that, given the established perspective mapping relationship between image pixel coordinates and sample plane coordinates, coordinate transformation and alignment are performed on the obtained wet and dry boundaries (pixel domain representation) for each frame. The specific process is as follows: First, the pixel-level point set of the wet and dry boundaries for that frame is read, and the necessary point ordering and refinement are performed in the image pixel coordinate system to obtain continuous boundary pixel polylines. Then, the perspective mapping relationship is invoked to map each boundary pixel point to the sample plane coordinate system, generating the original physical boundary point set for that frame.

[0088] Furthermore, using the origin of the sample plane coordinates as a reference, the physical boundary point set is translated as a whole, so that the water source reference point corresponds to the zero point. Then, the coordinate axes are aligned according to the defined axes, so that the main diffusion direction is consistent with the main axis of the sample plane coordinates, with the negative direction pointing towards the water source and the positive direction pointing towards the diffusion direction. Finally, the boundary coordinates of this frame are output in a unified data structure, including: timestamp, ordered list of boundary point coordinates, coordinate system definition and version information (origin, axis, scale factor, mapping relationship identifier), etc. If necessary, derived single-value quantities (such as the farthest / nearest distance along the main axis, the area enclosed by the boundary, etc.) are also saved for subsequent time series calculations and verification.

[0089] Furthermore, after transforming the boundary from the pixel domain to the sample plane and aligning the origin and axis, quantitative indicators such as position, distance, and area can be directly calculated under a unified physical coordinate system, avoiding scale inconsistencies caused by pixel units. Perspective mapping enables geometric correction of viewpoint and distance differences, significantly reducing projection deviations of the same boundary in different regions and frames, resulting in boundary morphology and position closer to the true physical distribution. All frames output boundary coordinates with a fixed origin and fixed axis, eliminating the need for coordinate conversion between frames. The boundary advancement trajectory is continuously traceable within the physical domain, facilitating robust estimation of subsequent dynamic indicators such as rate.

[0090] In one specific embodiment, the step "converting the wet-dry boundary into boundary coordinates based on the mapping relationship, tracking the boundary coordinates in the image sequence in time sequence, and calculating the quantitative index of the wicking rate of the sample" further includes the step of:

[0091] The camera moves along the preset path and continuously identifies the wet and dry boundaries. Based on the scanning offset distance of the camera, it performs translation compensation on the local sample plane coordinates, continuously outputs the boundary coordinates with timestamps, and calculates the quantitative index of the wicking rate of the sample based on the time series of the boundary coordinates.

[0092] The preset path is parallel to the sample surface.

[0093] Preferably, in this embodiment, it should be noted that: before the experiment begins, imaging geometry calibration is completed to obtain the perspective mapping relationship for pixel-sample plane conversion and the intrinsic and distortion parameters of the camera; subsequently, during the acquisition phase, the camera only performs translational scanning along a preset path parallel to the sample surface, without changing the imaging posture and focal length. Under this premise, the lens imaging model remains unchanged, the distortion parameters and intrinsic parameters can be directly reused, and the perspective mapping relationship is also steadily applicable under the same posture system (only translation compensation needs to be applied to different imaging positions), without the need for recalibration.

[0094] Furthermore, thermal imaging frames are acquired at each imaging position, and the wet / dry boundary (pixel domain) is identified. Based on a predetermined mapping relationship, the boundary pixels are converted into local sample planar coordinates. Translation compensation is applied to the local coordinates according to the scan offset distance recorded in this frame, and the boundary coordinates and timestamp of this frame are output with a unified origin and axis. Throughout the entire detection process, there is no need to stitch the entire image together; only coordinate-level fusion and temporal management are performed on the boundary coordinates. This achieves the imaging strategy of "high precision in a small area with single-shot capture, and the ability to move between frames to cover a larger test area," while ensuring that the scan offset is included in subsequent quantization calculations.

[0095] Specifically, since the scanning is translational and the attitude / focal length remains constant, the distortion parameters and perspective mapping relationship are reused, avoiding errors and overhead caused by frequent calibration, while ensuring that the geometric measurements at different imaging positions are on the same physical reference. A small field of view and high pixel density are used in a single frame to improve the accuracy of dry and wet boundary positioning; parallel translational scanning gradually covers the entire diffusion path of slender samples, achieving a balance between high precision and large range. Continuous output of timestamped boundary coordinates allows for direct temporal tracking within a unified coordinate system, calculating the change in boundary position over time to obtain a quantitative index of wicking rate; because steady-state reuse of geometric parameters and offset compensation are completed, rate evaluation is insensitive to the movement process, improving repeatability and comparability.

[0096] In one specific embodiment, steps S10, S20, and S30 are repeated to select an average quantization index to establish an accurate wicking rate quantization index.

[0097] Specifically, in this embodiment, it should be noted that this embodiment revolves around the complete process of "fixing the camera—calibrating the sample imaging geometry—establishing the image pixel coordinate system and the sample plane coordinate system and establishing the perspective mapping relationship—acquiring image sequences of temperature distribution—identifying dry and wet areas in the image sequence and separating the dry and wet boundaries—converting the dry and wet boundaries into boundary coordinates based on the mapping relationship—time-series tracking of boundary coordinates and calculating the wicking rate quantification index," performing multiple independent detections and selecting the average quantification index based on the results of multiple detections. Specifically:

[0098] Without changing the camera mounting posture and detection settings, S10, S20 and S30 are executed sequentially according to the same process as the first detection to obtain the wicking rate quantification index and the corresponding boundary coordinate time series of the first detection.

[0099] Replace the sample with a new one or restore the sample to its initial dry state conditions, and repeat the above process under the same settings to obtain the quantification index and time series of the wicking rate for the second test.

[0100] Complete the third and subsequent tests in the same way to ensure that each test generates a wicking rate quantification index using the same framing, coordinate system and discrimination rules.

[0101] In summary, performing multiple independent tests under the same process and unified coordinate / mapping benchmark, and then averaging them point by point after time alignment, can effectively reduce random noise and local jitter in a single test, making the quantification index of core absorption rate closer to the stable characteristics of the sample. Averaging multiple tests homogenizes the discrete effects caused by unstable identification of individual frame boundaries and instantaneous temperature disturbances; subsequent retests under the same settings result in an average curve with smaller deviations from key values, which is more conducive to batch-to-batch comparisons and quality assessment.

[0102] A thermal imaging-based textile fabric wicking rate detection device applies all the steps of the thermal imaging-based textile fabric wicking rate detection method described above. Since this embodiment includes all the detection methods of the above embodiments, it has all the beneficial effects of the above embodiments, and will not be repeated here.

[0103] The above description is merely an embodiment of this application. It should be noted that those skilled in the art can make improvements without departing from the inventive concept of this application, but these improvements all fall within the protection scope of this application.

Claims

1. A method for detecting the wicking rate of textile fabrics based on thermal imaging, characterized in that, include: The initial installation posture of the fixed camera is fixed and it can move along a preset path to calibrate the imaging geometry of the sample. Based on the imaging geometry, an image pixel coordinate system and a sample plane coordinate system are established, and the perspective mapping relationship between the image pixel coordinates and the sample plane coordinates is established. The dry areas of the sample are designated as dry zones, and the wet areas are designated as wet zones. The sample is brought into contact with the liquid, and an image sequence of the temperature distribution is acquired. The dry zones and wet zones are identified through the image sequence, and the dry-wet boundary is separated. Based on the mapping relationship, the dry and wet boundary is converted into boundary coordinates, the boundary coordinates in the image sequence are tracked in time, and the quantitative index of the wicking rate of the sample is calculated.

2. The method for detecting the wicking rate of textile fabrics based on thermal imaging according to claim 1, characterized in that, The step "calibrate the imaging geometry of the sample" includes the following steps: The sample is screened and its standard geometric dimensions are calibrated. The sample and the camera are placed in a stable and uniform environment, the sample posture is fixed, and the preset ambient temperature and humidity are set.

3. The method for detecting the wicking rate of textile fabrics based on thermal imaging according to claim 2, characterized in that, The step "calibrate the imaging geometry of the sample, and establish the image pixel coordinates and sample plane coordinates based on the imaging geometry" specifically includes the following steps: The imaging area of ​​the camera is obtained, the standard geometric size features are identified and the imaging geometry is calibrated, the origin and axis of the coordinate system are defined at the location of the sample near the water source, and the plane coordinates of the sample are established. Acquire thermal imaging image frames within the imaging area, and establish image pixel coordinates based on the thermal imaging image frames.

4. The method for detecting the wicking rate of textile fabrics based on thermal imaging according to claim 3, characterized in that, The step "establishing the mapping relationship between the image pixel coordinates and the sample plane coordinates" specifically includes the following steps: Identify at least four sets of non-collinear physical coordinate points of the sample plane coordinates, obtain pixel coordinates in the image, pair the pixel coordinates with the physical coordinates, establish pixel-physical correspondence points, and calculate the perspective mapping relationship between the sample plane coordinates and the corresponding coordinates of the image pixel coordinates based on the pixel-physical correspondence points.

5. The method for detecting the wicking rate of textile fabrics based on thermal imaging according to claim 4, characterized in that, The step "pairing the pixel coordinates with the physical coordinates and establishing the physical correspondence points of the pixels" is followed by the following step: The distortion parameters in the imaging geometry are calculated based on the physical corresponding points of the pixels. The thermal imaging frame is remapped according to the distortion parameters to obtain the corrected pixel coordinates. The perspective mapping relationship is calculated based on the corrected pixel coordinates and the sample plane coordinates.

6. The method for detecting the wicking rate of textile fabrics based on thermal imaging according to claim 1, characterized in that, The imaging area includes the water source and the expected diffusion area of ​​the sample.

7. The method for detecting the wicking rate of textile fabrics based on thermal imaging according to claim 1, characterized in that, The step "acquiring an image sequence of temperature distribution, identifying the dry and wet regions through the image sequence, and separating the dry and wet boundaries" includes the following specific steps: A preset temperature threshold is set, and a temperature reference value for the dry zone is obtained. Based on the temperature reference value and the temperature threshold, a separation temperature value is calculated to identify the wet zone and define the dry-wet boundary.

8. The method for detecting the wicking rate of textile fabrics based on thermal imaging according to claim 7, characterized in that, The step "converting the wet-dry boundary into boundary coordinates based on the mapping relationship, tracking the boundary coordinates in the image sequence over time, and calculating the quantitative index of the wicking rate of the sample" further includes the following steps: The camera moves along the preset path and continuously identifies the wet and dry boundaries. Based on the scanning offset distance of the camera, it performs translation compensation on the local sample plane coordinates, continuously outputs the boundary coordinates with timestamps, and calculates the quantitative index of the wicking rate of the sample based on the time series of the boundary coordinates. The preset path is parallel to the sample surface.

9. The method for detecting the wicking rate of textile fabrics based on thermal imaging according to claim 1, characterized in that, The step "converting the wet and dry boundary into boundary coordinates based on the mapping relationship" includes the following specific steps: Based on the perspective mapping relationship, the pixel coordinates are mapped to the sample plane coordinates to form a boundary point set. The boundary point set is then moved to align with the origin and axis of the sample plane coordinates to output the boundary coordinates corresponding to each image frame.

10. The textile fabric wicking rate detection device based on thermal imaging according to claim 1, characterized in that, The detection equipment is applied to a method for detecting the wicking rate of textile fabrics based on thermal imaging, as described in any one of claims 1-9.

Images