Method and system for analyzing root growth state image after seedling transplantation

Abstract

The invention discloses a method and a system for analyzing a root system growth state image after seedling transplantation, and particularly relates to the technical field of root system growth state image analysis. Based on an identifier between a single seedling and an image acquisition device, a root zone image time sequence acquisition and numbering system is constructed, and a shallow buried near-infrared imaging device is adopted to obtain a multi-frame image sequence; and through image rasterization, alignment resampling and structure enhancement processing, a root system atlas with space consistency and boundary separability is formed. Further extracting an edge migration vector field and a root tip active region, calculating a root tip active change rate and a root system increment density gradient, fusing to generate a recovery trend index, and combining a trend label and a structural feature vector to construct an activity grade judgment and response suggestion map, so as to realize intelligent analysis and management decision of a root system recovery state. The problems that root system state recognition lags behind, judgment subjectivity is high, and dynamic quantitative support is lacked are effectively solved.

Description

Technical Field

[0001] This invention relates to the field of root growth status image analysis technology, and more specifically, to a method and system for analyzing root growth status images after seedling transplantation. Background Technology

[0002] In seedling transplantation, the root system is the fundamental structure for maintaining the plant's water and nutrient absorption. The timeliness and stability of its growth recovery directly determine the seedling's adaptability and survival rate in the new environment. Especially in the early stages of transplantation, the root system is subjected to multiple influences, including environmental stress, soil disturbance, and physical damage, resulting in an extremely slow recovery process, and early changes in external morphology are difficult to detect. The above-ground appearance at this stage may not correspond to the actual state of the root system, making it impossible for traditional management methods relying on visual observation to identify potential root problems in a timely manner, thus affecting the scientific nature and timeliness of maintenance decisions.

[0003] Currently, the identification of root system status after seedling transplantation relies heavily on manual experience or endpoint survival statistics, lacking continuous, quantitative, and objective analytical methods. Especially in the early growth stages when roots are shallow, sparsely structured, and have unclear boundaries, existing image technologies struggle to extract effective features and accurately assess indicators such as root tip activity, capillary root density, and root spread trends, making it impossible to construct a dynamic evolution model of root growth trends. These problems limit the ability to detect and intervene in abnormal root conditions during seedling maintenance, hindering the improvement of overall transplanting effectiveness. Therefore, there is an urgent need to develop an intelligent identification method that can non-destructively perceive the early state of seedling roots through images and achieve qualitative and quantitative analysis. Summary of the Invention

[0004] To overcome the aforementioned deficiencies in the prior art, this invention provides a method and system for analyzing root growth status images after seedling transplantation. By constructing a multi-frame image time-series acquisition system based on shallow-buried near-infrared imaging, and combining the fusion analysis of root tip activity change rate and root incremental density gradient, a recovery trend index is generated and intelligent judgment of root zone activity level is achieved. This solves the problems of delayed root status recognition, strong subjectivity in judgment, and lack of dynamic quantitative support mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for analyzing root growth status images after seedling transplantation, comprising: Step S1: After the seedling transplantation is completed, a shallow-buried optical detection component is installed in the target root zone. Multiple frames of near-infrared images are collected in a continuous time period using a time-sharing control method, and the original image frames are established. Step S2: Perform structural equalization preprocessing on each frame of the original image frame, and construct a preprocessed image frame by smoothing the reflection distribution and softening the boundaries. Step S3: Using the preprocessed image frames as input, extract the migration vector field and regional increment features of the root region edge between consecutive image frames to form a root growth dynamic parameter sequence based on the time axis. The regional increment features include at least the root frame difference map, grayscale increment, density increment map and local density direction change rate map. Step S4: Calculate the root tip activity change rate (AAR) and root system incremental density gradient (IDG) respectively. The root tip activity change rate (AAR) is obtained based on the time derivative of the root tip normal vector. The root system incremental density gradient is obtained by extracting the density increase and gradient direction of the root system region in the partitioned grid. Input the root tip activity change rate and root system incremental density gradient into a preset frame-level mapping function, and output a comprehensive growth status index Γm that reflects the root growth activity of the corresponding image frame. Step S5: Construct a root zone activity level model based on the root zone growth trend status and the distribution map of comprehensive growth status indicators, and output a root zone structure map that includes growth status level, confidence level of confidence interval and response suggestions.

[0006] Preferably, when acquiring original image frames, a spatial index and acquisition time label are bound to each image frame based on the seedling identification tag to form a traceable dual-index image data structure, and the map is called based on the index in subsequent processing; frame-level resampling and registration operations are performed on the acquired original image frames, and an inter-frame aligned image sequence is constructed through nonlinear time interpolation and local elastic registration algorithms to improve the consistency and accuracy of subsequent dynamic parameter calculations.

[0007] Preferably, during the construction of the original image frames, the sampling method of the image frames is a differential resampling operation based on the intensity of pixel changes between frames. During the time period when the root region is growing rapidly or undergoing significant structural changes, the sampling density of the image frames is increased to enhance the temporal coverage of key dynamic changes. For static regions where the intensity of structural changes between consecutive frames is lower than a preset threshold, the sampling frequency is reduced to avoid redundant image acquisition. This ensures the complete expression of key growth information while improving the processing efficiency of the image sequence and the response focusing ability of the subsequent model.

[0008] Preferably, the preprocessing of the original image frame includes: The original image frame set is scheduled frame by frame and pixel normalized to construct a preprocessed image with a uniform brightness range; Perform reflectance feature enhancement and bidirectional moving average brightness smoothing to generate a reflectance feature enhanced image; The edge enhancement operator guided by the structural gradient direction extracts the edge features of the root region and generates the initial root region target contour map. Based on the spatial structure offset ΔRg between adjacent frames, the system dynamically identifies regions with unchanged continuous structures as false recognition signals, performs mask removal and cleaning, and outputs preprocessed image frames.

[0009] Preferably, the extraction of the root growth dynamic parameter sequence includes: Based on the structure enhancement map package, frame-level structure images and corresponding initial root region target contour maps RGm are extracted from the target frame set in a time series. Using any two adjacent frames in the inter-frame structure-aligned image sequence as processing units, for the corresponding initial root region target contour map, the edge point set is extracted, and the set of corresponding edge point pairs is generated by minimum distance addition matching. The vector migration value between all point pairs is calculated to construct the migration vector field Vm of the root region edge; the set of migration vector fields on the time axis is constructed for the entire sequence of frames. The frame difference map is formed by subtracting adjacent frames. The grayscale increment of the two frames within the initial root region target contour map RGm mask range is calculated. The original grid structure is partitioned, and the pixel increment density per unit area of ​​each sub-region is calculated to form a density increase map. At the same time, the gradient direction field is extracted in the density increase map to construct a local density direction change rate map and generate the root system incremental density gradient IDGm.

[0010] The preferred method for obtaining comprehensive growth status indicators is as follows: Based on the root tip activity change rate and root incremental density gradient output frame by frame from the target frame set, the root growth dynamic parameter sequence is reorganized on the time axis, and a frame-level mapping function Γm=f(AAR,IDGm) is constructed to generate a comprehensive growth status index Γm. The following combined expression is used in the calculation structure of the comprehensive growth state index Γm: ; Among them, the ln(1+α·AAR) term improves the sensitivity to weak changes through logarithmic transformation, while (1+β·IDG) retains the weighted influence of density growth on the overall structure formation; α is used to regulate the sensitivity to fluctuations in the root tip activity rate AAR, and β controls the threshold of the root incremental density gradient.

[0011] Preferably, the α and β coefficients are obtained in the following way: In the early stage of transplanting, root zone image sequences were selected from several representative samples, and the corresponding root tip activity change rate (AAR) and root incremental density gradient were extracted from each frame of the image. A gridded traversal was performed on the α and β values ​​under different combinations, the comprehensive growth state index Γm sequence corresponding to each combination was calculated, and the correlation between its change trend and subsequent real survival results was evaluated at the frame level. Based on survival rate data, paired labels are constructed, and the fluctuation characteristics of the comprehensive growth status index Γm sequence and growth outcome are mapped to the scoring model. The optimal parameter pair is determined by minimizing the prediction error. Using the optimal parameter pair as the initial setting, and combining the statistical characteristics of the sliding time window constructed from sampling data under different ecological conditions or soil disturbance, Bayesian optimization or genetic algorithm fine-tuning is performed to obtain a robust adaptive parameter adjustment mechanism. The final fitted parameter pairs are embedded into the mapping frame-level mapping function and deployed in an online image analysis system to generate a comprehensive growth status index.

[0012] Preferably, the current growth trend status of the root zone is categorized by label based on the comprehensive growth status index Γm, and the specific operation is as follows: After obtaining the comprehensive growth state index Γm, trend state identification processing is performed according to its distribution in the time series. This processing is based on the preset trend state judgment interval, dividing the numerical range of the comprehensive growth state index Γm into several level segments, with each level corresponding to a root zone growth trend state. During execution, the growth trend status of the root region is determined based on the slope of the change of the comprehensive growth status index Γm in consecutive frames and the interval landing point, and the root region growth trend status sequence is recorded in chronological order. The generated root region growth trend state sequence is subjected to temporal continuity analysis. If there are frequent switching of root region growth trend state or local fluctuations and instability, the sequence is corrected according to the comprehensive growth state index Γm value of adjacent frames and the boundary fluctuation situation, so that the output root region growth trend state has temporal consistency and structural rationality.

[0013] Preferably, based on the level boundary division, the fluctuation range and slope variance of the comprehensive growth status index Γm of each trend label within a specified time window are statistically analyzed to form a confidence index vector; each frame-level label is assigned a confidence level within a confidence interval according to the confidence index vector and marked in the label sequence as the stability basis for subsequent response suggestion judgment.

[0014] To achieve the objective of this invention, an image analysis system for the root growth status of seedlings after transplanting is provided, comprising: Image acquisition module: A shallow-buried optical detection component is installed in the target root zone after the seedling transplant is completed. Near-infrared images are continuously acquired through time-sharing control to generate a sequence of multiple original image frames. Image preprocessing module: Taking the original image frame sequence as input, it performs structural equalization processing on each frame image, including smoothing the reflection distribution and softening the root region edges, in order to improve the separation clarity between the root region and the background and construct a highly recognizable preprocessed image frame. Dynamic modeling module: Taking preprocessed image frames as input, it analyzes the migration vector field of the root region edge between adjacent frames and extracts the regional incremental features, which include root frame difference map, gray level increment, density increase map and local density direction change rate map, and constructs a dynamic parameter sequence for root growth. The growth state generation module takes the incremental features of the region as input, calculates the root tip activity change rate (AAR) and root system incremental density gradient (IDG), inputs the AAR and IDG to a preset frame-level mapping function, and outputs the comprehensive growth state index Γ corresponding to each frame image. m ; The grading analysis module constructs a root zone activity grading model based on the root zone growth trend status and the distribution map of comprehensive growth status indicators, and outputs a root zone structure map that includes growth status grading, confidence level of confidence intervals, and response suggestions.

[0015] The technical effects and advantages of this invention are as follows: (1) This invention achieves spatial consistency modeling and temporal comparability reconstruction of the dynamic behavior of root zone structure in the early stage of seedling transplantation by using a shallow-buried multi-frame near-infrared imaging device and an image rasterization resampling mechanism. This effectively solves the problems of lack of continuous observation of root recovery process, large spatial positioning error and inability to uniformly number and trace image processing results in traditional methods. At the same time, by using structure enhancement maps and root zone masks, the interference areas of soil and false edge signals are suppressed, making the subsequent dynamic migration analysis more robust and boundary fidelity. In the stage of extracting dynamic change parameters of root zone, the root tip activity change rate and root incremental density gradient are generated in conjunction with the explicit clustering of vector field and root tip active area, so that the dynamic change has structural explanatory power and morphological symmetry basis, and establishes a reliable data foundation for trend fusion analysis.

[0016] (2) This invention constructs a root region structure map containing growth state level, confidence level of confidence interval and response suggestions by adaptively fusing and normalizing the root tip activity change rate and root incremental density gradient, providing qualitative judgment basis and distributed response support for decision-making system, thereby improving the decision-making power and application of image analysis system. Attached Figure Description

[0017] Figure 1 This is a simplified flowchart of the image analysis method for root growth status of seedlings after transplanting, as described in this invention. Detailed Implementation

[0018] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0019] At the same time, it should be understood that, for ease of description, the dimensions of the various parts shown in the accompanying drawings are not drawn according to actual scale.

[0020] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the scope of this application and its application or use.

[0021] Techniques, methods, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and equipment should be considered part of the specification.

[0022] See Figure 1 A simplified flowchart of the image analysis method for root growth status of transplanted seedlings is provided in this invention. Figure 1 The image analysis method for the root growth status of seedlings after transplanting shown includes: Step S1: After the seedling transplantation is completed, a shallow-buried optical detection component is installed in the target root zone. Multiple frames of near-infrared images are collected in a continuous time period using a time-sharing control method, and original image frames are established to maintain the spatial distribution consistency and temporal comparability of the root zone images. Step S2: Perform structural equalization preprocessing on each frame of the original image frame, and enhance the separability of the root region by smoothing the reflection distribution and softening the boundaries to construct a highly recognizable preprocessed image frame. Step S3: Using the preprocessed image frames as input, extract the migration vector field and regional increment features of the root region edge between consecutive image frames to form a root growth dynamic parameter sequence based on the time axis. The regional increment features include at least the root frame difference map, grayscale increment, density increment map and local density direction change rate map. Step S4: Calculate the root tip activity change rate (AAR) and root system incremental density gradient (IDG) respectively. The root tip activity change rate (AAR) is obtained based on the time derivative of the root tip normal vector. The root system incremental density gradient is obtained by extracting the density increase and gradient direction of the root system region in the partitioned grid. Input the root tip activity change rate and root system incremental density gradient into a preset frame-level mapping function, and output a comprehensive growth status index Γm that reflects the root growth activity of the corresponding image frame. Step S5: Construct a root zone activity level model based on the root zone growth trend status and the distribution map of comprehensive growth status indicators, and output a root zone structure map that includes growth status level, confidence level of confidence interval and response suggestions, so as to provide qualitative judgment basis and distributed response support for the decision-making system.

[0023] Furthermore, step S1: the construction of the root region image temporal data acquisition mechanism includes: Sub-step 101: Deployment of detection components and location of data acquisition area Near-infrared imaging detection components are shallowly buried in the soil layer above the root zone after the seedlings are transplanted. The detection components include a multispectral imaging head, a time-division exposure control unit, and a time-series label encoding module. The boundary coordinates of the imaging area are set according to the physiological characteristics of the seedling root zone and the initial planting parameters of the plant, generating a corresponding root zone spatial identification label Map, and establishing a one-to-one mapping relationship with the current seedling number to ensure the consistency of the spatial distribution of the acquired images and the uniqueness of the identification. This identification label Map will be used for indexing in subsequent image matrix and index mapping. Sub-step 102: Multi-frame image acquisition and time-stamp encoding After generating the identification tag Map and completing the region boundary binding, the time-division exposure control mechanism of the detection component is started. Near-infrared image sequences containing continuous time sequence are acquired according to the preset imaging interval. Each frame of image is bound to the current acquisition timestamp Tk. At the same time, spatial classification is performed according to the identification tag Map to form an image data set Gseq={Gk|Map,Tk} with a dual index structure, where Gk represents the image frame of the root region at time k. Explanatory note: Image frame Gk enhances the reflection difference between root tissue and the surrounding soil background through the near-infrared band, which helps with root zone boundary enhancement and structure extraction in subsequent processing; this structure will become the basic data input for the entire image analysis process; Sub-step 103: Image rasterization and temporally unified resampling: To improve the spatial alignment accuracy of image analysis, image frame Gk is divided into blocks and rasterized according to the spatial grid to generate a normalized image matrix GGk={Gij^k}, where Gij^k is the pixel block of the (i,j)th grid unit in the image at time k. Then, the normalized image matrix GGk is uniformly resampled on the temporal axis. Using spline interpolation and inter-frame intensity matching strategies, non-uniformly sampled frames are uniformly mapped to the target frame set Tseq={T1, T2, ..., Tm} (representing the image data set corresponding to times 1, 2, ..., ... m), and an aligned image set GGalign={GGm} is generated. Explanatory note: This processing method is not a traditional uniform downsampling, but a differential resampling based on the rate of change of sampling intensity. This results in denser frame coverage during periods of rapid root growth or significant changes, thus preserving key dynamic change information. At the same time, it avoids redundant sampling of static, unchanging areas, improving data processing efficiency and model focus. Sub-step 104: Establishing a unified interface for image set import and atlas generation The aligned image set GGalign and its corresponding identifier Map, along with the target frame set Tseq, are used as a unified data packet input to the root region map cache system. A map call index table Index_Table={Map, Tseq, GGm} is constructed for subsequent steps in stage S2 to perform structural equalization preprocessing.

[0024] The overall technical synergy of step S1 is explained as follows: This step realizes the unified acquisition and marking archiving of seedling root zone images in the spatiotemporal dimension. Through the overall process of numbering and identification, time-series archiving, raster resampling, and function-based indexing, the image data structure is transformed into a standardized expression method that is "controllable, scalable, and reconfigurable". It provides a highly organized acquisition platform for the entire root system image analysis process.

[0025] Furthermore, step S2: image structure equalization preprocessing and root system target separability enhancement mechanism construction, includes: Sub-step 201: Image frame scheduling and structure initialization processing Using the image set GGalign completed in step S1 as input, individual seedlings are scheduled one by one according to the identification label Map, and the calling function is executed frame by frame according to the target frame set Tseq to extract the image frame GGm corresponding to each time Tm as the current processing target; at the same time, pixel value normalization and structural domain equalization are performed on the image frame GGm to construct a rasterized preprocessed image GGeqm under a uniform brightness range for subsequent reflection feature processing; Sub-step 202: Local reflection feature enhancement and regional brightness smoothing For the preprocessed image GGeqm, a region reflection fitting enhancement operation is performed. In each frame, the grid cell is used as the calculation window to adjust the overexposure or underexposure value of each cell edge. Brightness smoothing is performed based on a bidirectional neighborhood moving average filter to generate a reflection feature enhancement image, ensuring the stability of the root region texture features under background fluctuation conditions. Sub-step 203: Boundary-guided edge enhancement and coarse target extraction Based on the reflection feature-enhanced image, the edge enhancement operator guided by the structural gradient direction (DEGF) is used to extract the target edge features Em in the root region, and an initial target contour map RGm in the root region is generated based on the edge closure threshold to construct a frame-level coarse recognition root region mask; the initial target contour map in the root region is nested into the original image to generate a root region saliency map, realizing the integrated processing of structural enhancement and boundary imaging; The explanation is that this step achieves direction awareness and morphological integrity constraints for boundary extraction by guiding the gradient direction rather than dividing by the intensity threshold, effectively avoiding the breakage and misjudgment phenomena that occur in the root hair structure under the traditional Canny or Sobel operators; the initial root region target contour map RGm generated will be directly used as the mask basis for extracting dynamically changing parameters in step S3. Sub-step 204: Anti-spurious signal detection and significance consistency screening Structural consistency verification is performed on the root region saliency map. The spatial structural offset ΔRg of root region mask change in adjacent images is calculated using the statistical parameter of root region morphological difference between frames, ΔRg=||RGm−RG{m−1}||. If the spatial structure offset ΔRg is lower than the dynamic change threshold for multiple consecutive moments, the corresponding region is determined to be a false recognition signal, and mask removal is performed from the image to update it to a preprocessed image frame. Explanatory notes: The inter-frame structural variation threshold dynamic identification mechanism effectively avoids interference from false root signals caused by water droplets, reflections, and local shadows, improving the accuracy and continuity of subsequent dynamic parameter extraction; this ΔRg index will also be provided to the dynamic migration vector calculation in the S3 stage as one of the morphological drift reference parameters; Sub-step 205: Structural map storage and processing status registration All preprocessed image frames are packaged with their corresponding identifiers (Map), target frame set (Tseq), initial root region target contour map (RGm), root region target edge features (Em), and spatial structure offset (ΔRg) to generate a structure enhancement map package. This package is then uniformly registered in the structure map index, forming a preprocessed map scheduling interface. Subsequent steps can use this scheduling interface to extract the cleaned root region image and mask information in a time series manner, which can then be used as direct input for dynamic parameter sequence analysis.

[0026] This step demonstrates the synergistic effect of the techniques: Step S2 achieves a complete transition from spatial organization (S1 output) to enhanced image separability. On the one hand, it preserves the highly stable structure of the root region boundary details, and on the other hand, it removes dynamic interference background signals. The data ultimately provided to stage S3 has stronger morphological consistency and texture continuity. Especially in the multi-layered nested process of image brightness normalization + local reflectance rate adjustment + structure-guided edge extraction + cross-frame consistency verification, the technical features of each step have a high degree of coupling effect and are not independently superimposed.

[0027] Furthermore, step S3 includes the following: Sub-step 301: Inter-frame image deconstruction and root region structure synchronization alignment processing The structure enhancement map package corresponding to the target seedling is obtained by calling the interface. The frame-level structure image and the corresponding initial root region target contour map RGm are extracted from the target frame set according to the time sequence. In order to ensure the spatial consistency of each frame image in subsequent morphological comparison, pixel-level registration processing is performed. An image alignment algorithm based on a combination of local rigid constraints and global elastic fitting is adopted to generate an inter-frame structure alignment image sequence ={Gĝ_m}, ensuring that the root region structure maintains the consistency of position and scale on the time axis. Sub-step 302: Extraction of root region edge migration vector field and construction of temporal encoding Using any two adjacent frames (Gĝ_{m}, Gĝ_{m+1}) in the inter-frame structure-aligned image sequence as processing units, for the corresponding initial root region target contour maps RGm and RG{m+1}, edge point sets {P_i^m} and {P_j^{m+1}} are extracted, and the set of corresponding edge point pairs C_ij is generated by minimum distance addition matching; the vector migration value Vij=P_j^{m+1}−P_i^m between all point pairs is calculated to construct the migration vector field Vm of the root region edge; and the set of migration vector fields Vseq={Vm} on the time axis is constructed for the entire sequence of frames. Explanatory notes: The vector migration field not only includes positional differences, but also implies dynamic information such as root tip growth direction and edge expansion trend; the migration vector field set Vseq will serve as the core supporting parameter for calculating the root tip activity change rate AAR, and has high interpretability and physical mapping capability; Sub-step 303: Calculation of regional incremental features The regional incremental features include at least the root frame difference map, gray-level increment, density increase map, and local density direction change rate map. The frame difference map ΔGGm is formed by subtracting image frame Gĝ_{m+1} from image frame Gĝ_{m}. The gray-level increment ΔI_ij between the two within the initial root region target contour map RGm mask range is calculated. Following the original grid structure partitioning, the pixel increment density per unit area is calculated for each sub-region to form the density increase map ρ_m. Simultaneously, gradient direction field extraction is performed on the density increase map ρ_m to construct the local density direction change rate map φ_m. Combining ρ_m and φ_m generates the root system incremental density gradient IDGm, which is used for subsequent comprehensive trend determination. Explanatory note: The region incremental feature is a necessary input for step S4. Its calculation depends on the pixel-level changes and spatial distribution patterns between frames in the current step. Therefore, it needs to be executed under the combined effect of the grid structure and the edge mask to ensure the consistency of region division.

[0028] Furthermore, step S4 includes: Sub-step 401: Determination of active root tip region and generation of initial value of root tip activity change rate Based on the clustering characteristics of vector amplitude and direction in the migration vector field Vm, a spatial density weighting method is used to perform clustering analysis on edge point groups to identify regions in continuous frames where the vector exceeds the threshold and the direction is stable, and these regions are marked as root tip active candidate regions. For each root tip active candidate region, the mean vector amplitude and the rate of change of normal angle are calculated to form the root tip active change rate AAR, and its position index Tk in the time axis is recorded. Sub-step 402: Generation of comprehensive growth state indices and construction of structural interfaces Based on the root tip activity change rate AAR and root incremental density gradient IDGm output frame by frame by frame from the target frame set Tseq, they are reorganized into a root growth dynamic parameter sequence on the time axis, and a frame-level mapping function Γm=f(AAR, IDGm) is constructed to generate a comprehensive growth state index Γm. The structured root update index map U_map={Γ1, Γ2, ..., Γm} is obtained and incorporated into the map index for subsequent step S5. The frame-level mapping function refers to taking each frame as a unit, using the root tip activity change rate (AAR) and root incremental density gradient (IDGm) corresponding to that frame as inputs, and calculating the comprehensive growth state index Γm that reflects the root growth activity level of that frame through a preset function model f. This frame-level mapping function not only reflects the dynamic changes in root tip activity, but also comprehensively considers the spatiotemporal distribution characteristics of root structure density, thus enabling each frame to have a quantitatively analyzable growth state expression capability, which is convenient for subsequent trend calculations. Among them, the structured root system update index map refers to an ordered set U_map consisting of a series of Γm calculated by a frame-level mapping function. It is arranged in the order of the target frame set Tseq according to the time axis and records the comprehensive growth status index of the root region in each frame during the entire target observation period. This map not only retains the temporal index information of the original image frames, but also embeds the dynamic feature parameters of the root region, making it a high-level structured data representation that supports trend evolution modeling, regional level determination and subsequent response analysis.

[0029] Furthermore, to achieve transferable analysis capabilities under diverse environmental and species conditions, this invention employs the following combined expression in the calculation structure of the comprehensive growth state index Γm:

[0030] Among them, the ln(1+α·AAR) term improves the sensitivity to weak changes through logarithmic transformation, which is especially suitable for the case where the initial fluctuation of the root tip growth signal of seedlings is not obvious, while (1+β·IDG) retains the weighted influence of density growth on the overall structure formation; the product of the two constitutes a nonlinear coupling mechanism, which enables the model to have higher discrimination and trend amplification ability when dealing with strong and weak root regions. The coefficients α and β are not fixed constants, but can be adaptively set according to the application object, root zone imaging quality and seedling type. α is usually used to adjust the sensitivity to AAR fluctuations, and β controls the threshold of structural response. The two can be used for sensitivity regression within a specified sliding time window to improve the stability and generalization ability of index calculation.

[0031] In practical applications, to obtain α and β coefficients suitable for specific seedling species and planting conditions, this invention uses the following parameter optimization process for modeling training and fitting: In the early stage of transplanting, root zone image sequences were selected from several representative samples, and the corresponding root tip activity change rate (AAR) and root incremental density gradient were extracted from each frame of the image. A gridded traversal was performed on the α and β values ​​under different combinations, the comprehensive growth state index Γm sequence corresponding to each combination was calculated, and the correlation between its change trend and subsequent real survival results was evaluated at the frame level. Based on the survival rate data, paired labels are constructed, and the fluctuation characteristics of the comprehensive growth status index Γm sequence and the growth outcome are mapped to the scoring model. The optimal parameter pair (α*, β*) is determined by minimizing the prediction error. The scoring model, as explained, is a mathematical function or statistical learning framework built upon historical labeled data. It measures the consistency or predictive ability between the comprehensive growth status index Γm generated by different parameter combinations and actual physiological performance (such as survival status, root zone recovery level, etc.). The scoring model uses Γm features extracted from sample frame sequences (such as mean, variation rate, trend slope, etc.) as input variables and the actual labeled growth performance as the output root zone growth trend status. A fitting function is obtained through training using regression or classification algorithms (such as logistic regression, support vector machines, or gradient boosting trees). Evaluation metrics output by the scoring function (such as prediction accuracy, mean squared error, or F1 score) are used to quantify the Γm calculation effect under different parameter settings. Finally, the parameter combination with the optimal evaluation metrics is selected as the model's optimal solution, thus achieving a closed-loop mapping between the root index generation mechanism and physiological performance, ensuring that the calculation results have biological relevance and practical guiding value. Using the optimal parameter pair as the initial setting, and combining the statistical characteristics of the sliding time window constructed from sampling data under different ecological conditions or soil disturbance, Bayesian optimization or genetic algorithm fine-tuning is performed to obtain a robust adaptive parameter adjustment mechanism. The final fitted parameter pairs are embedded into the mapping frame-level mapping function and deployed in an online image analysis system to generate a comprehensive growth status index.

[0032] Step S4 Summary and Technical Effects: This step focuses on the temporal structural changes between frames of the image. From spatial registration, edge matching, dynamic vector extraction, active region modeling, and incremental density direction calculation, to the final generation of a structured update index map U_map, it completes the technical leap from structural images to dynamic behavior mapping. Unlike traditional frame difference analysis based solely on grayscale changes, this scheme introduces edge migration and spatial directionality analysis mechanisms, making root change analysis more interpretable and morphologically stable, and is particularly suitable for early transplant root zone scenarios with complex structures and discontinuous changes.

[0033] Furthermore, step S5: The current growth trend status of the root zone is categorized and labeled based on the comprehensive growth status index Γm. The specific operation is as follows: After obtaining the comprehensive growth state index Γm, trend state identification processing is performed according to its distribution in the time series. This processing is based on the preset trend state judgment interval, dividing the numerical range of the comprehensive growth state index Γm into several level segments, with each level corresponding to a root zone growth trend state. During execution, the growth trend status of the root region is determined based on the slope of the change of the comprehensive growth status index Γm in consecutive frames and the interval landing point, and the root region growth trend status sequence is recorded in chronological order. The generated root zone growth trend state sequence is subjected to temporal continuity analysis. If there are frequent switching of root zone growth trend state or local fluctuations and instability, it is further corrected according to the comprehensive growth state index Γm value of adjacent frames and the boundary fluctuation situation, so that the output root zone growth trend state has temporal consistency and structural rationality. The above root zone growth trend state sequence is used to represent the growth trend type of the root zone in the current monitoring period, providing a basis for subsequent growth level classification and maintenance response.

[0034] Furthermore, step S5: Root region activity level model construction and structural map output To achieve the hierarchical classification of root zone growth trend status and the partitioned output of regional response suggestions, this step, based on the comprehensive growth status index sequence Γm and its trend label results generated in step S4, and combined with time evolution stability analysis, constructs a composite root zone structure map that includes level determination, confidence interval extraction, and response suggestion generation; specifically including: Sub-step 501: Trend level generation and level boundary delineation Based on the full sequence distribution of the comprehensive growth state index Γm, a hierarchical model based on interval division is constructed. The value of the comprehensive growth state index Γm is divided into multiple continuous level segments according to the statistical percentile method. Each level corresponds to a growth state level label and is bound to the corresponding trend evolution direction (enhancement, decay, or stability). To ensure the stability of the boundary interval judgment, the boundary is corrected by combining the slope of the change of the comprehensive growth state index sequence Γm at the trend label switching position to obtain the final boundary interval of each level. Sub-step 502: Confidence interval extraction and confidence weighting Based on the classification of grade boundaries, the fluctuation range and slope variance of the comprehensive growth status index Γm of each trend label within a specified time window are statistically analyzed to form a confidence index vector. Based on the confidence index vector, each frame-level label is assigned a confidence level within a confidence interval and marked in the label sequence as a basis for the stability of subsequent response suggestion judgment. Sub-step 503: Structural map spatial mapping and response suggestion generation By combining the spatial raster division structure of the original image frames, the trend labels with labeled levels and confidence are mapped to the corresponding sub-regions of the root region image. Auxiliary features such as the mean value of the comprehensive growth status index sequence Γm and the density change amplitude within the region are integrated to comprehensively determine the growth activity level. Based on the preset response rule library, corresponding response suggestion labels are generated for each sub-region, such as "under observation", "priority intervention" or "no processing required", and encapsulated into a root region structure map containing four fields: location coordinates, level value, confidence, and response suggestion.

[0035] To achieve the above objectives, the present invention provides the following technical solution: an image analysis system for root growth status of seedlings after transplanting, comprising: Image acquisition module: A shallow-buried optical detection component is installed in the target root zone after the seedling transplant is completed. Near-infrared images are continuously acquired through time-sharing control to generate a sequence of multiple original image frames. The acquisition process ensures the consistency of the spatial distribution of the images and their comparability on the time axis. The output is a sequence of original image frames arranged in chronological order, with time labels and spatial indexes, providing a basic data source for subsequent image processing and dynamic change recognition. Image preprocessing module: Taking the original image frame sequence as input, it performs structural equalization processing on each frame image, including smoothing the reflection distribution and softening the root region edges, in order to improve the separation clarity between the root region and the background and construct a highly recognizable preprocessed image frame. Dynamic modeling module: Taking preprocessed image frames as input, it analyzes the migration vector field of the root region edge between adjacent frames and extracts the regional incremental features, which include root frame difference map, gray level increment, density increase map and local density direction change rate map, and constructs a dynamic parameter sequence for root growth. The growth state generation module takes the incremental features of the region as input, calculates the root tip activity change rate (AAR) and root system incremental density gradient (IDG), inputs the AAR and IDG to a preset frame-level mapping function, and outputs the comprehensive growth state index Γ corresponding to each frame image. m ; The grading analysis module constructs a root zone activity grading model based on the root zone growth trend status and the distribution map of comprehensive growth status indicators. It outputs a root zone structure map that includes growth status grading, confidence level of confidence intervals, and response suggestions, providing qualitative judgment basis and distributed response support for the decision-making system.

[0036] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for analyzing root growth status images after seedling transplantation, characterized in that, include: Step S1: After the seedling transplantation is completed, a shallow-buried optical detection component is installed in the target root zone. Multiple frames of near-infrared images are collected in a continuous time period using a time-sharing control method, and the original image frames are established. Step S2: Perform structural equalization preprocessing on each frame of the original image frame, and construct a preprocessed image frame by smoothing the reflection distribution and softening the boundaries. Step S3: Using the preprocessed image frames as input, extract the migration vector field and regional increment features of the root region edge between consecutive image frames to form a root growth dynamic parameter sequence based on the time axis. The regional increment features include at least the root frame difference map, grayscale increment, density increment map and local density direction change rate map. Step S4: Calculate the root tip activity change rate (AAR) and root system incremental density gradient (IDG) respectively. The root tip activity change rate (AAR) is obtained based on the time derivative of the root tip normal vector. The root system incremental density gradient is obtained by extracting the density increase and gradient direction of the root system region in the partitioned grid. Input the root tip activity change rate and root system incremental density gradient into a preset frame-level mapping function, and output a comprehensive growth status index Γm that reflects the root growth activity of the corresponding image frame. Step S5: Construct a root zone activity level model based on the root zone growth trend status and the distribution map of comprehensive growth status indicators, and output a root zone structure map that includes growth status level, confidence level of confidence interval and response suggestions.

2. The method for analyzing root growth status images after seedling transplantation according to claim 1, characterized in that, When acquiring raw image frames, each image frame is bound to a spatial index and a time tag based on the seedling identification tag, forming a traceable dual-index image data structure. In subsequent processing, the image data is retrieved based on this index. Frame-level resampling and registration operations are performed on the acquired raw image frames. Inter-frame aligned image sequences are constructed through nonlinear time interpolation and local elastic registration algorithms to improve the consistency and accuracy of subsequent dynamic parameter calculations.

3. The method for analyzing root growth status images after seedling transplantation according to claim 2, characterized in that, During the construction of the original image frames, the sampling method of the image frames is a differential resampling operation based on the intensity of pixel changes between frames. During the time period when the root region is growing rapidly or undergoing significant structural changes, the sampling density of the image frames is increased to enhance the temporal coverage of key dynamic changes. For static regions where the intensity of structural changes between consecutive frames is lower than a preset threshold, the sampling frequency is reduced to avoid redundant image acquisition. This ensures the complete expression of key growth information while improving the processing efficiency of the image sequence and the response and focusing ability of the subsequent model.

4. The method for analyzing root growth status images after seedling transplantation according to claim 3, characterized in that, The preprocessing of the original image frame includes: The original image frame set is scheduled frame by frame and pixel normalized to construct a preprocessed image with a uniform brightness range; Perform reflectance feature enhancement and bidirectional moving average brightness smoothing to generate a reflectance feature enhanced image; The edge enhancement operator guided by the structural gradient direction extracts the edge features of the root region and generates the initial root region target contour map. Based on the spatial structure offset ΔRg between adjacent frames, the system dynamically identifies regions with unchanged continuous structures as false recognition signals, performs mask removal and cleaning, and outputs preprocessed image frames.

5. The method for analyzing root growth status images after seedling transplantation according to claim 4, characterized in that, The extraction of the root growth dynamic parameter sequence includes: Based on the structure enhancement map package, frame-level structure images and corresponding initial root region target contour maps RGm are extracted from the target frame set in a time series. Using any two adjacent frames in the inter-frame structure-aligned image sequence as processing units, for the corresponding initial root region target contour map, the edge point set is extracted, and the set of corresponding edge point pairs is generated by minimum distance addition matching. The vector migration value between all point pairs is calculated to construct the migration vector field Vm of the root region edge; the set of migration vector fields on the time axis is constructed for the entire sequence of frames. The frame difference map is formed by subtracting adjacent frames. The grayscale increment of the two frames within the initial root region target contour map RGm mask range is calculated. The original grid structure is partitioned, and the pixel increment density per unit area of ​​each sub-region is calculated to form a density increase map. At the same time, the gradient direction field is extracted in the density increase map to construct a local density direction change rate map and generate the root system incremental density gradient IDGm.

6. The method for analyzing root growth status images after seedling transplantation according to claim 1, characterized in that, The comprehensive growth status index is obtained as follows: Based on the root tip activity change rate and root incremental density gradient output frame by frame from the target frame set, the root growth dynamic parameter sequence is reorganized on the time axis, and a frame-level mapping function Γm=f(AAR,IDGm) is constructed to generate a comprehensive growth status index Γm. The following combined expression is used in the calculation structure of the comprehensive growth state index Γm: ; Among them, the ln(1+α·AAR) term improves the sensitivity to weak changes through logarithmic transformation, while (1+β·IDG) retains the weighted influence of density growth on the overall structure formation; α is used to regulate the sensitivity to fluctuations in the root tip activity rate AAR, and β controls the threshold of the root incremental density gradient.

7. The method for analyzing root growth status images after seedling transplantation according to claim 6, characterized in that, The α and β coefficients are obtained as follows: In the early stage of transplanting, root zone image sequences were selected from several representative samples, and the corresponding root tip activity change rate (AAR) and root incremental density gradient were extracted from each frame of the image. A gridded traversal was performed on the α and β values ​​under different combinations, the comprehensive growth state index Γm sequence corresponding to each combination was calculated, and the correlation between its change trend and subsequent real survival results was evaluated at the frame level. Based on survival rate data, paired labels are constructed, and the fluctuation characteristics of the comprehensive growth status index Γm sequence and growth outcome are mapped to the scoring model. The optimal parameter pair is determined by minimizing the prediction error. Using the optimal parameter pair as the initial setting, and combining the statistical characteristics of the sliding time window constructed from sampling data under different ecological conditions or soil disturbance, Bayesian optimization or genetic algorithm fine-tuning is performed to obtain a robust adaptive parameter adjustment mechanism. The final fitted parameter pairs are embedded into the mapping frame-level mapping function and deployed in an online image analysis system to generate a comprehensive growth status index.

8. The method for analyzing root growth status images after seedling transplantation according to claim 6, characterized in that, The current growth trend of the root zone is categorized and labeled based on the comprehensive growth status index Γm. The specific operation is as follows: After obtaining the comprehensive growth state index Γm, trend state identification processing is performed according to its distribution in the time series. This processing is based on the preset trend state judgment interval, dividing the numerical range of the comprehensive growth state index Γm into several level segments, with each level corresponding to a root zone growth trend state. During execution, the growth trend status of the root region is determined based on the slope of the change of the comprehensive growth status index Γm in consecutive frames and the interval landing point, and the root region growth trend status sequence is recorded in chronological order. The generated root region growth trend state sequence is subjected to temporal continuity analysis. If there are frequent switching of root region growth trend state or local fluctuations and instability, the sequence is corrected according to the comprehensive growth state index Γm value of adjacent frames and the boundary fluctuation situation, so that the output root region growth trend state has temporal consistency and structural rationality.

9. The method for analyzing root growth status images after seedling transplantation according to claim 8, characterized in that, Based on the classification of grade boundaries, the fluctuation range and slope variance of the comprehensive growth status index Γm of each trend label within a specified time window are statistically analyzed to form a confidence index vector. Based on the confidence index vector, each frame-level label is assigned a confidence level within a confidence interval and marked in the label sequence as a stability basis for subsequent response suggestion judgment.

10. A root growth status image analysis system for seedlings after transplanting, characterized in that, include: Image acquisition module: A shallow-buried optical detection component is installed in the target root zone after the seedling transplant is completed. Near-infrared images are continuously acquired through time-sharing control to generate a sequence of multiple original image frames. Image preprocessing module: Taking the original image frame sequence as input, it performs structural equalization processing on each frame image, including smoothing the reflection distribution and softening the root region edges, in order to improve the separation clarity between the root region and the background and construct a highly recognizable preprocessed image frame. Dynamic modeling module: Taking preprocessed image frames as input, it analyzes the migration vector field of the root region edge between adjacent frames and extracts the regional incremental features, which include root frame difference map, gray level increment, density increase map and local density direction change rate map, and constructs a dynamic parameter sequence for root growth. The growth state generation module takes the incremental features of the region as input, calculates the root tip activity change rate (AAR) and root system incremental density gradient (IDG), inputs the AAR and IDG to a preset frame-level mapping function, and outputs the comprehensive growth state index Γ corresponding to each frame image. m ; The grading analysis module constructs a root zone activity grading model based on the root zone growth trend status and the distribution map of comprehensive growth status indicators, and outputs a root zone structure map that includes growth status grading, confidence level of confidence intervals, and response suggestions.

Images