Method for detecting the absorption rate of a water-absorbing mat based on infrared image technology

Abstract

This invention discloses a method for detecting the absorption rate of absorbent pads based on infrared imaging technology, belonging to the field of absorbent pad detection technology. The method involves acquiring a sequence of multiple infrared thermal images of the absorbent pad during liquid absorption using an infrared thermal imager; based on the temperature distribution in each frame of the infrared thermal image, obtaining the temperature value of each pixel on the surface of the absorbent pad and tracking the dynamic thermal response curve of the temperature value changing over time; constructing the heat absorption response gradient field of the absorbent pad at different times based on the temporal variation characteristics of the dynamic thermal response curve; extracting the liquid diffusion front contour line of the absorbent pad based on the spatial distribution of the heat absorption response gradient field; and determining the absorption rate of the absorbent pad based on the displacement of the liquid diffusion front contour line between adjacent frames. This method utilizes infrared thermal imaging to non-contactly capture the heat absorption changes caused by liquid diffusion, ensuring the objectivity of the absorption rate detection.

Description

Technical Field

[0001] This invention relates to the field of absorbent pad testing technology, specifically a method for detecting the absorption rate of absorbent pads based on infrared imaging technology. Background Technology

[0002] The absorption rate of absorbent pads is a key indicator for evaluating their liquid absorption performance, directly affecting their applicability in medical, hygiene, and cleaning fields. Existing methods for detecting the absorption rate of absorbent pads largely rely on visible light imaging technology combined with dyeing liquids. This involves adding dye to the absorbent pad and continuously capturing color images, then using image processing algorithms to identify changes in the boundary of the dyed area over time, thereby calculating the liquid diffusion rate. This method requires adding dye to the liquid to enhance image contrast, which is not only cumbersome but also potentially alters the surface tension and viscosity of the liquid, affecting the true representation of the absorption process. Furthermore, visible light imaging is sensitive to lighting conditions; the color, texture, and surface undulations of the absorbent pad itself can interfere with image segmentation accuracy, leading to unstable diffusion front extraction. Another known method uses an infrared thermal imager to monitor changes in the surface temperature of the absorbent pad. By setting a fixed temperature threshold, it directly divides the wetted and unwetted areas, using a binarized temperature image to track the liquid diffusion range. While this method avoids dyeing interference, relying solely on the temperature amplitude threshold is easily affected by ambient temperature fluctuations, the thermal properties of the absorbent pad material itself, and initial temperature differences within the pad. Near the liquid diffusion front, there is often a temperature transition zone caused by thermal conduction. A fixed threshold is difficult to accurately define the true liquid invasion boundary, which can easily cause boundary jumps or fragmentation. This makes the reliability of the front displacement calculated frame by frame insufficient, ultimately affecting the repeatability and accuracy of the absorption rate detection value.

[0003] Based on the above problems, it is necessary to provide a detection method that can accurately characterize the dynamic temperature changes during liquid diffusion without introducing staining interference, and stably extract the true diffusion front, so as to solve the problems of front positioning deviation and unstable absorption rate calculation caused by thermal conduction ambiguity and threshold misjudgment. Summary of the Invention

[0004] This invention aims to provide a method for detecting the absorption rate of absorbent pads based on infrared imaging technology, in order to solve the problems in existing detection methods that rely on dyeing liquids and visible light images, which are easily affected by light and the color and texture of the absorbent pad itself, as well as the unstable boundary extraction and large deviation in absorption rate calculation caused by directly extracting the diffusion front based on temperature thresholds due to the influence of the thermal conduction transition zone.

[0005] To achieve the above objectives, the present invention provides the following technical solution: The present invention provides a method for detecting the absorption rate of an absorbent pad based on infrared imaging technology. This method utilizes the minute temperature changes accompanying the liquid absorption process and achieves non-contact, precise measurement of the absorption rate by analyzing infrared thermal image sequences. The steps include: acquiring a continuous multi-frame infrared thermal image sequence of the absorbent pad during the liquid absorption process using an infrared thermal imager to obtain the spatiotemporal temperature distribution information of the entire absorption process.

[0006] Based on the temperature distribution reflected in each frame of the infrared thermal image, the temperature values ​​of each pixel on the surface of the absorbent pad are obtained, and the dynamic thermal response curve of the temperature value at each pixel location over time is tracked. In a preferred embodiment, when acquiring the dynamic thermal response curve, pixel-level spatial registration is first performed on each frame of the infrared thermal image sequence to ensure strict alignment of the absorbent pad's spatial position in each frame. The temperature values ​​of all pixels in the area where the absorbent pad is located are extracted from each aligned frame. The temperature values ​​of the same pixel location in consecutively arranged frames are combined in chronological order to form a time series of temperature values ​​for that pixel location. This time series serves as the dynamic thermal response curve for the corresponding pixel location. This method eliminates positional deviations caused by pad movement or camera shake, ensuring accurate tracking of the temperature change trajectory of each pixel.

[0007] Based on the obtained dynamic thermal response curves, temporal variation features are extracted to construct the endothermic response gradient field of the absorbent pad at different times. As a technical solution of this invention, the process of constructing the endothermic response gradient field includes: calculating the first-order time-domain derivative of the dynamic thermal response curve at each pixel location at each frame time, as the instantaneous heating rate of that pixel location at that frame time; calculating the second-order time-domain derivative of the dynamic thermal response curve at each pixel location at each frame time, as the heating acceleration of that pixel location at that frame time; determining the endothermic response intensity value of each pixel location at each frame time based on the instantaneous heating rate and heating acceleration; arranging the endothermic response intensity values ​​of all pixels within the absorbent pad region in the same frame time according to spatial coordinates to form the endothermic response gradient field for that frame time. The endothermic response intensity value can simultaneously reflect the rate and strength of temperature change caused by liquid diffusion, and is more sensitive to capturing the movement of the liquid front than simply relying on temperature values ​​or simple temperature differences.

[0008] When determining the endothermic response intensity value, a weighted fusion and normalization process is preferred: the instantaneous heating rate is multiplied by a preset heating rate weighting coefficient to obtain the weighted heating rate, and the heating acceleration is multiplied by a preset heating acceleration weighting coefficient to obtain the weighted heating acceleration. The two are added together to obtain the initial endothermic response intensity value. Then, the initial endothermic response intensity values ​​of all pixels at all frame times are normalized to a preset numerical range, which is used as the final endothermic response intensity value. Through weighted normalization, the units can be unified and the comparability of response features can be enhanced, so that the detection results at different times and in different batches have a consistent evaluation benchmark.

[0009] Based on the spatial distribution of the endothermic response gradient field, the liquid diffusion front contour of the absorbent pad is extracted. Specifically, spatial gradient magnitude calculation is performed on the endothermic response gradient field at each frame time to obtain the gradient magnitude field at each frame time; the pixel positions in the gradient magnitude field with gradient magnitude greater than a preset magnitude threshold are marked as candidate front pixels; connected component analysis is performed on all candidate front pixels to extract the outer boundary line of the connected component with the largest area, and this outer boundary line is used as the liquid diffusion front contour of that frame time. A preferred method for determining the preset magnitude threshold is to use the sum of the mean and standard deviation of the gradient magnitudes of all pixels in the gradient magnitude field at each frame time as the magnitude threshold for that frame time, and the threshold for different frames time times is determined independently. This adaptive threshold strategy can adapt to the changes in the overall magnitude of the endothermic response gradient field over time, ensuring the stability and accuracy of the front contour extraction. When calculating the gradient magnitude field, for each pixel location, the spatial partial derivatives of its endothermic response intensity value in the horizontal and vertical directions are calculated. The square root of the sum of the squares of the partial derivatives in the two directions is used to obtain the gradient magnitude at that pixel location. Arranging the gradient magnitudes of all pixels at the same frame according to their spatial coordinates constitutes the gradient magnitude field. The extracted liquid diffusion front contour clearly depicts the dynamic boundary of the liquid spreading in the absorbent pad.

[0010] After obtaining the liquid diffusion front contour, the absorption rate of the absorbent pad is determined based on the displacement of this contour between adjacent frames. The process is as follows: calculate the geometric center coordinates of the liquid diffusion front contour at each frame; calculate the Euclidean distance between the geometric center coordinates of two adjacent frames in time, as the front displacement between adjacent frames; divide the front displacement between each adjacent frame by the corresponding time interval between adjacent frames to obtain the interval diffusion rate of each adjacent frame interval; use the arithmetic mean of the interval diffusion rates of all adjacent frame intervals as the absorption rate of the absorbent pad. When calculating the geometric center coordinates, obtain the spatial coordinate set of all pixels enclosed by the contour, and calculate the arithmetic mean of the horizontal and vertical coordinate values ​​of all pixels in the set; these two values ​​constitute the geometric center coordinates. By focusing on the displacement of the geometric center of the front contour, the overall trend of liquid diffusion in all directions can be comprehensively reflected, overcoming the deviation that may be caused by measurement in a single direction. Furthermore, before calculating the arithmetic mean of the interval diffusion rates, abnormal interval diffusion rates that exceed the preset upper limit threshold or fall below the preset lower limit threshold in all adjacent frame interval diffusion rates are removed, thereby eliminating abnormal displacement data caused by occasional factors such as image noise and instantaneous disturbances, making the final absorption rate more robust and reliable.

[0011] This invention, through pixel-level dynamic analysis and multi-level spatiotemporal feature extraction of infrared thermal image sequences, can extract the diffusion front and its movement rate, which are essential to the absorption process of the reaction liquid, from the original temperature information step by step. The entire detection process does not require contact with the absorbent pad and is not affected by the material color or surface condition. It has high sensitivity and high repeatability and is suitable for rapid and objective evaluation of the absorption performance of various liquid-absorbing materials.

[0012] The technical effects and advantages provided by the present invention in the above technical solution are as follows: After recording a multi-frame temperature image sequence of the absorbent pad's liquid absorption process using an infrared thermal imager, a dynamic thermal response curve showing the temperature change over time was constructed for each pixel. Based on this, the first and second time-domain derivatives of the dynamic thermal response curve for each pixel at each frame were calculated, serving as the instantaneous heating rate and heating acceleration, respectively. These two derivatives were then weighted and synthesized to obtain the endothermic response intensity value for each pixel at each time point. On one hand, the instantaneous heating rate reflects the speed of temperature change caused by liquid wetting, highlighting the active endothermic region; on the other hand, the heating acceleration reflects the dynamic trend of temperature change, distinguishing between continuous heating caused by actual liquid absorption and slow temperature drift caused by heat conduction. Combining these two factors to construct an endothermic response gradient field ensures that the endothermic response intensity value not only exhibits a high response at the point of direct liquid wetting but also effectively suppresses the static background region far from the diffusion front and the low-intensity response disturbance caused by simple thermal diffusion. The resulting endothermic response gradient field, compared to the original temperature field, more concentratedly reflects the dynamic heat exchange region caused by liquid diffusion, providing a feature field with strong anti-interference capabilities for extracting the true diffusion front. After obtaining the endothermic response gradient field for each frame, the spatial gradient magnitude is calculated to obtain the gradient magnitude field. The sum of the mean and standard deviation of the gradient magnitudes of all pixels in each frame's gradient magnitude field is used as an adaptive magnitude threshold to mark candidate front pixels. Then, the outer boundary line of the connected domain with the largest area is extracted as the liquid diffusion front contour line through connected component analysis. Pixels with drastic changes in endothermic response intensity are identified as diffusion fronts. The adaptive threshold can automatically adjust according to the distribution characteristics of the response field at different times, avoiding over-extraction or under-extraction of fixed thresholds at different diffusion stages. The processing method of selecting the outer boundary of the largest connected domain excludes sporadic high-gradient response regions formed by differences in the local fiber structure of the absorbent pad or seepage through liquid microchannels, ensuring that the final front contour line is consistent, continuous, and corresponds to the main liquid diffusion region. The diffusion rate of the interval is calculated based on the geometric center displacement of the leading edge contour line, and the arithmetic mean is taken as the absorption rate of the absorbent pad. After eliminating abnormal jump frame intervals, the obtained absorption rate value can better reflect the overall average liquid absorption speed of the absorbent pad, and the repeatability and reliability of the test results are enhanced. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0014] Figure 1 This is a flowchart of a water-absorbing pad absorption rate detection method based on infrared imaging technology; Figure 2 This is a flowchart for obtaining the dynamic thermal response curves of each pixel on the surface of the absorbent pad; Figure 3 This is a flowchart of the endothermic response gradient field construction process; Figure 4 This is a flowchart of liquid diffusion front contour extraction based on endothermic response gradient field; Figure 5 This is a flowchart for calculating the absorption rate of an absorbent pad; Figure 6 This is a flowchart of the process for removing outliers in the diffusion rate and calculating the absorption rate. Detailed Implementation

[0015] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0016] See Figure 1 This invention provides a method for detecting the absorption rate of an absorbent pad based on infrared imaging technology. The method includes the following steps: acquiring a multi-frame sequence of infrared thermal images of the absorbent pad during liquid absorption using an infrared thermal imager; obtaining the temperature value of each pixel on the surface of the absorbent pad based on the temperature distribution in each frame of the infrared thermal image, and tracking the dynamic thermal response curve of the temperature value changing over time; constructing the heat absorption response gradient field of the absorbent pad at different times based on the temporal variation characteristics of the dynamic thermal response curve; extracting the liquid diffusion front contour line of the absorbent pad based on the spatial distribution of the heat absorption response gradient field; and determining the absorption rate of the absorbent pad based on the displacement of the liquid diffusion front contour line between adjacent frames.

[0017] Example 1: See Figure 2 The process of acquiring the temperature values ​​of each pixel on the surface of the absorbent pad and tracking the dynamic thermal response curve of the temperature values ​​over time is as follows: In the specific implementation, pixel-level spatial registration is performed on each frame of the infrared thermal image sequence. Using the first frame as a reference frame, the boundary contour of the absorbent pad is automatically detected, and multiple corner features are extracted as registration control points based on this contour. For each subsequent frame, a normalized cross-correlation matching algorithm is used to locate the corresponding source points in the reference frame. Based on these source point pairs, the two-dimensional affine transformation parameters between adjacent frames are estimated. These parameters include horizontal translation, vertical translation, rotation angle, and scaling factor. Using the estimated affine transformation parameters, a bilinear interpolation algorithm is used to perform a spatial geometric transformation on each subsequent frame, ensuring that the same physical location of the absorbent pad in all frames is mapped to the same pixel coordinates, thus achieving mutual alignment of the absorbent pad's spatial position in each frame. The temperature values ​​of all pixels in the area where the absorbent pad is located are then extracted from each aligned frame. A binary mask is pre-created based on the physical shape of the absorbent pad. The number of rows and columns of the binary mask matches the number of rows and columns of the infrared thermal image. Pixels belonging to the absorbent pad area in the binary mask are set to 1, and pixels belonging to the background area are set to 0. For each aligned frame of the infrared thermal image, all pixels with a value of 1 in the binary mask are traversed. The digital signal value of the corresponding pixel is read from the infrared thermal image, and the temperature conversion function provided in the infrared thermal imager's factory calibration file is called to convert the digital signal value into a Celsius temperature value, thus obtaining the temperature value of each pixel in the absorbent pad area of ​​that frame of the infrared thermal image. The temperature values ​​of the same pixel in each frame of the infrared thermal image, which are arranged sequentially in time, are combined in chronological order to form a time series of temperature values ​​for each pixel position. This time series of temperature values ​​for each pixel position is used as the dynamic thermal response curve for each pixel position. Let the total number of frames in the infrared thermal image sequence be N frames, the frame acquisition time interval be Δt, and the absorbent pad area within the binary mask contain M pixels. For the m-th pixel within the binary mask, the spatial coordinates of this pixel in the infrared thermal image are represented by row index i and column index j. The temperature values ​​at coordinate (i,j) in the infrared thermal images of the 1st frame, 2nd frame, up to the Nth frame are arranged in ascending order of frame number to obtain a temperature value time series of length N, denoted as D(i,j)[k], where k is the frame number, k=1,2,…,N; D(i,j)[k] represents the temperature value at coordinate (i,j) in the k-th frame of the infrared thermal image, in degrees Celsius; the value range of row index i is from 1 to the total number of rows in the infrared thermal image, and the value range of column index j is from 1 to the total number of columns in the infrared thermal image. The temperature value time series D(i,j)[k] completely records the dynamic response of the pixel at coordinate (i,j) to the temperature change over time during the entire liquid absorption process. This temperature value time series is the dynamic thermal response curve of the pixel at coordinate (i,j).The above temperature value time series construction operation was performed on all M pixels within the binary mask, resulting in a total of M dynamic thermal response curves.

[0018] Example 2: See Figure 3 The process of constructing the endothermic response gradient field based on the time-series variation characteristics of the dynamic thermal response curve is as follows: In practical implementation, starting from the acquired dynamic thermal response curves of each pixel location, which represent a discrete sequence of temperature values ​​changing with frame number, a functional relationship between temperature and time is first established to calculate the time derivative. For the p-th pixel location within the binary mask, the temperature value at that pixel location at the k-th frame is denoted as... The corresponding real time point is ,in , where is the frame acquisition time interval of the infrared thermal image sequence, in seconds. Calculate the first-order temporal derivative of the dynamic thermal response curve at each pixel location at each frame time. Optionally, the central difference method is used to calculate the first-order temporal derivative at the k-th frame time; for the 1st and Nth frames, forward or backward differencing is used. For the p-th pixel location at the k-th frame time, when When, the first time-domain derivative The calculation method is as follows: ;when hour, ;when hour, Calculated This represents the instantaneous heating rate at the p-th pixel location at frame k, expressed in degrees Celsius per second. The second-order time-domain derivative of the dynamic thermal response curve for each pixel location is calculated at each frame. Using a method similar to the first-order time-domain derivative, the heating rate sequence is differentiated again to obtain the heating acceleration. For the p-th pixel location at frame k, when... When, the second time-domain derivative The calculation method is as follows: ;when hour, ;when hour, Calculated This refers to the heating acceleration at the p-th pixel location at frame k, expressed in degrees Celsius per square second. When determining the endothermic response intensity of each pixel location at each frame time, preset heating rate weighting coefficients and preset heating acceleration weighting coefficients are introduced. The heating rate weighting coefficient is denoted as... The weighting coefficient for heating acceleration is denoted as , and All are positive real numbers, and satisfy the following conditions: In one way of taking a value, The value is 0.6. The value is set to 0.4. This set of values ​​is based on the following: During the liquid absorption process of the absorbent pad, the temperature change trend is mainly dominated by the heating rate. Simultaneously, the heating acceleration reflects the non-steady-state characteristics of heat transfer. Through multiple experiments comparing the stability of the leading edge contour extraction and the consistency of diffusion speed with manual observation under different weight combinations, it was determined that the allocation of 0.6 and 0.4 allows the heat absorption response intensity value to more accurately capture the temperature gradient abrupt change caused by liquid diffusion. For the p-th pixel position at frame k, the instantaneous heating rate is first... Weighting coefficient of heating rate Multiply to obtain the weighted heating rate. ; accelerate heating Weighting coefficient of heating acceleration Multiplying them together yields the weighted heating acceleration. Then, add the weighted heating rate and the weighted heating acceleration to obtain the initial endothermic response intensity value of the p-th pixel at the k-th frame. ,Right now: in, This represents the index number of the pixel position within the binary mask. , This represents the total number of pixels within the absorbent pad area. The frame number, , This represents the total number of frames in the infrared thermal image sequence. The instantaneous temperature rise rate of the p-th pixel at the k-th frame is expressed in degrees Celsius per second. Let be the heating acceleration of the p-th pixel at the k-th frame, expressed in degrees Celsius per second squared. This is the heating rate weighting coefficient, with a value range of 0 < <1; This is the weighting coefficient for heating acceleration, with a value range of 0 < <1, and ; Let be the initial endothermic response intensity value at the p-th pixel location at frame k. After obtaining the initial endothermic response intensity values ​​for all pixel locations at all frame times, normalize all initial endothermic response intensity values. Optionally, the normalization process uses the min-max normalization method, applying it to all pixels. The values ​​are linearly mapped to a preset numerical range [0,1]. Specifically, the operation involves iterating through all p and all k values ​​to find the minimum value among all initial endothermic response intensity values. and maximum value For each group (p,k), calculate the normalized endothermic response intensity value. .like and If they are equal, then all Set to 0. The value after normalization. This represents the heat absorption response intensity value of the p-th pixel at the k-th frame, with a dimension of 1 and a value range of [0,1]. The heat absorption response intensity values ​​of all pixels within the absorbent pad region at the same frame are arranged according to their spatial coordinates to form the heat absorption response gradient field for that frame. For the k-th frame, a two-dimensional matrix is ​​constructed, with the same number of rows and columns as the infrared thermal image. For the p-th pixel within the binary mask, whose spatial coordinates are row index r and column index c, the heat absorption response intensity value of this pixel at the k-th frame is... Fill the r-th row and c-th column of the matrix; for background pixel positions outside the binary mask, directly assign a value of 0. The resulting two-dimensional matrix is ​​the endothermic response gradient field at the k-th frame, and a total of N endothermic response gradient fields are obtained.

[0019] Example 3: See Figure 4 The process of extracting the liquid diffusion front contour line based on the spatial distribution of the endothermic response gradient field is as follows: In practice, spatial gradient magnitude calculation is performed on the endothermic response gradient field at each frame time to obtain the gradient magnitude field at each frame time. For the endothermic response gradient field at the k-th frame time, this endothermic response gradient field is a two-dimensional matrix, where the element in the u-th row and v-th column is the endothermic response intensity value at that coordinate position, denoted as . Where u is the row index, ranging from 1 to the total number of rows H of the infrared thermal image, v is the column index, ranging from 1 to the total number of columns W of the infrared thermal image, and k is the frame number. The spatial partial derivatives of each pixel in the endothermic response gradient field are calculated in the horizontal and vertical directions. The horizontal spatial partial derivative is obtained by convolving the endothermic response gradient field with a central difference kernel [-1, 0, 1] in the horizontal direction. For the pixel position in the u-th row and v-th column, the horizontal spatial partial derivative is... The calculation method is as follows: when hour, When v=1, When v=W, The spatial partial derivatives in the vertical direction are obtained by convolving the transpose of the central difference kernel [-1,0,1] with the endothermic response gradient field in the vertical direction. For the pixel position in the u-th row and v-th column, the spatial partial derivatives in the vertical direction are... The calculation method is as follows: when hour, When u=1, When u=H, The gradient magnitude of each pixel is obtained by adding the squares of the horizontal and vertical spatial partial derivatives and then taking the square root. The gradient magnitude of the pixel at row u, column v at time k is denoted as... The calculation formula is: in, Let be the horizontal spatial partial derivative of the pixel at the u-th row and v-th column in the endothermic response gradient field at time k. Let be the spatial partial derivative in the vertical direction of the pixel at the u-th row and v-th column in the endothermic response gradient field at time k. Let be the gradient magnitude of the pixel at row u and column v in the endothermic response gradient field at time k. Arrange the gradient magnitudes of all pixel positions within the same frame according to their spatial coordinates, i.e., all... The values ​​are arranged into a two-dimensional matrix of H rows and W columns according to the row index u and column index v. This two-dimensional matrix is ​​the gradient magnitude field at the k-th frame. The above spatial gradient magnitude calculation is performed on the endothermic response gradient fields at all N frames, resulting in a total of N gradient magnitude fields.

[0020] After obtaining the gradient magnitude field for each frame, the pixel positions in the gradient magnitude field of each frame whose gradient magnitude is greater than a preset magnitude threshold are marked as candidate leading edge pixels. The preset magnitude threshold is the sum of the mean and standard deviation of the gradient magnitudes of all pixels in the gradient magnitude field of each frame. The preset magnitude threshold for different frame times is determined separately based on the gradient magnitude field of each frame time. For the gradient magnitude field of the k-th frame, this gradient magnitude field contains a total of Calculate the gradient magnitude. The arithmetic mean of the gradient magnitudes is denoted as . Calculate this The standard deviation of each gradient magnitude is denoted as . Then the preset amplitude threshold at time k is... for: Iterate through all pixel positions in the gradient magnitude field at time k. For the pixel position in row u and column v, if the gradient magnitude of that pixel position is... Greater than Then mark the position of that pixel as a candidate leading edge pixel; if the gradient magnitude Less than or equal to If the pixel position is not specified, it will not be considered a candidate leading edge pixel. The above marking operation is performed on the gradient magnitude field for all N frame times to obtain the set of candidate leading edge pixels for each frame time.

[0021] Connectivity analysis is performed on all candidate leading edge pixels at each frame time, extracting the outer boundary line of the connected component with the largest area, and using this outer boundary line as the liquid diffusion front contour line for that frame time. For the k-th frame time, the candidate leading edge pixels at that frame time constitute a binary image. In the binary image, the position marked as a candidate leading edge pixel has a value of 1, and the other positions have a value of 0. Connectivity analysis adopts the 8-connectivity definition, that is, if two pixels with a value of 1 are adjacent in the horizontal, vertical, or diagonal directions, they belong to the same connected component. A two-pass scanning method is used to mark the connected components of the binary image at the k-th frame time: the first scan traverses the binary image row by row and column by column. When a pixel with a value of 1 is encountered, the marking status of the left and upper neighbors of the pixel is checked, and a new temporary label is assigned to the pixel or the smallest label already existing in the neighborhood is inherited; when both the left and upper neighbors have labels and the labels are different, the equivalence relationship between the two labels is recorded; the second scan unifies the labels with the equivalence relationship into the smallest label. After two scans, the label number of each connected component and the number of pixels contained in each component are obtained. The connected component with the most pixels is selected as the largest connected component. The outer boundary line of the largest connected component is extracted as follows: morphological boundary extraction is performed on all pixel positions within the largest connected component. A 3x3 structuring element is used to perform an erosion operation on the largest connected component, and then the erosion result is subtracted from the largest connected component to obtain the set of pixel positions on the boundary of the largest connected component. The set of pixel positions on the boundary is connected in a counterclockwise or clockwise order to form a closed outer boundary line. This closed outer boundary line is the contour line of the liquid diffusion front at the k-th frame.

[0022] Example 4: See Figure 5When calculating the geometric center coordinates of the liquid diffusion front contour at each frame, for a given frame, the spatial coordinate set of all pixels enclosed by the liquid diffusion front contour is obtained. In practice, a flood filling algorithm is used to fill the area inside the liquid diffusion front contour. Using any pixel on the contour as a seed point, a search is performed in the 8-neighborhood direction. If a neighboring pixel is located inside the contour and has not been visited, it is added to the filling area, and the search continues until all pixels within the closed area enclosed by the contour are visited. The row and column indices of all pixels within the filling area are recorded to form a spatial coordinate set. Each element in the spatial coordinate set is a tuple (r, c), where r represents the row index of the pixel in the infrared thermal image, and c represents the column index of the pixel in the infrared thermal image. Calculate the arithmetic mean of the horizontal coordinates of all pixels in the spatial coordinate set. Then, sum the values ​​of all column indices c in the spatial coordinate set and divide by the total number of pixels Q in the spatial coordinate set to obtain the horizontal coordinate of the geometric center point, denoted as . Where f is the frame number. Calculate the arithmetic mean of the vertical coordinates of all pixels in the spatial coordinate set. Then, sum the values ​​of all row indices r in the spatial coordinate set and divide by Q to obtain the vertical coordinate of the geometric center point, denoted as Q. . Horizontal coordinate values and vertical coordinate values The geometric center coordinates of the liquid diffusion front profile that makes up this frame. .

[0023] The Euclidean distance between the geometric center points of the liquid diffusion front contour at two adjacent time points is calculated sequentially, and this Euclidean distance is used as the front displacement between adjacent frames. For time points f and f+1, the front displacement is... The calculation formula is: in, The frame number, ; Let f be the horizontal coordinate of the geometric center point of the liquid diffusion front profile at time f. The vertical coordinates of the geometric center point of the liquid diffusion front profile at time f. Let be the horizontal coordinate of the geometric center point of the liquid diffusion front contour at time f+1. The vertical coordinates of the geometric center point of the liquid diffusion front profile at time f+1. It represents the leading-edge displacement between frame f and frame f+1, in pixels.

[0024] Dividing the leading-edge displacement between adjacent frames by the time interval between the corresponding adjacent frames yields the interval spread rate for each adjacent frame interval. The frame acquisition time interval for the infrared thermal image sequence is... The unit is seconds. The value is a known fixed value. For the interval between time f and time f+1, the interval diffusion rate is... , The unit is pixels per second.

[0025] Calculate the arithmetic mean of the interval diffusion rates across all adjacent frame intervals, and use this arithmetic mean as the absorption rate of the absorbent pad. The interval diffusion rate sequence across all adjacent frame intervals is as follows: ,Will to common The values ​​are summed to obtain the total diffusion rate of the interval. Then, the total diffusion rate of the interval is divided by 1 / 2. The arithmetic mean is obtained, which is the absorption rate of the absorbent pad.

[0026] Example 5: See Figure 6 Before calculating the arithmetic mean of the interval spread rates across all adjacent frame intervals, the total number of interval spread rates across all adjacent frame intervals is first obtained. The total number of frames in the infrared thermal image sequence is... The number of adjacent frame intervals is Therefore, the total number of interval spread rates for all adjacent frame intervals is The interval spread rate of all adjacent frame intervals is denoted as... .

[0027] Abnormal interval spread rates exceeding a preset upper threshold or falling below a preset lower threshold are removed from all adjacent frame intervals. The preset upper and lower thresholds are determined based on the statistical distribution characteristics of the interval spread rate sequence. In specific implementation, all interval spread rates are calculated... The median of the diffusion rates in each interval is denoted as . ; Calculate all The first quartile of the diffusion rate in each interval is denoted as The first quartile is the value at the 25th percentile after arranging the interval diffusion rate sequence in ascending order; calculate all... The third quartile of the diffusion rate in each interval is denoted as . The third and fourth quartiles are the values ​​at the 75th percentile after arranging the interval diffusion rate sequence in ascending order; the interquartile range is calculated. Preset upper limit threshold The calculation method is as follows: Preset lower threshold The calculation method is as follows: ,in The preset threshold adjustment factor, The range of values ​​for is real numbers greater than 0. In one possible value range, The value is 1.5. The rationale for setting it to 1.5 is as follows: In the box plot method for univariate outlier detection, using 1.5 times the interquartile range as the inner limit effectively marks observations that deviate significantly from the main distribution. This standard is widely used in statistical data processing and has a reasonable ability to identify abnormal diffusion rate values ​​caused by occasional shaking, partial obstruction, or transient noise from the thermal imager in the scenario of detecting the absorption rate of absorbent pads. (Tour all) The diffusion rate in the interval, for the ... Diffusion rate in each region ,like or Then The diffusion rate of the abnormal region is identified and eliminated; if Then keep After removing abnormal interval diffusion rates, the remaining interval diffusion rates constitute the set of effective interval diffusion rates. Let the set of effective interval diffusion rates contain a total of There are 10 values, among which .

[0028] The remaining interval diffusion rates after removing outliers are summed to obtain the total interval diffusion rate. The effective interval diffusion rate set is then... The values ​​are added one by one, and the sum is recorded as the total diffusion rate of the interval. Sum of the interval diffusion rates Divide by the number of remaining interval diffusion rates The arithmetic mean of the diffusion rates within the interval is obtained, and this arithmetic mean is taken as the absorption rate of the absorbent pad, denoted as . The calculation formula is: in, This represents the index number of the element in the set of effective interval diffusion rates. ; The total number of valid interval diffusion rates after removing abnormal interval diffusion rates; The first in the set of effective interval diffusion rates The effective diffusion rate is expressed in pixels per second. The absorbency rate of the absorbent pad is expressed in pixels per second.

[0029] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for detecting the absorption rate of an absorbent pad based on infrared imaging technology, characterized in that, The method includes the following steps: A sequence of multiple infrared thermal images of the absorbent pad during the liquid absorption process was acquired using an infrared thermal imager. Based on the temperature distribution in each frame of the infrared thermal image, the temperature value of each pixel on the surface of the absorbent pad is obtained, and the dynamic thermal response curve of the temperature value changing over time is tracked, including: Perform pixel-level spatial registration on each frame of the infrared thermal image sequence to align the spatial positions of the absorbent pads in each frame of the infrared thermal image. Extract the temperature values ​​of all pixels in the area where the absorbent pad is located from each frame of the aligned infrared thermal image. The temperature values ​​of the same pixel location in each frame of infrared thermal image arranged continuously in time sequence are combined in time order to form a time series of temperature values ​​for each pixel location, and the time series of temperature values ​​for each pixel location is used as the dynamic thermal response curve for each pixel location. Based on the temporal variation characteristics of the dynamic thermal response curve, the endothermic response gradient field of the absorbent pad at different times is constructed, including: Calculate the first-order time-domain derivative of the dynamic thermal response curve of each pixel location at each frame time, and use it as the instantaneous heating rate of each pixel location at each frame time. Calculate the second time-domain derivative of the dynamic thermal response curve of each pixel location at each frame time, and use it as the heating acceleration of each pixel location at each frame time. Based on the instantaneous heating rate and heating acceleration of each pixel position at each frame time, determine the endothermic response intensity value of each pixel position at each frame time. The heat absorption response intensity values ​​of all pixels in the area where the absorbent pad is located in the same frame are arranged according to the spatial coordinates of each pixel to form the heat absorption response gradient field of that frame. Based on the spatial distribution of the endothermic response gradient field, the liquid diffusion front contour of the absorbent pad is extracted, including: For the endothermic response gradient field at each frame time, the spatial gradient magnitude is calculated to obtain the gradient magnitude field at each frame time. The positions of pixels whose gradient magnitude is greater than a preset magnitude threshold in the gradient magnitude field at each frame time are marked as candidate leading edge pixels; Perform connected component analysis on all candidate leading edge pixels at each frame time, extract the outer boundary line of the connected component with the largest area, and use the outer boundary line as the liquid diffusion leading edge contour line at that frame time; The absorption rate of the absorbent pad is determined based on the displacement of the liquid diffusion front profile between adjacent frames, including: Calculate the coordinates of the geometric center point of the liquid diffusion front profile at each frame time. The Euclidean distance between the geometric center points of the liquid diffusion front profile at two adjacent time points is calculated sequentially, and the Euclidean distance is used as the front displacement between adjacent frames. Divide the leading edge displacement between each adjacent frame by the time interval between the corresponding adjacent frames to obtain the interval diffusion rate of each adjacent frame interval. Calculate the arithmetic mean of the interval diffusion rates of all adjacent frame intervals, and use the arithmetic mean as the absorption rate of the absorbent pad.

2. The method for detecting the absorption rate of an absorbent pad based on infrared imaging technology according to claim 1, characterized in that, The step of determining the heat absorption response intensity value of each pixel position at each frame time includes: The weighted heating rate is obtained by multiplying the instantaneous heating rate of each pixel position at each frame time with the preset heating rate weighting coefficient. The weighted heating acceleration is obtained by multiplying the heating acceleration of each pixel position at each frame time with a preset heating acceleration weighting coefficient. The weighted heating rate and the weighted heating acceleration are added together to obtain the initial endothermic response intensity value of each pixel position at each frame time. The initial heat absorption response intensity values ​​of all pixel locations at all frame times are normalized so that all initial heat absorption response intensity values ​​are mapped to a preset value range. The normalized values ​​are then used as the heat absorption response intensity values ​​of each pixel location at each frame time.

3. The method for detecting the absorption rate of an absorbent pad based on infrared imaging technology according to claim 2, characterized in that, The step of obtaining the gradient magnitude field at each frame time includes: For the endothermic response gradient field at each frame time, calculate the spatial partial derivative of each pixel location in the horizontal direction and the spatial partial derivative in the vertical direction. The square root of the sum of the squares of the horizontal and vertical spatial partial derivatives at each pixel location is used to obtain the gradient magnitude at each pixel location. The gradient magnitudes of all pixels at the same frame are arranged according to the spatial coordinates of each pixel to form the gradient magnitude field at that frame.

4. The method for detecting the absorption rate of an absorbent pad based on infrared imaging technology according to claim 3, characterized in that, The preset amplitude threshold is the sum of the mean and standard deviation of the gradient amplitude of all pixels in the gradient amplitude field at each frame time, and the preset amplitude threshold at different frame times is determined separately according to the gradient amplitude field at each frame time.

5. The method for detecting the absorption rate of an absorbent pad based on infrared imaging technology according to claim 4, characterized in that, The step of calculating the geometric center point coordinates of the liquid diffusion front contour line at each frame time includes: For the liquid diffusion front contour line at each frame, obtain the set of spatial coordinates of all pixels enclosed by the contour line. Calculate the arithmetic mean of the horizontal coordinates of all pixels in the spatial coordinate set, and use it as the horizontal coordinate of the geometric center point; Calculate the arithmetic mean of the vertical coordinates of all pixels in the spatial coordinate set, and use it as the vertical coordinate of the geometric center point. The horizontal and vertical coordinate values ​​are used to form the geometric center point coordinates of the liquid diffusion front contour line at that frame time.

6. The method for detecting the absorption rate of an absorbent pad based on infrared imaging technology according to claim 5, characterized in that, The step of calculating the arithmetic mean of the interval spread rate of all adjacent frame intervals includes: Obtain the total number of interval spread rates for all adjacent frame intervals; The interval spread rate of all adjacent frame intervals is summed to obtain the total interval spread rate; Divide the sum of the interval diffusion rates by the total number of intervals to obtain the arithmetic mean of the interval diffusion rates; Before calculating the arithmetic mean, abnormal interval spread rates that are greater than a preset upper threshold or less than a preset lower threshold are removed from the interval spread rates of all adjacent frame intervals.

Images