A kit for detecting abnormality of nitrofurazone metabolite in aquatic products

Abstract

This invention relates to the field of computer image processing technology, and more particularly to a method and system for identifying multi-well anomalies in a reagent kit for nitrofurantoin metabolites in aquatic products. The method includes the following steps: acquiring single-well image data; reconstructing the pore location using polar coordinates from the single-well image data to obtain pore location reconstruction data; constructing a radial optical density spectrum based on the pore location reconstruction data to obtain radial optical density spectrum data; extracting radial structural features from the radial optical density spectrum data to obtain radial structural feature data; verifying concentric topological features based on the radial structural feature data to obtain annular structure data; determining the true color development of the annular structure data to obtain true color development data; and determining multi-well location color development coordination based on the true color development data to obtain multi-well location color development coordination data. This invention, through radial structural modeling and concentric topological verification of pore location color development, effectively distinguishes between true color development and matrix interference structures, improving identification accuracy.

Description

Technical Field

[0001] This invention relates to the field of computer image processing technology, and in particular to a method for identifying multi-pore site anomalies in a reagent kit for nitrofurantoin metabolites in aquatic products. Background Technology

[0002] With the expansion of aquaculture, the overuse of antibiotics such as nifuroslidone has become increasingly prominent, and the residues of their metabolites in aquatic products pose a potential risk to food safety. To achieve rapid detection of these substances, reagent kits are widely used due to their ease of operation and rapid response. These kits perform qualitative or semi-quantitative analysis of target metabolites through colorimetric reactions, meaning the detection results are determined by observing color changes within the wells.

[0003] However, in actual testing, due to the complex composition of aquatic product samples, especially under conditions of high fat, high protein, or particulate impurities, the color development results are easily affected by factors such as matrix interference, well wall adhesion, uneven illumination, and imaging errors. This leads to non-ideal color development structures within the wells, such as rings, edge enhancement, and localized deposition. These structures resemble the true color development in the image, thus affecting the accuracy of the test results. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes a method for identifying multi-well site anomalies in a reagent kits for nitrofurantoin metabolites in aquatic products, thereby resolving at least one of the aforementioned technical problems.

[0005] This application provides a method for identifying multi-well site anomalies in a reagent kit for nitrofurantoin metabolites in aquatic products, including the following steps: Step S1: Acquire single-hole image data; reconstruct the hole position using polar coordinates from the single-hole image data to obtain the hole position reconstruction data; Step S2: Construct a radial optical density spectrum based on the pore reconstruction data to obtain radial optical density spectrum data; extract radial structural features from the radial optical density spectrum data to obtain radial structural feature data; Step S3: Verify the concentric topological features based on the radial structural feature data to obtain the annular structure data; determine the true color rendering of the annular structure data to obtain the true color rendering data. Step S4: Based on the actual color development data, determine the color development synergy of multiple pores to obtain the color development synergy data of multiple pores.

[0006] This invention reconstructs single-well image data using polar coordinates and constructs a radial optical density spectrum based on this, effectively mapping the color development information of well locations from two-dimensional space to a one-dimensional structured signal, thus enhancing the ability to characterize the color development distribution patterns. Through radial structural feature extraction and concentric topological feature verification, it can distinguish between annular color development and unstructured perturbations at the structural level, effectively suppressing pseudo-color development phenomena introduced by matrix interference, uneven illumination, or acquisition errors. Combined with a true color development determination mechanism, the results are screened from the perspectives of color development saliency and energy distribution consistency, improving the accuracy and stability of single-well determination. Through multi-well location color development collaborative determination, the single-well identification results are extended to the plate-level spatial distribution level, and the inter-well correlation is used to correct isolated color development or structural mismatch results, thereby achieving effective identification of abnormal well locations.

[0007] Preferably, step S1 specifically includes: Acquire single-hole image data; Gradient vector flow processing is applied to single-aperture image data to obtain center localization data; Radial sampling is defined on the center positioning data to obtain radial sampling data; Pixel mapping is performed based on radial sampling data to obtain aperture reconstruction data.

[0008] This invention achieves stable center localization of apertures even under conditions of blurred edges, uneven illumination, or background interference by processing single-aperture image data using gradient vector flow. Compared to traditional methods based on geometric contours or threshold segmentation, this significantly improves the accuracy and robustness of center extraction. By defining radial sampling, the aperture region is transformed from a two-dimensional irregular distribution into a radially structured representation based on the center, allowing analysis to revolve around the physical diffusion path of color development, enhancing data consistency and interpretability. Combining pixel mapping to construct aperture reconstruction data enables a structured rearrangement of color information in the original image, effectively eliminating the effects of aperture offset, tilt, and local distortion.

[0009] Preferably, the radial optical density spectrum construction in step S2 is specifically as follows: Angle domain integral dimensionality reduction is performed based on the hole position reconstruction data to obtain angle domain integral data; Optical density data is obtained by performing optical density conversion on the integral data in the angle domain; Radial segmentation feature data is obtained by performing radial segmentation feature mapping based on optical density data; Radial optical density spectrum data is obtained by convolution smoothing based on radial segmented feature data.

[0010] This invention employs angular domain integral dimensionality reduction processing on the reconstructed aperture data, effectively compressing the original two-dimensional color distribution in the angular dimension. This preserves radial variation characteristics while eliminating the influence of local directional perturbations, thereby improving the consistency and stability of color information representation. Density conversion maps pixel intensity to more physically meaningful optical response quantities, making subsequent analysis more closely aligned with the actual color development mechanism. Radial piecewise feature mapping is used to represent the color distribution within different radius intervals, facilitating a precise description of the color change trend from the center to the edge. Convolutional smoothing suppresses random noise and local fluctuations, resulting in a more continuous and stable radial density spectrum.

[0011] Preferably, the radial structural feature extraction in step S2 specifically involves: Radial gradient data is obtained by calculating the radial gradient from the radial optical density spectrum data. Bimodal structure identification is performed based on radial gradient data to obtain bimodal structure data. Based on the bimodal structure data, the bimodal spacing feature is constructed to obtain the bimodal spacing feature data; The centroid radius of optical density is calculated based on the radial optical density spectrum data to obtain the centroid radius data of optical density. The radial gradient data, bimodal spacing feature data, and optical density centroid radius data are vectorized to obtain radial structural feature data.

[0012] This invention utilizes gradient calculation of the radial optical density spectrum to represent the variation trend of color intensity from the center to the edge, effectively highlighting color boundaries and structural transition points. The system identifies bimodal structures, distinguishing between single color structures and multi-layered structures, avoiding misjudgments caused by relying solely on overall intensity. Constructing bimodal spacing features helps quantify the spatial separation between different peaks, representing potential multi-source color development or interference superposition relationships. Calculating the centroid radius of optical density allows for a global description of the distribution center of color energy, enhancing the perception of overall structural shifts. Vectorizing the radial gradient, bimodal spacing, and centroid radius enables unified encoding of multi-dimensional structural information, elevating the color structure from a single feature description to a composite structural representation.

[0013] Preferably, the bimodal structure identification specifically involves: Multi-scale gradient expansion is performed on the radial gradient data to obtain multi-scale gradient data; Cross-scale validation is performed on multi-scale gradient data to obtain cross-scale validation data; Topological adjacency analysis was performed on the cross-scale test data to obtain peak level data. Structural semantic labeling is performed on the peak-level data to obtain peak labeling data; The peak calibration data is subjected to bimodal coupling processing to obtain bimodal structure data.

[0014] This invention employs multi-scale unfolding of radial gradient data to simultaneously capture the detailed changes and overall trends of the colorimetric structure at different scales, effectively avoiding the problems of local noise amplification or structural information loss caused by single-scale analysis. A cross-scale verification mechanism filters peaks that are stable across multiple scales, improving the robustness of peak identification and reducing the impact of spurious peaks introduced by random perturbations or sampling errors. Topological adjacency analysis models the spatial hierarchy between peaks, ensuring that peaks no longer exist in isolation but form an organizational relationship with internal and external hierarchical structures. Combined with structural semantic labeling, peaks at different locations are assigned physical meanings such as center diffusion or boundary enhancement, realizing the transformation from purely numerical features to structural semantic features. Bimodal coupling processing provides unified constraints and filtering of the correlation between peaks, effectively distinguishing between true bimodal colorimetric structures and spurious bimodal phenomena caused by noise or local perturbations.

[0015] Preferably, the dual-peak coupling process specifically involves: Peak pairing data is obtained by matching inner and outer peaks based on peak calibration data; Perform outer peak boundary adjacency analysis on the peak pairing data to obtain boundary adjacency data; External peak lag features are extracted from boundary adjacency data to obtain lag feature data; Boundary dynamic viscosity analysis was performed based on hysteresis characteristic data to obtain morphological stability data; The validity of the bimodal structure was screened based on the morphological stability data to obtain the bimodal structure data.

[0016] This invention pairs inner and outer peaks in peak calibration data, transforming the originally discrete peak information into a combination with a clear structural relationship, effectively establishing a basic correlation framework for the colorimetric structure. Through outer peak boundary adjacency analysis, the spatial attachment relationship between the outer peak and the pore wall can be identified, thus distinguishing the structural differences between edge enhancement caused by interface effects and those formed by normal diffusion. By extracting the hysteresis features of the outer peaks, the time or response delay information in the color formation process is mapped into calculable structural features, making the judgment of peak origin no longer limited to static distribution. Combined with boundary dynamic viscosity analysis, the retention and attachment behavior of the outer peaks in the boundary region is modeled, reflecting the stability and persistence characteristics of the colorimetric substance near the pore wall. Through bimodal validity screening based on morphological stability, the true bilayer colorimetric structure and the pseudo-bimodal structure can be distinguished.

[0017] Preferably, the concentric topological feature verification in step S3 specifically includes: Obtain reagent kit parameter data; Based on the reagent kit parameter data, an ideal diffusion dynamics simulation was performed to obtain an ideal concentric ring model; Based on the radial structural feature data and the ideal concentric ring model, the structural feature space is mapped to obtain the feature mapping model; Rotation-invariant topological verification of the feature mapping model was performed to obtain the ring structure data.

[0018] This invention constructs an ideal diffusion dynamics model using reagent kit parameter data, establishing a correspondence between the color development behavior during actual detection and the physical diffusion mechanism. This eliminates the reliance on empirical thresholds for judgment, providing clear mechanistic support. An ideal concentric ring model is constructed as a reference structure, and the extracted radial structural features are mapped to a unified feature space. This aligns the actual color development distribution with the theoretical diffusion morphology, effectively identifying abnormal structures that deviate from the ideal diffusion law. Rotation-invariant topological verification eliminates the influence of aperture orientation, imaging angle, and local inhomogeneities on the judgment results, ensuring the ring structure identification is stable against rotational perturbations.

[0019] Preferably, the determination of true color development in step S3 specifically involves: The annular structure data is transformed into a reference color space to obtain reference color space data; Colorimetric significance data are obtained by comparing the colorimetric significance based on the reference color space data. The energy distribution isotropically verified on the colorimetric significance data to obtain the true colorimetric data.

[0020] This invention decouples the brightness and color information in the original image by performing a reference color space transformation on the annular structure data. This allows color differences to be stably expressed under a unified colorimetric reference system, effectively reducing the impact of illumination variations and imaging condition differences on the judgment results. Through color saliency comparison, the color response differences between the annular region and the background region are quantitatively analyzed, highlighting the response characteristics of the true color-developed region and suppressing weak response fluctuations caused by noise or local interference. By verifying the isotropic nature of energy distribution, the color-developed structure is constrained and judged from the perspective of overall annular consistency, effectively distinguishing between true color development formed by uniform diffusion and non-uniform structures caused by local deposition or boundary adhesion.

[0021] Preferably, step S4 specifically includes: Based on the actual color development data, the topological mapping of the color development pore group is performed to obtain the color development pore group data; A pore adjacency map was constructed from the colorimetric pore site group data to obtain pore adjacency map data; Local support evaluation was performed on the well site adjacency map data to obtain local support data; Colorimetric co-identification is performed based on local support data to obtain multi-pore colorimetric co-identification data.

[0022] This invention utilizes topological mapping of real color development data to transform the originally scattered single-well determination results into a set of wells with spatial structural relationships, effectively establishing a holistic representation of the color development distribution at the board level. By constructing a well adjacency map, the row and column relationships and neighborhood associations between wells are explicitly modeled, enabling the organization and analysis of color development information at the structural level. Through local support evaluation, the degree of color development support obtained by each well within its neighborhood is quantified, thereby identifying color development regions with spatial consistency and suppressing misjudgments caused by isolated color development or local interference. Through collaborative color development recognition, single-well results are elevated to multi-well collaborative determination results, ensuring that anomaly identification does not rely solely on single-point features.

[0023] The beneficial effects of this invention are as follows: In step S1, the two-dimensional distribution in the original image is uniformly mapped to a radial structure expression based on the aperture center by reconstructing the aperture position polar coordinates, eliminating the influence of aperture position offset, tilt, and imaging distortion on subsequent analysis; In step S2, the radial optical density spectrum is constructed and radial structural features are extracted, transforming the color information from spatial distribution into a structured signal, enabling key features such as bimodal structure and centroid offset to be accurately represented, thereby providing a highly discriminative description of complex color patterns; The overall process embodies a solution that moves from dispersion to order, revealing the truth through form. The analysis process involves: in step S3, verifying concentric topological features using an ideal diffusion model, and combining chromatic saliency and isotropic constraints to determine true color development. This elevates the identification process from simple feature judgment to joint verification of physical mechanism and spatial structure, effectively suppressing misjudgments caused by matrix interference, boundary attachment, and imaging artifacts. In step S4, through pore topological mapping and adjacency map construction, the single-pore judgment result is extended to the multi-pore collaborative level. Local support relationships are used to correct isolated color development, achieving panel-level anomaly identification, prioritizing shape and determining truth based on structure. Attached Figure Description

[0024] Other features, objects, and advantages of this application will become more apparent from the following detailed description of the non-limiting embodiments, taken with reference to the accompanying drawings: Figure 1 A flowchart illustrating the steps of a multi-well site anomaly identification method for a nitrofurantoin metabolite kit in aquatic products according to an embodiment is shown. Figure 2 A flowchart illustrating the steps of a method for reconstructing the polar coordinates of a hole position according to an embodiment is shown. Figure 3 A flowchart illustrating the steps of a radial optical density spectrum construction method according to an embodiment is shown; Figure 4A flowchart illustrating the steps of a concentric topology feature verification method according to an embodiment is shown. Figure 5 A flowchart illustrating the steps of a multi-pore site colorimetric synergistic determination method according to an embodiment is shown. Detailed Implementation

[0025] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0026] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. Functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0027] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0028] Please see Figures 1 to 5 This application provides a method for identifying multi-well site anomalies in a reagent kit for nitrofurantoin metabolites in aquatic products, comprising the following steps: Step S1: Acquire single-hole image data; reconstruct the hole position using polar coordinates from the single-hole image data to obtain the hole position reconstruction data; In one embodiment, the system extracts the target single-well region from the overall image of the reagent kit plate. For example, the overall image of the reagent kit plate is preprocessed, including grayscale conversion and adaptive threshold segmentation, to obtain the initial candidate region of the well position. A circular detection method (such as based on Hough circle transform or edge contour fitting) is used to identify the center position and radius parameters of all well positions. For each detected well position, a bounding box is constructed by expanding a certain boundary (such as expanding the radius by 1.1-1.2 times) according to its center coordinates and radius, thereby extracting the corresponding single-well image region. The region is then grayscale processed and edge enhancement is performed. Gradient vector flow is used to perform convergence calculations on the aperture edges, forming an edge gradient distribution field. For example, the system calculates the initial gradient field of a single-aperture grayscale image, obtaining the gradient direction and magnitude of each pixel, constructing a gradient vector flow field. Iterative optimization is performed using smoothing and gradient constraints to diffuse gradient information towards the edges in flat regions while maintaining directional stability at the edges. Starting from each pixel, path tracking is performed along the opposite gradient direction (e.g., iterating 5-10 steps or until convergence), gradually converging the paths to the edge region. The distribution of termination positions for all paths is statistically analyzed, forming an edge convergence density map, which is the edge gradient distribution field. The center coordinates of the aperture are then determined based on the gradient convergence positions. , The x-coordinate of the center point of the hole (image column coordinates). Here is the ordinate (image row coordinate) of the center point of the aperture. Using this center as the origin, a polar coordinate sampling grid is constructed, consisting of angular and radial directions, with the radius covering the effective area of ​​the aperture. The system maps each pixel in the original image to its corresponding polar coordinate position in a radial pattern from the center outwards. , The x-coordinate (or column coordinate) of the target pixel. This represents the x-coordinate of the center point of the hole (or the lateral position of the origin of the polar coordinate system). This is the radial distance (or the radius value relative to the center point). For angle variables (or polar coordinate angle positions). The vertical coordinate (or image row coordinate) of the target pixel. Using the ordinate of the center point of the borehole (or the longitudinal position of the origin of the polar coordinate system), an unfolded image is generated with radius as the horizontal dimension and angle as the vertical dimension, serving as the borehole reconstruction data. This process enables a uniform radial representation of the color distribution within the borehole.

[0029] Step S2: Construct a radial optical density spectrum based on the pore reconstruction data to obtain radial optical density spectrum data; extract radial structural features from the radial optical density spectrum data to obtain radial structural feature data; In one embodiment, the system performs statistical processing along the angular direction based on aperture reconstruction data, summarizing the grayscale values ​​of all angular pixels at each radius position and calculating the average value, thereby obtaining a grayscale distribution sequence that varies with the radius. The grayscale values ​​are then logarithmically transformed using a reference brightness to obtain the corresponding radial optical density distribution, such as... , This represents the optical density value (or radial optical density value) at the i-th radius position. This is a reference brightness value (or a reference incident light intensity). This represents the original grayscale value (or radial average brightness value) at the i-th radius position. This is the i-th radial sampling position (or discrete radius point). This is a numerically stable term (or a zero-bias parameter). The optical density sequence is segmented according to a preset radius step size, and adjacent segments are smoothed using a one-dimensional smoothing method to construct radial optical density spectrum data. Difference operations are performed on this radial optical density spectrum to obtain a radial gradient sequence representing the changing trend, and local extrema locations where the density changes from rising to falling are identified for extracting bimodal spacing features. Based on the weighted relationship between the optical density at each radius position and the corresponding radius, the centroid position spectrum of the optical density distribution is calculated, such as... , It is the centroid radius (or the centroid position) of the optical density distribution. This represents the i-th radial sampling position (or a discrete point on the radius). Let be the optical density value corresponding to the i-th radius position. The radial gradient features, bimodal spacing features, and centroid position features are combined to form radial structural feature data.

[0030] Step S3: Verify the concentric topological features based on the radial structural feature data to obtain the annular structure data; determine the true color rendering of the annular structure data to obtain the true color rendering data. In one embodiment, the system reads reagent kit parameter data, including pore size, reagent diffusion time, and standard reaction liquid volume, and constructs a corresponding ideal concentric ring model based on preset diffusion rules. For example, the preset diffusion rules are based on the radial diffusion of the reagent system within the pores, specifically obtained through experimental calibration or fitting with historical standard samples. Multiple sets of standard reaction data are selected, and under different reaction times (e.g., 5 min, 10 min, 15 min) and volume conditions, the distribution curves of color intensity changing with radius are statistically analyzed and normalized to extract the average diffusion trend. Based on this trend, a radial distribution template with a high center, decreasing outwards, and attenuating at the boundaries is established, and several equivalent rings (e.g., 3-6 layers) are divided according to intensity stratification. The typical radius range, width ratio, and energy distribution of each ring are used as ideal characteristic parameters to form an ideal concentric ring model. The radial structural characteristic data are mapped to the feature space of this ideal model, and the peak position, ring width, and centroid position in the actual structure are compared and analyzed with the ideal model to calculate the deviation. When the actual structure maintains closure of the ring zones and a stable radius distribution under rotational conditions (the system extracts radial response sections in multiple angular directions (e.g., sampling every 10°, for a total of 36 directions), and statistically analyzes the peak radius and number of ring zone layers in each direction; if the deviation of the main peak radius corresponding to each direction does not exceed 5% of the effective radius (e.g., |Δr|≤0.05R), and the number of ring zone layers is consistent (difference not exceeding 1 layer), and there are no breaks or gaps exceeding 60° continuously, then it is judged to have good closure. The above radius deviation is compared with the corresponding radius in the ideal model. If the overall average deviation is less than a preset threshold (e.g., 10%), it indicates that the actual structure is consistent with the ideal diffusion, thus satisfying the concentric topological characteristics. That is, rotational consistency is used to verify structural integrity, and deviation is used to verify structural rationality; both serve as the judgment criteria), it is determined to satisfy the concentric topological characteristics, and the ring zone structure data is obtained. The system performs a reference color space transformation on the annular structure region, extracts the color difference between the annular region and the background region, and analyzes the consistency of color distribution at various angles along the circumferential direction. For example, the system divides the annular region into multiple sectors (e.g., 12 or 24 equal parts) along the angular direction, calculates the average color difference value within each sector, calculates the mean and standard deviation of all sectors, and calculates the deviation of each sector from the overall mean. If more than 80% of the sector color differences fall within the mean ± 20%, and there are no consecutive low response or high anomaly regions exceeding 90°, the circumferential distribution is considered consistent. When the color difference reaches a preset significance level, and the circumferential distribution fluctuation (when the fluctuation value is lower than the preset tolerance (e.g., the standard deviation is less than 15% of the mean), it indicates good circumferential distribution consistency; otherwise, it indicates the presence of local anomalies) is within the allowable range, it is determined to be true color data; otherwise, it is marked as a pseudo-annular zone or interference structure.

[0031] Step S4: Based on the actual color development data, determine the color development synergy of multiple pores to obtain the color development synergy data of multiple pores.

[0032] In one embodiment, the system backfills the actual colorimetric results obtained from each well according to the original row and column positions of the reagent plate, forming a corresponding colorimetric well distribution matrix. Using each well as a node, a well adjacency map is constructed based on the vertical, horizontal, and vertical adjacency relationships on the plate surface, establishing connections between adjacent wells. For each well determined to have actual colorimetric results, the number of wells in its neighborhood that also show actual colorimetric results is counted, and combined with the total number of wells in that well's neighborhood, its neighborhood support level is calculated, such as... , Let be the neighborhood support (or local color support index) of the i-th pore. This represents the number of holes (or adjacent holes) in the neighborhood of the i-th hole that are determined to be truly colored. This represents the total number of holes (or the maximum number of adjacent holes) in the neighborhood of the i-th hole. When the neighborhood support of a target hole reaches a preset threshold, and it maintains consistency with its adjacent holes in terms of color intensity level or structural type, it is identified as a cooperatively color-developing hole. If it appears alone and its neighborhood support is insufficient, it is marked as an isolated color-developing hole. The system outputs a set of cooperatively color-developing holes, isolated hole markers, and multi-hole color-developing cooperation data for the corresponding board position. When a hole meets the color-developing conditions at the single-hole level, but its neighborhood support is lower than a preset threshold (e.g., support less than 0.25 or the number of adjacent color-developing holes less than 2), or its color intensity and structural type are significantly inconsistent with surrounding holes (e.g., intensity deviation exceeding 30% or structural type mismatch), it can be identified as an abnormal hole. Conversely, when multiple adjacent holes simultaneously meet the support threshold (e.g., support greater than 0.5) and have consistent structural characteristics, they can be identified as a true cooperatively color-developing region. Based on this determination result, it can be used for reagent reaction validity verification, batch detection consistency analysis, and automatic removal of abnormal well positions.

[0033] Preferably, step S1 specifically includes: Step S11: Acquire single-hole image data; In one embodiment, the system cropes a region from the entire reagent kit image based on a preset well template. According to the standard layout information of the reagent plate (including row and column coordinates or well spacing parameters), the initial center position of the target well is determined, and a sub-image of a fixed size is cropped from this position as the single-well image data. The system performs grayscale processing on the sub-image, uniformly converting color information into grayscale representation; median filtering is used to denoise the image; and histogram equalization is used to normalize the image brightness, ensuring that images of different well positions have a consistent dynamic range and contrast. This results in single-well image data with uniform size and stable brightness distribution.

[0034] Step S12: Perform gradient vector flow processing on the single-aperture image data to obtain center localization data; In one embodiment, the system calculates pixel gradient information for a single-hole grayscale image to obtain an initial vector field representing the direction and magnitude of brightness changes, such as... , It is the image gradient vector (or brightness gradient vector field). This represents the gradient component (or horizontal gradient component) of the image in the x-direction. This represents the gradient component (or vertical gradient component) of the image in the y-direction. Based on this initial vector field, a gradient vector flow field is constructed. Through iterative optimization by setting smoothness and gradient consistency constraints, the vector field maintains its original gradient direction in edge regions and diffuses towards the boundary in flat regions, thus forming a globally continuous guiding field, such as... , This is the gradient vector flow energy function (or optimization objective function). These are the weight parameters (or regularization coefficients) for smoothing constraints. It represents the spatial gradient of the vector field in the x-component direction (or the u-component gradient). This is the spatial gradient (or v-component gradient) of the vector field in the y-component direction. This refers to the optimized gradient vector flow field (or guiding vector field). The initial image gradient field serves as a data constraint. Based on this gradient vector flow field, the system traces the path of each pixel in the image along the reverse direction of the vector to obtain the corresponding streamline convergence positions and calculates the distribution density of each convergence position. The region with the highest convergence density is selected, and its geometric center is calculated as the hole center position, outputting the center positioning data. This ensures stable center positioning even with incomplete edges or low contrast.

[0035] Step S13: Define radial sampling for the center positioning data to obtain radial sampling data; In one embodiment, the system uses the center positioning result as the origin of polar coordinates and establishes a radial sampling frame around this center. The effective radius range of the aperture position is determined based on aperture parameters or edge detection results, such as... , It is a radial position variable (or a continuous radius variable). The effective radius of the aperture (which can be determined by aperture parameters or edge detection results) is defined, and within this range, a radial distance interval from the center outwards is divided; simultaneously, the angular direction is uniformly divided according to a preset step size, forming multiple discrete directions, such as... , This represents the k-th angle sample value (or discrete angle position). Pi is a constant. For angle sampling index (or angular discrete number). For the number of angle samples, The system generates sampling positions point by point along the radial direction from the center at fixed intervals in each angular direction, constructing a set of sampling points covering different radii and directions, such as... , , This is the j-th radial sampling position (or discrete radius point). For radial sampling index (or radial discrete number). This represents the radial sampling interval (or radius step). By forming a polar coordinate sampling grid as described above, radial sampling data is obtained, ensuring uniform distribution in all directions and covering the colored area from the center to the edge.

[0036] Step S14: Perform pixel mapping based on radial sampling data to obtain aperture reconstruction data.

[0037] In one embodiment, the system converts the polar coordinate positions corresponding to the radial sampling data back to the original image coordinate system. Based on the center position and the radial distance and angle information of each sampling point, it determines the corresponding pixel position in the original image and reads the corresponding grayscale value, such as... , The x-coordinate (or column coordinate) of the target pixel. The x-coordinate (or center column coordinate) of the hole center point. This represents the radial sampling distance (or radius sampling value) of the j-th sample. The vertical coordinate (or image row coordinate) of the target pixel. The vertical coordinate (or center row coordinate) of the hole center point. This represents the k-th angular sample value (or angular sampling angle). For sample points falling at non-integer coordinate positions, bilinear interpolation is used to calculate their pixel grayscale. The system arranges the radial sampling sequences in the same angular direction into a row, and stacks them row by row according to angular order to construct a two-dimensional matrix structure, obtaining the hole position reconstruction data. The horizontal direction represents the radius change from the center to the edge, and the vertical direction represents different angular directions. Through this reconstruction process, the original circular color distribution is transformed into a regular rectangular expression, allowing the radial variation features to be continuously unfolded in space, reducing the impact of rotation and positional offset.

[0038] Preferably, the radial optical density spectrum construction in step S2 is specifically as follows: Step S21: Perform dimensionality reduction by angular domain integration based on the hole position reconstruction data to obtain angular domain integral data; In one embodiment, the system takes aperture reconstruction data as input, which includes pixel distributions at different radius positions and corresponding angular directions, such as... Reconstruct the pixel intensity (or optical density) value at the radius r and angle θ position in the image using polar coordinates. This is a function for reconstructing images in polar coordinates (or a function for representing data related to aperture reconstruction). This represents the radial position (or radius coordinates). This refers to the angular position (or angular coordinates). The system statistically processes the pixel values ​​of each radius layer along the angular direction, calculating the average value of all angular pixels corresponding to that radius to obtain a representative intensity value for that radius position, such as... , This represents the angular domain integral intensity value (or radial representative intensity) at the i-th radius position. This represents the number of angle sampling points (or the total number of angular samples) corresponding to this radius layer, and this represents the angle sampling index (or the k-th angular sampling point). This is the i-th radial sampling position (or discrete radius point). Let k be the sampling position (or discrete angle point) of the k-th angle. Through the above process, the original two-dimensional reconstructed data is compressed into a one-dimensional sequence that varies only with the radius, resulting in angle-domain integral data.

[0039] Step S22: Perform optical density conversion on the angular domain integral data to obtain optical density data; In one embodiment, the system converts the angular domain integral data into an optical density sequence. For the integral intensity value at each radius position, the system performs a logarithmic transformation based on a preset reference brightness to convert it into a corresponding optical density value. The reference brightness can be selected as the average brightness of the blank aperture or the average value of the brightest region in the image, such as... , Let i be the optical density value at the i-th radius position. This is a base-2 logarithmic transformation function (logarithmic transformation operator). This is a reference brightness value (or a reference incident light intensity). This represents the original brightness value (or integrated intensity value) at the i-th radius position. This represents the i-th radial sampling position (or radius sampling point). This is a numerically stable term (or a zero-bias parameter). Through this transformation, areas with deeper color and lower brightness are mapped to higher optical density values, while areas with lighter color and higher brightness correspond to lower optical density values, thereby enhancing the expression of color intensity.

[0040] Step S23: Perform radial segmentation feature mapping based on optical density data to obtain radial segmentation feature data; In one embodiment, the system divides the effective radius range from the center to the edge into multiple continuous intervals based on optical density data along the radial direction. The interval division method can use a fixed interval or a preset number of segments. For each segment, the system statistically analyzes the optical density characteristics within that interval, including the average optical density, maximum value, minimum value, and the variation range within the interval. For example, the average optical density of the j-th segment can be expressed as: , It represents the average optical density characteristic value (or segmental characteristic value) of the j-th radial segment. This represents the number of sampling points (or the number of statistical samples) within the j-th segment. This represents the i-th radial sampling position (or radius sampling point). This represents the starting radius (or lower bound) of the j-th segment. This is the ending radius position (or upper bound of the interval) of the j-th segment. Let be the optical density value corresponding to the i-th radius position. The system calculates the degree of difference between adjacent segments. The statistical results of each segment are combined in radial order to form radial segment feature data.

[0041] Step S24: Perform convolution smoothing based on the radial segmented feature data to obtain radial optical density spectrum data.

[0042] In one embodiment, the system employs a one-dimensional convolution method to smooth the radial segmented feature data, addressing potential local fluctuations and random noise. A pre-defined smoothing convolution kernel is selected, and the weights within the kernel are normalized to a fixed sum. The system then uses this convolution kernel to perform sliding calculations on the radial segmented feature sequence, weighting and summing the feature values ​​at the current position and its neighborhood according to their respective weights to obtain the smoothed radial spectrum value, such as... , The j-th radial spectral value after smoothing. This is the kernel index (or weight number). This represents the kernel length (or kernel size). Let t be the convolution weight (or kernel weight coefficient). These are the neighborhood feature values ​​(or convolution input feature values). For radial position index (or spectral position number). This is the index of the center position of the convolution kernel. The convolution kernel can be symmetrically structured to maintain the morphological characteristics of the true peak values ​​while suppressing high-frequency noise. The kernel weights satisfy a symmetry constraint, meaning that for a convolution kernel of length m, there exists... ,in A Gaussian symmetric convolution kernel is selected, with weights distributed according to a squared decay law from the center. For example, when the kernel length is 5, the corresponding weights are set to [0.06, 0.24, 0.40, 0.24, 0.06], and normalization is performed to make the weight sum equal to 1. For sequence boundary positions, mirror extension or neighbor complementation is used. After the above processing, continuous and stable radial optical density spectrum data are obtained.

[0043] Preferably, the radial structural feature extraction in step S2 specifically involves: Radial gradient data is obtained by calculating the radial gradient from the radial optical density spectrum data. In one embodiment, the system takes radial optical density spectrum data as input, where each data point corresponds to a radial position sampled at fixed intervals. To characterize the color development trend along the radial direction, the system performs first-order difference processing on the optical density spectrum, calculates the change amplitude between adjacent positions, and thus obtains the rate of change at each radius, such as... , This represents the radial gradient value (or the rate of change of optical density). This is the optical density value (or backward optical density value) of the next sampling point. It is the optical density value of the previous sampling point (or the forward optical density value). The radial sampling interval is used; for the middle position, estimation is performed using the difference method between adjacent points, while for the positions at both ends of the sequence, a one-sided difference method is used respectively. The obtained gradient values ​​are used to reflect the direction and intensity of the change in optical density. A positive gradient indicates enhanced outward color development, while a negative gradient indicates weakened outward color development. The system sets a change threshold, and small changes are judged as stable regions. The system obtains radial gradient data.

[0044] Bimodal structure identification is performed based on radial gradient data to obtain bimodal structure data. In one embodiment, the system identifies local extrema in the optical density spectrum based on radial gradient data. The gradient sequence is traversed along the radial direction; when a given location has a positive gradient preceding it and the current gradient becomes zero or negative, that location is designated as a peak point. , This is the gradient value (or forward gradient value) of the previous sampling point. This is the gradient value of the current sampling point (or the gradient value at the current location). The radius of the previous sampling point. This refers to the radius (or peak position) of the current sampling point. The system determines whether the optical density at this position is simultaneously greater than the values ​​on both its adjacent sides, and requires that its peak amplitude exceeds a preset threshold, thereby filtering out valid peaks. , This represents the optical density value (or current color intensity) at the current sampling point. This is the optical density value of the previous sampling point. This is the optical density value at the next sampling point. This represents the radius (or peak position) of the current sampling point. The radius of the previous sampling point. This represents the radius of the next sampling point. All valid peaks are sorted according to their radius size, and the existence of a two-peak structure is analyzed. When a significant trough region is detected between two peaks, and the light density at the trough is lower than that at the peak, it is determined to be a bimodal structure, and the positions of the inner and outer peaks are determined. The bimodal structure data is output to characterize the radially layered distribution of color development.

[0045] Based on the bimodal structure data, the bimodal spacing feature is constructed to obtain the bimodal spacing feature data; In one embodiment, the system calculates the radial distance between the inner peak position and the outer peak position based on the identified bimodal structure data, and uses the radius difference between the two as the bimodal spacing feature, such as... , This refers to the bimodal spacing (or radial interpeak distance characteristic). This refers to the outer peak radius position (or outer peak radial position). This refers to the inner peak radius position (or inner peak radial position). The system constrains the peak positions, requiring the inner peak to be located near the center and the outer peak near the edge. The relevant range can be divided proportionally based on the aperture radius. When the bimodal positions satisfy the above spatial distribution relationship, their spacing is retained as a valid feature; otherwise, it is determined as an invalid bimodal structure and discarded. The system outputs the bimodal spacing feature data.

[0046] The centroid radius of optical density is calculated based on the radial optical density spectrum data to obtain the centroid radius data of optical density. In one embodiment, the system represents the overall distribution of color development along the radial direction based on radial optical density spectral data. The system uses the optical density values ​​at each sampling radius as weights to perform a weighted average calculation on the corresponding radius locations, thereby obtaining the centroid location of the optical density distribution, such as... , It is the centroid radius of optical density (or the centroid position of colorimetric distribution). This is the radial sampling index (or sampling point number). This represents the radius (or radial sampling position) of the i-th sampling point. radius The optical density value (or radial optical density intensity) at a given location. When the centroid is closer to the center, it indicates that the color is mainly concentrated in the inner region; when the centroid is closer to the outer region, it indicates that the color diffuses towards the edge. The system selects sampling points with optical density higher than a preset threshold for calculation. The system obtains the optical density centroid radius data.

[0047] The radial gradient data, bimodal spacing feature data, and optical density centroid radius data are vectorized to obtain radial structural feature data.

[0048] In one embodiment, the system performs unified encoding processing on radial gradient data, bimodal spacing feature data, and optical density centroid radius data. First, the radial gradient sequence is truncated or resampled according to a preset length to form a gradient feature vector of fixed dimensions. The bimodal spacing feature and optical density centroid radius are then concatenated with the gradient feature vector as additional features to construct a structural feature vector. The system normalizes each feature component; for example, features involving radii are proportionally converted to the effective radius of the aperture. The system obtains radial structural feature data, achieving a unified expression of local variation characteristics, structural hierarchy relationships, and overall distribution location.

[0049] Preferably, the bimodal structure identification specifically involves: Multi-scale gradient expansion is performed on the radial gradient data to obtain multi-scale gradient data; In one embodiment, the system takes radial gradient data as input and constructs multi-scale processing based on the changing characteristics of the colorimetric structure at different scales. Multiple sets of scale windows of different lengths are preset, and the original radial optical density spectrum is smoothed at each scale to obtain the optical density sequence at the corresponding scale. Gradient calculation is performed on the optical density sequence corresponding to each scale to obtain the gradient results at different scales, such as... , This represents the radial gradient value (or multi-scale gradient value) at the k-th scale. This represents the backward-smoothed optical density value (or the right-side neighborhood smoothing value) at the k-th scale. In scale The smoothed radial optical density spectrum The radial sampling interval (or radius from the walk distance) is the radial sampling interval. This consists of a set of local neighborhood windows of different lengths. Multi-scale gradient data is formed by combining and arranging gradient sequences at multiple scales.

[0050] Cross-scale validation is performed on multi-scale gradient data to obtain cross-scale validation data; In one embodiment, the system performs cross-scale consistency verification on candidate peak positions based on multi-scale gradient data. In the gradient results at each scale, peak positions that satisfy the condition of gradient changing from positive to negative are identified, and their corresponding radius positions are recorded. The system compares the positions of these peaks across different scales. When a peak appears near the same radius at multiple adjacent scales, and its positional deviation is within a preset tolerance range (…), the system considers the peak position to be consistent across different scales. , This refers to the peak radius position (or the peak radial center position). A radius tolerance threshold (or position deviation tolerance range) is used to classify peaks as cross-scale stable peaks. For peaks that meet this condition, the system records their corresponding scale range and position distribution information, forming cross-scale verification data. This process effectively eliminates noise peaks that appear only at a single scale.

[0051] Topological adjacency analysis was performed on the cross-scale test data to obtain peak level data. In one embodiment, the system uses stable peaks obtained from cross-scale verification as nodes in the radial structure and constructs topological adjacency relationships based on the size patterns of each peak in the radial direction. The system sorts the peaks according to their radial positions, establishing an adjacency connection between two adjacent nodes without any other stable peaks in between. Based on the distance of each peak from the center of the borehole location, a hierarchical identifier is assigned; peaks closer to the center are classified as lower-level nodes, and peaks farther from the center as higher-level nodes, forming an ordered hierarchical structure according to the radius size pattern. When multiple peaks exist in the same borehole location, the system organizes them into a continuous hierarchical chain structure. The peak hierarchy data output by the system includes at least the peak position, the adjacency relationships between nodes, and the corresponding hierarchical order information.

[0052] Structural semantic labeling is performed on the peak-level data to obtain peak labeling data; In one embodiment, the system performs structural semantic labeling on each peak based on peak hierarchy data. According to the radius of the peak's location, the aperture is divided into a central region, a transition region, and an edge region; this division can be proportionally set according to the effective radius of the aperture. The system assigns corresponding semantic labels to peaks based on their region and hierarchy: peaks located in the central region and with lower hierarchical levels are labeled as inner peaks, such as... , This refers to the peak radius position (or peak radial position). This is the threshold for the proportion of the central area (or the coefficient for dividing the inner area). The effective radius of the pore location (or the standard pore diameter radius) is used. Peaks located in the edge region and at higher levels are marked as outer peaks, such as... , This refers to the peak radius position (or peak radial position). This is the threshold for the proportion of the edge region (or the coefficient for dividing the outer region). The effective radius of the pore location (or the standard pore diameter radius) is used. Peak values ​​in the transition zone are marked as intermediate peaks or undetermined peaks, such as... , This is the threshold for the proportion of the central area (or the coefficient for dividing the inner area). The effective radius of the hole (or the standard hole diameter radius). This refers to the peak radius position (or peak radial position). This is the threshold for the edge region proportion (or the outer region division coefficient). The system records the peak intensity corresponding to each peak and its stability information in multi-scale analysis. The system obtains peak calibration data.

[0053] In one embodiment, the effective radius R of the aperture is set to a normalized scale of 1, and the central region proportion threshold is... A value of 0.35 can be used as the threshold for the proportion of the edge region. A radius of 0.70 can be chosen. Therefore, peaks with a radius less than 0.35R are labeled as inner peaks, peaks with a radius greater than or equal to 0.70R are labeled as outer peaks, and peaks with a radius between 0.35R and 0.70R are labeled as intermediate peaks or undetermined peaks. For example, a peak with a radius of 0.22R can be identified as an inner peak; a peak with a radius of 0.81R can be identified as an outer peak; and a peak with a radius of 0.54R can be identified as an intermediate peak. If one peak is located in the inner peak region and the other in the outer peak region, and the hierarchical order increases from the inside to the outside, then they can be considered as paired peaks.

[0054] The peak calibration data is subjected to bimodal coupling processing to obtain bimodal structure data.

[0055] In one embodiment, the system performs bimodal coupling processing on the inner and outer peaks based on peak calibration data. The corresponding inner and outer peak nodes are extracted from the calibration results and paired according to radius size rules. It is then determined whether the paired peaks satisfy the basic hierarchical relationship, i.e., the inner peak is located inside the outer peak. For combinations that meet the conditions, the radial interval between the two peaks is calculated, and it is determined whether this interval is within a preset effective range. The system detects whether there is a unique valley region between the two peaks, and the optical density of this region is significantly lower than that of the peaks on both sides, to confirm that it constitutes a peak-valley-peak structure. When the pairing result simultaneously satisfies the hierarchical relationship, interval constraint, and valley separation condition, it is determined to be a valid bimodal structure, and information such as the inner peak position, outer peak position, and corresponding spacing is output; otherwise, it is marked as an invalid or pseudo-bimodal structure.

[0056] Preferably, the dual-peak coupling process specifically involves: Peak pairing data is obtained by matching inner and outer peaks based on peak calibration data; In one embodiment, the system performs pairing processing of inner and outer peaks based on peak calibration data. The system extracts sets of peaks labeled as inner and outer peaks respectively. For each inner peak, peaks located outside it are selected from the outer peak set according to their radius size, prioritizing those without interference from other main peaks and with a continuous hierarchical order. When an inner peak corresponds to multiple outer peaks, the system filters based on peak intensity and their matching degree in the edge region, prioritizing the retention of outer peaks with higher intensity and positions more within a preset range. Combinations that do not meet the sequential relationship or span too many levels are discarded. Each valid inner-outer peak combination is recorded as the corresponding peak position and intensity information, forming peak pairing data.

[0057] Perform outer peak boundary adjacency analysis on the peak pairing data to obtain boundary adjacency data; In one embodiment, the system performs boundary adjacency analysis on the position of the outer peak in the peak pairing data. Based on the difference between the radius of the outer peak and the effective radius of the hole, the radial distance between the outer peak and the hole boundary is calculated. When the distance is less than a preset threshold, the outer peak is determined to be in the adjacent region close to the hole wall. The system retrospectively analyzes the angular distribution of the corresponding radius layer in the polar coordinate reconstruction diagram, and statistically analyzes the coverage of the outer peak in each angular direction. When the region is continuously distributed in the angular direction and the coverage ratio exceeds a preset range (the radius layer where the outer peak is located is divided into several equal sectors along the angular direction (e.g., 36 sectors, each corresponding to 10°), the number of sectors with colorimetric response intensity exceeding the threshold (considered as high-coverage sectors) is counted, and their proportion of all sectors is calculated. Continuous coverage is determined when the following two conditions are met: first, the high-coverage sector forms a continuous segment in the angular direction with a continuous span of not less than 180°; second, the proportion of high-coverage sectors is not less than 60%. When both the continuity and coverage ratio conditions are met, the outer peak is considered to have a significant continuous distribution characteristic in the angular direction), and the outer peak is determined to have boundary adhesion characteristics. That is, the outer peak shows a tendency to form stable adhesion or extension along the pore wall region in spatial distribution, indicating that the chromogenic substance is locally enriched at the boundary due to interfacial effects (such as wetting, surface tension, or evaporation differences). The system jointly records the location of the outer peak, its distance from the boundary, and the angular coverage to form boundary adjacency data.

[0058] External peak lag features are extracted from boundary adjacency data to obtain lag feature data; In one embodiment, the system extracts the hysteresis characteristics of outer peaks based on boundary adjacency data. For each pair of inner and outer peaks, the radial interval of the outer peak relative to the inner peak is calculated to characterize the degree of delay of the outer peak relative to the inner peak, such as... , It is the radial delay (or the hysteresis distance of the outer peak). The radius of the outer peak (or the radial position of the outer peak). The radius (or radial position) of the inner peak is used as the reference point. Simultaneously, considering the location of the valley between the two peaks, the system analyzes whether the optical density decreases after the inner peak and then rises again near the boundary to form the outer peak, thus determining whether hysteresis enhancement exists. The system identifies the starting position of the outer peak's rising segment, i.e., the radius where the decrease transitions to a significant increase, and calculates the interval of this position relative to the inner peak as the initial hysteresis characteristic. , This is the initial lag (or lag start interval). The starting radius (or the lagging starting position) of the outer peak rising segment. The radius of the inner peak is defined as the radius of the inner peak. When all the above-mentioned delay characteristics are within a preset range, and the outer peak is located in the boundary adjacent region, it can be determined that the outer peak has obvious hysteresis formation characteristics. The radial delay degree, initial hysteresis characteristics, and valley value separation are recorded to form hysteresis characteristic data.

[0059] Boundary dynamic viscosity analysis was performed based on hysteresis characteristic data to obtain morphological stability data; In one embodiment, the system performs boundary dynamic viscosity analysis on the outer peak based on hysteresis characteristic data and boundary adjacency relationships. Specifically, when the outer peak is located in the vicinity of the pore boundary and its formation is significantly delayed relative to the inner peak, the system further analyzes the radial distribution pattern of the outer peak. By determining the radius positions corresponding to the outer peak when it reaches half its peak height, its coverage area is calculated to characterize the broadening degree of the outer peak, such as... , The half-width at half-maximum (or the broadening parameter of the outer peak) is the outer peak. It is the radius of the right half-height position of the outer peak (or the radius of the right boundary). The radius of the left half-height position (or left boundary radius) of the outer peak is defined. Simultaneously, the rates of change of the rising and falling segments of the outer peak are compared. When the changes on both sides are relatively gradual, it indicates that the outer peak exhibits a diffusion-type distribution characteristic. The system combines the proximity of the outer peak to the boundary, the degree of hysteresis formation, and the characteristics of peak broadening and smoothing to assess its morphological stability. The closer an outer peak is to the boundary, the more pronounced the hysteresis, and the more gently distributed it is, the more likely it is to originate from the interface retention effect. For example, the radial distance between the outer peak and the pore boundary is less than 10% of the pore diameter as a criterion for proximity to the boundary; the radial interval between the inner peak and the outer peak is greater than 15% of the pore diameter as a criterion for pronounced hysteresis; and the ratio of the half-width at half-maximum of the outer peak to the pore diameter is greater than 20%, and the absolute values ​​of the gradients in the rising and falling segments are both less than a preset threshold (e.g., 30% of the overall maximum gradient) as a criterion for gentle distribution. When all three conditions are met, the outer peak is determined to originate from the interface retention effect. The relationship between the interface retention effect and the morphological stability data is that interface retention causes the diffusion of chromogenic substances to be restricted and to accumulate in the boundary region, thereby forming a broadened and gently changing stable peak shape. Therefore, its corresponding morphological stability index shows high stability but low structural authenticity. The system obtains morphological stability data, which characterizes the radial distribution and formation of the outer peak structure. This data includes at least parameters such as the distance between the outer peak and the boundary, inter-peak hysteresis, peak broadening, and gradient smoothness. The system can normalize these parameters and combine them to form a stability evaluation vector. Based on preset rules, it outputs stability labels, for example, labeling outer peaks that satisfy the interface retention characteristics as highly stable pseudo-structures, while labeling outer peaks that do not meet these conditions but have clear structures as true structures.

[0060] The validity of the bimodal structure was screened based on the morphological stability data to obtain the bimodal structure data.

[0061] In one embodiment, the system performs validity screening of bimodal structures based on morphological stability data. For each pair of inner and outer peaks, it determines whether the outer peak simultaneously possesses characteristics such as proximity to the boundary, significant lag in formation, and broadening and gradual change in peak shape. When all the above conditions are met, the outer peak is identified as a pseudo-outer peak formed by interface stagnation, and the corresponding peak pair is removed from the valid set. For outer peaks that do not exhibit obvious boundary stagnation characteristics, the system checks whether there is a clear valley separation and a relatively independent peak shape structure between them and the inner peak. If the conditions are met, the pair is retained as a valid bimodal structure. The positions of the screened inner peaks, outer peaks, inter-peak distances, and corresponding validity markers are output uniformly to form bimodal structure data. Through this processing, the true bilayer color development structure and the pseudo-bimodal structure caused by boundary effects can be effectively distinguished.

[0062] Preferably, the concentric topological feature verification in step S3 specifically includes: Step S31: Obtain reagent kit parameter data; In one embodiment, the system acquires parameter data corresponding to the target kit, representing the basic conditions of the pore structure and colorimetric reaction process. The parameter data includes at least pore geometry parameters, reaction liquid volume parameters, colorimetric reaction time parameters, and standard diffusion reference parameters. Specifically, the pore geometry parameters characterize the pore size, effective reaction area, and pore wall structure; the reaction liquid volume parameters define the initial spreading state of the liquid within the pores; the colorimetric reaction time parameters describe the duration of colorimetric diffusion; and the standard diffusion reference parameters represent the diffusion trend of the chromogenic substance from the center outwards under normal conditions. The system can acquire these parameters by reading the kit specification file, calling a pre-calibrated database, or importing experimental calibration results, and then uniformly organizes them to form the kit parameter data. , This refers to the kit parameter set / kit parameter data. The effective radius of the hole position. The volume of the reacting liquid. For the duration of color development, For diffusion scale parameters, This is the boundary attenuation parameter.

[0063] Step S32: Perform ideal diffusion dynamics simulation based on the reagent kit parameter data to obtain an ideal concentric ring model; In one embodiment, the system constructs an ideal diffusion kinetic model based on reagent kit parameter data, representing the theoretical distribution of the chromogenic substance diffusing outward from the center under normal reaction conditions. The well sites are considered as circular regions with a fixed radius originating from the center. It is assumed that the color intensity changes only with the radius, and a radial relationship is constructed based on diffusion scale parameters and boundary attenuation characteristics, causing the color intensity to gradually decrease from the center outward, with attenuation occurring near the well wall. , It is the radial color intensity distribution function (or radial function). The initial intensity at the center, It is an exponential decay function (or exponential transformation function). This refers to the radial distance (or radius position). For diffusion scale parameters, The boundary correction term characterizes the natural attenuation trend near the well wall. The initial central intensity represents the theoretical maximum response intensity of the chromogenic substance at the center of the well at the start of the reaction, and its value is derived from the standard reaction calibration information in the kit parameter data. The initial central intensity can be determined by the average colorimetric intensity of the blank control well or standard positive well in the initial stage, for example, taking the average pixel value of the central region radius less than 10% of the well diameter as the initial reference value. Under different reagents or different detection items, due to the different types of reactants and colorimetric mechanisms, the corresponding initial central intensities are allowed to differ, but they should remain consistent within the same batch of tests or fluctuate within a preset range (e.g., ±5%). The system calculates the intensity / value at different positions within the entire radius and divides it into multiple continuous concentric layers according to the intensity variation range, with each layer corresponding to an ideal annular region. The system obtains an ideal concentric ring model composed of multiple concentric rings. This model can serve as a reference template for the ideal colorimetric structure.

[0064] Step S33: Perform structural feature space mapping based on radial structural feature data and the ideal concentric ring model to obtain the feature mapping model; In one embodiment, the system maps the actually extracted radial structural feature data to an ideal concentric ring model into a unified structural feature space, establishing a correspondence between the two. The system normalizes actual features such as radial gradient, bimodal spacing, and optical density centroid radius; for example, it scales radius-related parameters according to the effective radius of the aperture, forming a dimensionless feature representation. It extracts corresponding theoretical features from the ideal concentric ring model, including ideal peak positions, concentric ring spacing, and energy center distribution, and constructs a theoretical feature representation, such as... , This is the actual radial structure eigenvector. This is the first radial gradient feature (or inner gradient strength feature). This is the second radial gradient feature (or outer gradient strength feature). This represents the nth-order radial gradient feature (or multi-scale gradient feature component). This is a feature of the bimodal spacing (or interpeak distance). This represents the centroid radius of optical density (or the location of the energy center). The actual and theoretical characteristics are projected onto the same feature space. , For the ideal structural feature vector, For the ideal first radial gradient feature, For the ideal second radial gradient feature, For the ideal nth order radial gradient feature, For the ideal bimodal spacing characteristics, Given the ideal optical density centroid radius, a feature mapping model is established by calculating the degree of difference or positional deviation between the two.

[0065] Step S34: Perform rotation-invariant topological verification on the feature mapping model to obtain the ring structure data.

[0066] In one embodiment, the system performs rotation-invariant topological verification of the colorimetric structure based on a feature mapping model. The system extracts multiple radial response sections from the polar coordinate reconstruction data at different angles and maps them to a unified feature space for comparative analysis. If the target structure is a real concentric ring, its characteristic performance at different angles should be consistent, i.e., the number of rings is the same, the corresponding peak positions are similar, and the overall structure remains closed and continuous. Based on this, the system judges the closure of the concentric rings, the radius distribution deviation, and the stability of the peak level in each angle section. For example, the system identifies the corresponding radial peak position in each angle section and judges whether the peak is continuously present in adjacent angle directions. When the peaks of the same level appear continuously in an angle range exceeding a preset proportion (e.g., 80%), the concentric ring is determined to be a closed structure. The radius distribution deviation is obtained by statistically analyzing the dispersion of the radius positions of the peaks of the same level in different angle directions. For example, it calculates the difference or standard deviation between the maximum and minimum radii. When the deviation is less than a preset threshold (e.g., 5% of the hole radius), the radius distribution is considered stable. The peak level stability is obtained by comparing the number of peaks and the consistency of their level order in each angle section. When the number of peaks is consistent in most angles (e.g., exceeding 85%) and the level order does not overlap or become missing, the level is determined to be stable. When the results at each angle meet the preset consistency requirements of the above indicators, and there is no obvious structural break or unilateral shift (when the range of continuously missing angles exceeds a preset threshold (e.g., 30°), it is considered that there is a structural break; the unilateral shift is obtained by comparing whether the color intensity distribution in different angular directions shows obvious asymmetry, for example, when the average color intensity in one semicircular region is significantly higher than that in another semicircular region (e.g., a difference of more than 20%), it is determined that there is a unilateral shift), the color structure is determined to satisfy the rotation-invariant topological condition. The system outputs information such as the number of ring layers, the radius of the main ring, and structural stability markers to form ring structure data, distinguishing between real rings and pseudo-structures generated by local disturbances.

[0067] Preferably, the determination of true color development in step S3 specifically involves: The annular structure data is transformed into a reference color space to obtain reference color space data; In one embodiment, the system takes the annular structure region obtained through concentric topology verification as input and extracts the corresponding pixel set from the original single-hole image. The system converts the pixels from the RGB color space to a reference color space, such as Lab or HSV, and preferentially uses the components representing color attributes for analysis, such as... , This refers to the luminance component (or lightness component). This is the first chromaticity component (or red-green channel component). This is the second chromaticity component (or the yellow-blue channel component). The system statistically analyzes the chromaticity components of all pixels within the annular region to obtain the average chromaticity characteristics of that region, such as... , This represents the average chromaticity feature (or chromaticity vector) of the annular region. The average value of the first chromaticity component in the annular region. This represents the average value of the second chromaticity component within the annular region; simultaneously, an adjacent region outside the annular region is selected as a background reference, and its average chromaticity characteristics are calculated, such as... , This represents the average chromaticity feature (or background chromaticity vector) of the background region. The average value of the first chromaticity component of the background area. This represents the average value of the second chromaticity component of the background region. By distinguishing and expressing the chromaticity information of the annular region and the background region, reference chromaticity space data is obtained.

[0068] Colorimetric significance data are obtained by comparing the colorimetric significance based on the reference color space data. In one embodiment, the system performs a significant comparison of the chromaticity differences between the annular region and the background region based on reference chromaticity space data. The degree of difference between the average chromaticity of the annular region and the average chromaticity of the background region is calculated to characterize the color rendering intensity of the annular region relative to the background. , The colorimetric saliency intensity of the annular region relative to the background region. It is the average value of channel a in the annular region (or the average chromaticity component a of the annular region). This represents the average value of channel a in the background region (or the average chromaticity component a of the background). This represents the average value of channel b in the annular region (or the average chromaticity component b of the annular region). The value is the average value of channel b in the background region (or the average chromaticity component b in the background). When using different chromaticity spaces, the corresponding chromaticity components can be selected for difference calculation. The system divides the ring region into multiple sub-segments along the angular direction, calculates the chromaticity difference of each segment, and counts the proportion of segments that reach the significance threshold. When the overall chromaticity difference exceeds the preset threshold and the proportion of segments meeting the conditions reaches the requirement, the ring is determined to have significant color rendering characteristics. The system outputs color rendering significance data, including the overall chromaticity difference level, the distribution of each segment, and the significance judgment mark, confirming that the ring is a real color rendering structure rather than a pseudo-structure caused by changes in brightness or texture.

[0069] The energy distribution isotropically verified on the colorimetric significance data to obtain the true colorimetric data.

[0070] In one embodiment, after confirming that the annular bands have significant color rendering differences, the system verifies the isotropy of their circumferential energy distribution. Using the radius of the annular band as a reference, the annular band region is divided into multiple sectors along the angular direction, and the cumulative color rendering intensity value within each sector is statistically analyzed to represent the color rendering energy level in that direction, such as... , The color rendering energy value within each sector For pixels (or pixel indices). Let k be the set of pixels in the k-th sector. This addresses pixel-level chromaticity differences. The system calculates the average energy of all sectors and compares the deviation of each sector from the average. When the energy deviation of most sectors is within a preset tolerance range, and there is no continuous large-scale energy loss or unilateral concentration, the ring is determined to have a uniform distribution characteristic in the ring direction. The system outputs true color rendering data, including the energy distribution of each sector, isotropic verification results, and overall color rendering judgment markers, ensuring that only structures with significant and uniform chromaticity are recognized as true color rendering.

[0071] Preferably, step S4 specifically includes: Step S41: Perform topological mapping of the color development aperture group based on the actual color development data to obtain the color development aperture group data; In one embodiment, the system backfills the single-well true color development determination results obtained in the aforementioned steps into the original well layout structure of the reagent kit. Based on the row and column structure of the reagent plate, each well is assigned a corresponding two-dimensional position identifier, and its true color development status is marked according to the determination result. Simultaneously, all wells determined to be true color development are extracted to form a set of color development wells, retaining additional information such as their corresponding color development intensity and structural type. The system organizes and maps this set according to the spatial relationship of the wells on the plate surface, forming color development well group data with topological distribution characteristics. The data includes at least the well position, color development state, and its distribution on the overall plate surface.

[0072] Step S42: Construct a well adjacency map from the colorimetric well site group data to obtain well site adjacency map data; In one embodiment, the system constructs a hole adjacency map using the color-developing hole location group data as a node set. All holes determined to be truly color-developing are treated as nodes in the map, and connections are established between nodes based on their spatial position on the board surface. When two holes are adjacent in row and column coordinates, such as vertically or horizontally adjacent, the system establishes a connection between them; when necessary, this can be extended to include diagonal adjacency relationships, such as... , The row coordinates (row index) of the target hole position. The row coordinates (adjacent row index) of adjacent holes. The column coordinates (column index) of the target hole position. This refers to the column coordinates (adjacent column indexes) of adjacent well sites. The system traverses all colored well sites, determines their adjacent well sites one by one, and records the set of adjacent nodes corresponding to each node, forming a well site adjacency graph structure. The well site adjacency graph data output by the system includes at least the node set, the connection relationship between nodes, and the adjacency information of each node, representing the spatial connection and distribution characteristics of the colored well sites.

[0073] Step S43: Perform local support evaluation on the well site adjacency map data to obtain local support data; In one embodiment, the system performs local support evaluation on each colorimetric well based on a well adjacency map. For each colorimetric well node, the number of wells within its adjacent range that are also determined to be truly colorimetric is counted. , This represents the number of neighboring colored nodes. Given the set of adjacent nodes (or the number of adjacent nodes), and combining this with the maximum number of adjacent nodes in the complete neighborhood, its local support is calculated to characterize the colorimetric aggregation of this pore location in the surrounding area, such as... , This represents local support (or color support coefficient). This represents the number of neighboring colored nodes. This represents the maximum neighborhood capacity (or the number of complete neighborhood nodes). The system determines the size of the connected region where a node is located by traversing the graph, i.e., counting the total number of colorimetric holes directly or indirectly connected to it. When a hole has high local support and its connected region is large, it can be determined that it is in a stable colorimetric cluster region; otherwise, it is an isolated colorimetric point. The system outputs local support data, including the support level of each node and the corresponding connected region size information.

[0074] Step S44: Perform colorimetric co-identification based on local support data to obtain multi-well colorimetric co-identification data.

[0075] In one embodiment, the system performs collaborative identification of colorimetric wells based on local support data. For each colorimetric well, the system judges its colorimetric support level and the size of its connected region. If the colorimetric support level in its neighborhood reaches a preset threshold and the size of its connected region is not less than the minimum size requirement, the well is identified as a collaborative colorimetric well; otherwise, it is marked as an isolated colorimetric well or an abnormal candidate well. For all wells that meet the conditions, the system identifies the entire connected sub-region to which it belongs as a collaborative colorimetric region. The system outputs multi-well colorimetric collaborative data, which includes at least the set of collaborative colorimetric wells, the spatial distribution range of each collaborative region, and isolated well marking information. Through this processing, the system extends from single-well judgment to plate-level collaborative analysis, enabling the identification of situations where local colorimetry is established but the overall distribution is abnormal. At the operational or sample level, uneven sample addition, non-centered droplet diffusion, or local contamination (such as dust, bubbles, etc.) may cause a structure resembling real colorimetry to form inside a well, but due to the lack of support from adjacent wells, it lacks spatial consistency. Regarding pore wall and boundary effects, wetting hysteresis or liquid climbing along the pore wall can easily form pseudo-ring structures at the edges, manifesting as colorimetric features within a single pore, but lacking overall continuity in distribution. Under imaging or environmental interference, such as local reflections, shadows, or sensor noise, randomly distributed false colorimetric points may also occur, their spatial locations being discrete rather than patchy. At the boundary between weak positives and false positives, true responses show a regional consistency trend, while false signals tend to appear in isolation. Through plate-level spatial collaborative constraints, colorimetric determination is expanded from single-point features to regional structural consistency determination, effectively suppressing isolated colorimetric responses caused by local anomalous signals, boundary retention effects, or imaging perturbations, thereby improving the system's ability to distinguish true colorimetric patterns.

[0076] Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended application documents rather than the foregoing description. Thus, it is intended that all variations falling within the meaning and scope of the equivalents of the application documents be incorporated into the invention.

[0077] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A method for identifying multi-well site anomalies in a reagent kit for nitrofurantoin metabolites in aquatic products, characterized in that, Includes the following steps: Step S1: Acquire single-hole image data; reconstruct the hole position using polar coordinates from the single-hole image data to obtain the hole position reconstruction data; Step S2: Construct a radial optical density spectrum based on the pore reconstruction data to obtain radial optical density spectrum data; extract radial structural features from the radial optical density spectrum data to obtain radial structural feature data; Step S3: Verify the concentric topological features based on the radial structural feature data to obtain the annular structure data; determine the true color rendering of the annular structure data to obtain the true color rendering data. Step S4: Based on the actual color development data, determine the color development synergy of multiple pores to obtain the color development synergy data of multiple pores.

2. The method according to claim 1, characterized in that, Step S1 is as follows: Acquire single-hole image data; Gradient vector flow processing is applied to single-aperture image data to obtain center localization data; Radial sampling is defined on the center positioning data to obtain radial sampling data; Pixel mapping is performed based on radial sampling data to obtain aperture reconstruction data.

3. The method according to claim 1, characterized in that, The construction of the radial optical density spectrum in step S2 is specifically as follows: Angle domain integral dimensionality reduction is performed based on the hole position reconstruction data to obtain angle domain integral data; Optical density data is obtained by performing optical density conversion on the integral data in the angle domain; Radial segmentation feature data is obtained by performing radial segmentation feature mapping based on optical density data; Radial optical density spectrum data is obtained by convolution smoothing based on radial segmented feature data.

4. The method according to claim 1, characterized in that, The radial structure feature extraction in step S2 is specifically as follows: Radial gradient data is obtained by calculating the radial gradient from the radial optical density spectrum data. Bimodal structure identification is performed based on radial gradient data to obtain bimodal structure data. Based on the bimodal structure data, the bimodal spacing feature is constructed to obtain the bimodal spacing feature data; The centroid radius of optical density is calculated based on the radial optical density spectrum data to obtain the centroid radius data of optical density. The radial gradient data, bimodal spacing feature data, and optical density centroid radius data are vectorized to obtain radial structural feature data.

5. The method according to claim 4, characterized in that, The specific steps for identifying bimodal structures are: Multi-scale gradient expansion is performed on the radial gradient data to obtain multi-scale gradient data; Cross-scale validation is performed on multi-scale gradient data to obtain cross-scale validation data; Topological adjacency analysis was performed on the cross-scale test data to obtain peak level data. Structural semantic labeling is performed on the peak-level data to obtain peak labeling data; The peak calibration data is subjected to bimodal coupling processing to obtain bimodal structure data.

6. The method according to claim 5, characterized in that, The specific process of dual-peak coupling is as follows: Peak pairing data is obtained by matching inner and outer peaks based on peak calibration data; Perform outer peak boundary adjacency analysis on the peak pairing data to obtain boundary adjacency data; External peak lag features are extracted from boundary adjacency data to obtain lag feature data; Boundary dynamic viscosity analysis was performed based on hysteresis characteristic data to obtain morphological stability data; The validity of the bimodal structure was screened based on the morphological stability data to obtain the bimodal structure data.

7. The method according to claim 1, characterized in that, The concentric topological feature verification in step S3 is as follows: Obtain reagent kit parameter data; Based on the reagent kit parameter data, an ideal diffusion dynamics simulation was performed to obtain an ideal concentric ring model; Based on the radial structural feature data and the ideal concentric ring model, the structural feature space is mapped to obtain the feature mapping model; Rotation-invariant topological verification of the feature mapping model was performed to obtain the ring structure data.

8. The method according to claim 1, characterized in that, The specific steps for determining the true color in step S3 are as follows: The annular structure data is transformed into a reference color space to obtain reference color space data; Colorimetric significance data are obtained by comparing the colorimetric significance based on the reference color space data. The energy distribution isotropically verified on the colorimetric significance data to obtain the true colorimetric data.

9. The method according to claim 1, characterized in that, Step S4 is as follows: Based on the actual color development data, the topological mapping of the color development pore group is performed to obtain the color development pore group data; A pore adjacency map was constructed from the colorimetric pore site group data to obtain pore adjacency map data; Local support evaluation was performed on the well site adjacency map data to obtain local support data; Colorimetric co-identification is performed based on local support data to obtain multi-pore colorimetric co-identification data.

Images