Tree ring width measurement error correction method

Abstract

The invention relates to the field of tree chronology and tree survey research, in particular to a tree ring width measurement error correction method, which comprises the following steps of: acquiring a ring sample image; performing multi-band adaptive enhancement processing on the annual ring sample image to obtain an enhanced annual ring image; based on the enhanced annual ring image, adopting a multi-strategy boundary detection method to identify an annual ring boundary; according to the annual ring boundary, dynamic measurement path configuration is determined, and the dynamic measurement path configuration comprises a plurality of measurement paths; obtaining coordinates of intersection points of the annual ring boundary and the plurality of measurement paths; based on the intersection point coordinates, a complete annual ring boundary curve is reconstructed through multi-dimensional curve fitting; calculating local inclination angle distribution according to the annual ring boundary curve; based on the local inclination angle distribution, using a regionalized adaptive correction model to correct the annual ring width; and generating corrected annual ring width data and a quality evaluation report, and adopting a multi-strategy boundary detection method to realize automatic and high-precision identification of the annual ring boundary.

Description

Technical Field

[0001] The present invention relates to the fields of dendrochronology and forest mensuration, and particularly to a method for correcting the measurement error of tree ring width, which is used to improve the accuracy of measuring the width of inclined tree rings. Background Art

[0002] The measurement of tree ring width is the basic data for dendrochronology and forest mensuration research. Currently, the common method for detecting tree rings in China is to take samples with an increment borer. When taking samples with an increment borer, the situation of drilling off often occurs because the position of the pith cannot be correctly judged. Most of the tree rings obtained in this way are in an inclined state (the tangential angle is not zero).

[0003] In the prior art, for example, CN101922916A discloses a method for correcting the measurement error of tree ring width. This method sets two measurement lines above and below the scanned tree ring image, measures the upper and lower coordinates of the intersection points of each tree ring with the measurement lines, and calculates the average value to obtain the middle coordinate. The difference between the abscissas of adjacent middle coordinates is the width of the inclined tree ring to be corrected. The inclination angle of the tree ring boundary, that is, the tangential angle, can be calculated according to the difference between the vertical and horizontal coordinates of the upper and lower two endpoints. The average value of the inclination angles of the inner and outer boundaries of each tree ring is used as the inclination angle of the tree ring. According to the principle of trigonometric functions, the inclined tree rings are corrected.

[0004] However, the above method has the following deficiencies: 1) Only two parallel lines with a fixed spacing are used as the measurement benchmark, which cannot adapt to samples of different sizes and shapes; 2) It is necessary to manually mark the intersection points of each tree ring and the measurement lines, with a large workload and easy to introduce human errors; 3) Lack the ability to automatically identify the boundaries of tree rings, especially for fuzzy or atypical tree rings; 4) Simply take the average value of the inclination angles of the two boundaries, ignoring the complexity of the tree ring morphology; 5) Have high requirements for the quality of the original image and lack an image enhancement and preprocessing mechanism; 6) Unable to effectively identify and process abnormal situations such as false rings and missing rings. These problems limit its accuracy and efficiency in practical applications. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for correcting the measurement error of tree ring width to solve the problems existing in the prior art.

[0006] A method for correcting the measurement error of tree ring width includes: Obtain a tree ring sample image; Perform multi-band adaptive enhancement processing on the tree ring sample image to obtain an enhanced tree ring image; Based on the enhanced tree ring image, use a multi-strategy boundary detection method to identify the boundaries of the tree rings; According to the tree ring boundaries, determine a dynamic measurement path configuration, where the dynamic measurement path configuration includes multiple measurement paths; Obtain the intersection coordinates of the annual ring boundary and the multiple measurement paths; Based on the intersection coordinates, reconstruct the complete annual ring boundary curve through multi-dimensional curve fitting; Calculate the local inclination angle distribution according to the annual ring boundary curve; Based on the local inclination angle distribution, use a regionalized adaptive correction model to correct the annual ring width; Generate the corrected annual ring width data and a quality assessment report.

[0007] Preferably, the multi-band adaptive enhancement processing of the annual ring sample image specifically includes: Decompose the annual ring sample image into three frequency bands: high frequency, medium frequency, and low frequency; Perform detail enhancement and noise suppression processing on the high frequency band; Perform edge sharpening and texture enhancement processing on the medium frequency band; Perform contrast optimization and illumination equalization processing on the low frequency band; Based on the tree species characteristic parameters, adaptively adjust the enhancement parameters of each frequency band; Fuse the processed three frequency bands to obtain the enhanced annual ring image.

[0008] Preferably, the multi-strategy boundary detection method is used to identify the annual ring boundary, specifically including: Perform edge detection based on gradient change to obtain the first boundary detection result; Perform region segmentation based on texture features to obtain the second boundary detection result; Perform periodic detection based on morphological features to obtain the third boundary detection result; Perform weighted fusion on the first boundary detection result, the second boundary detection result, and the third boundary detection result to obtain a fused boundary; Perform continuity optimization and anomaly detection on the fused boundary to obtain the annual ring boundary.

[0009] Preferably, the continuity optimization and anomaly detection of the fused boundary specifically include: Detect the breakpoints and isolated points in the fused boundary; Based on the boundary continuity constraint, repair the breakpoints and remove the isolated points; Identify the abnormal annual rings in the fused boundary, including false rings, missing rings, and traumatic rings; According to the annual ring growth law, mark or correct the abnormal annual rings; Generate a boundary quality score to identify the areas that require manual intervention.

[0010] Preferably, the determination of the dynamic measurement path configuration specifically includes: Analyze the morphological characteristics of the annual ring boundary, including the overall shape, eccentricity, and complexity; According to the morphological characteristics and sample size, calculate the number of optimal measurement paths, ranging from 3 to 9; For circular annual rings, configure a radial measurement path layout; For elliptical annual rings, increase the measurement path density along the long and short axes; For irregular annual rings, dynamically adjust the measurement path distribution according to the morphological complexity; Evaluate the coverage uniformity of the measurement paths to ensure that key areas are fully measured.

[0011] Preferably, the reconstruction of the complete annual ring boundary curve by multi-dimensional curve fitting specifically includes: Based on the intersection coordinates, determine the preliminary contour of the annual ring boundary; According to the complexity of the annual ring boundary, adaptively select curve fitting methods, including: When the boundary is relatively regular, use piecewise polynomial fitting; When the boundary is moderately complex, use spline curve fitting; When the boundary is highly complex, use non-parametric curve fitting; Optimize the fitting result to ensure that the boundary is smooth and continuous and conforms to the biological growth law; Generate a mathematical model of the complete annual ring boundary to support the accurate coordinate calculation of any point.

[0012] Preferably, the calculation of the local inclination angle distribution specifically includes: Uniformly sample multiple points on the annual ring boundary curve; Calculate the tangent direction at each sampling point to obtain the local tangent vector; Based on the local tangent vector, calculate the local inclination angle of each sampling point; Establish a local inclination angle distribution map to describe the inclination characteristics of the entire annual ring boundary; Identify the regions of sudden change in the inclination angle, divide the inclination angle regions, and provide a basis for the selection of the correction model.

[0013] Preferably, the correction of the annual ring width using a regionalized adaptive correction model specifically includes: According to the local inclination angle distribution, divide the annual ring boundary into multiple correction regions; For regions with small changes in the inclination angle, apply a uniform inclination model; For regions with a gradient change in the inclination angle, apply a gradient inclination model; For regions with a sudden change in the inclination angle, apply a piecewise inclination model; Calculate the correction parameters for each region separately and perform precise correction; Calculate the true width between adjacent annual ring boundaries and generate corrected annual ring width data.

[0014] Preferably, the generation of the corrected annual ring width data and the quality assessment report specifically includes: Evaluate various sources of uncertainty in the measurement process, including uncertainties caused by image resolution, boundary recognition, tilt angle measurement, and correction model selection; Calculate the combined uncertainty of each annual ring width; Grade the quality of the annual ring width data according to the magnitude of the uncertainty; Generate a standard format output containing the original data, corrected data, correction parameters, and quality ratings; Provide data visualization results to intuitively display the measurement and correction effects.

[0015] Preferably, before the multi-band adaptive enhancement processing, it further includes: Obtain tree species characteristic information; Extract the corresponding image processing parameters from the pre-established tree species characteristic database according to the tree species characteristic information; Configure the parameters of the multi-band adaptive enhancement processing using the image processing parameters; Among them, the tree species characteristic information includes tree species type, annual ring density, early and late wood contrast, and boundary clarity.

[0016] The present invention realizes high-precision automatic correction of annual ring width measurement by introducing technologies such as multi-band adaptive image enhancement, multi-strategy boundary detection, dynamic measurement path configuration, multi-dimensional curve fitting reconstruction, and regionalized adaptive correction.

[0017] The present invention has the following beneficial effects: 1. Through multi-band adaptive enhancement processing, the quality of the annual ring image is significantly improved, making the boundary features clearer, laying a foundation for subsequent analysis; 2. Adopting a multi-strategy boundary detection method, automatic and high-precision recognition of annual ring boundaries is achieved, reducing manual intervention and improving work efficiency; 3. Introducing dynamic measurement path configuration, the measurement strategy is adaptively adjusted according to the morphological characteristics of the annual rings, adapting to different morphological annual ring samples; 4. Reconstruct the complete annual ring boundary through multi-dimensional curve fitting to comprehensively capture the morphological characteristics of the annual rings and improve the integrity of the measurement; 5. Based on the local tilt angle distribution, using a regionalized adaptive correction model, refined correction is achieved, greatly improving the accuracy of annual ring width measurement; 6. A complete quality assessment system is established to provide reliable uncertainty assessment for measurement results and enhance data credibility.

[0018] In summary, for the tree ring width measurement error correction method provided by the present invention, compared with the prior art, the measurement accuracy is increased by 40% - 60%, the degree of automation is significantly improved, the adaptability is stronger, and a more reliable data basis is provided for dendrochronology and paleoclimate research. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a flowchart of the tree ring width measurement error correction method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0021] Please refer to Figure 1 , Figure 1 is a flowchart of the tree ring width measurement error correction method provided by the present invention. As Figure 1 shown, the method includes the following steps: Step S101: Obtain the tree ring sample image; Step S102: Perform multi - band adaptive enhancement processing on the tree ring sample image to obtain the enhanced tree ring image; Step S103: Based on the enhanced tree ring image, use a multi - strategy boundary detection method to identify the tree ring boundary; Step S104: Determine the dynamic measurement path configuration according to the tree ring boundary, where the dynamic measurement path configuration includes multiple measurement paths; Step S105: Obtain the intersection coordinates of the tree ring boundary and the multiple measurement paths; Step S106: Based on the intersection coordinates, reconstruct the complete tree ring boundary curve through multi - dimensional curve fitting; Step S107: Calculate the local inclination angle distribution according to the tree ring boundary curve; Step S108: Based on the local inclination angle distribution, use a regionalized adaptive correction model to correct the tree ring width; Step S109: Generate the corrected tree ring width data and quality assessment report.

[0022] In one embodiment of the present invention, in step S101, the acquisition of the annual ring sample image can be achieved in various ways, including but not limited to high-resolution scanners, microscope cameras or digital cameras. Preferably, a scanner with a resolution of not less than 1200 dpi is used to acquire the annual ring sample image to ensure that the details of the annual ring boundaries in the image are clearly distinguishable. When acquiring the image, the standard color card and scale information can be collected simultaneously for subsequent image calibration to ensure the accuracy of the measurement.

[0023] In step S102, multi-band adaptive enhancement processing is performed on the annual ring sample image, specifically including: The annual ring sample image is decomposed into three frequency bands: high frequency, medium frequency and low frequency. In a preferred embodiment of the present invention, wavelet transform is used for image decomposition. Specifically, Daubechies wavelets (such as db4 or db6) can be selected for multi-scale decomposition, and the image is decomposed into 3-5 scale levels. The mathematical expression of wavelet transform is: , Where: is the wavelet coefficient, is the scale parameter, is the translation parameter, is the input signal (in this case, the image), is the conjugate of the wavelet basis function. Detail enhancement and noise suppression processing are performed on the high-frequency band. The high-frequency band contains the detail information of the annual ring boundaries, but also contains noise. The present invention uses an adaptive threshold method for processing to enhance the boundary details while suppressing the noise. Preferably, the following formula is used to adjust the high-frequency coefficients: , Where: is the adjusted high-frequency coefficient, is the original high-frequency coefficient, is the pixel coordinate, is the enhancement factor (usually taking values between 0.5-2.0), is the noise threshold (empirical value: 10-15 percentile of the absolute value of the high-frequency coefficient), is the maximum threshold (usually 95 percentile of the absolute value of the high-frequency coefficient), is the indicator function, which takes the value of 1 when the condition is met, otherwise 0.

[0024] Edge sharpening and texture enhancement processing are performed on the medium-frequency band. The medium-frequency band contains the main structural information of the annual rings. Enhancing this band helps to highlight the annual ring boundaries. The present invention uses a directional enhancement filter for processing, and its mathematical expression is: , Where: is the adjusted intermediate frequency coefficient, is the original intermediate frequency coefficient, is the enhancement coefficient (usually taking values from 0.3 to 0.8), is the directional weight, related to the consistency of the local gradient direction, and the calculation formula is: , where: is the point 's neighborhood (usually taking or ), represents the gradient operator, is a small constant (such as 0.0001) to prevent the denominator from being zero. Contrast optimization and illumination equalization processing are performed on the low-frequency band. The low-frequency band mainly contains the background information and overall brightness distribution of the image. The present invention adopts a method combining histogram equalization and gamma correction, and its processing formula is: , where: is the adjusted low-frequency coefficient, is the original low-frequency coefficient, represents the histogram equalization operation, is the gamma correction coefficient (usually between 0.8 and 1.2, adjusted according to the image characteristics), is the scaling coefficient, used to adjust the overall brightness level, and usually takes values such that the average brightness after processing is close to that of the original image.

[0025] Based on the tree species characteristic parameters, the enhancement parameters of each frequency band are adaptively adjusted. The present invention establishes a database containing the characteristic parameters of the annual rings of common tree species, such as the typical parameters of coniferous trees (pine, spruce, etc.) and broad-leaved trees (oak, Manchurian ash, etc.). For coniferous trees, since the boundary between early and late wood is usually relatively clear, the intermediate frequency enhancement coefficient can be appropriately increased (such as taking 0.6 - 0.8); for broad-leaved trees, since the boundary is relatively blurred and high-frequency detail enhancement is required, the high-frequency enhancement factor can be appropriately increased (such as taking 1.5 - 2.0). The processed three frequency bands are fused to obtain the enhanced annual ring image. Preferably, the following formula is used for frequency band fusion: , where: is the enhanced image, representing the pixel value at the position after processing; IDWT represents the inverse discrete wavelet transform, which is the inverse operation of the wavelet transform and is used to reconstruct the coefficients in the transform domain into an image in the spatial domain; are the processed low-frequency, intermediate frequency, and high-frequency coefficients respectively.

[0026] In a preferred embodiment of the present invention, before step S102, it further includes: obtaining tree species characteristic information; extracting corresponding image processing parameters from a pre-established tree species characteristic database according to the tree species characteristic information; configuring parameters for the multi-band adaptive enhancement processing using the image processing parameters; wherein the tree species characteristic information includes tree species type, annual ring density, earlywood-latewood contrast, and boundary clarity.

[0027] The tree species characteristic information can be obtained in various ways, such as user input or automatic recognition. Automatic recognition can be achieved by using machine learning methods based on image texture features and annual ring morphological features. In an embodiment of the present invention, a random forest classifier based on texture features is used for tree species recognition, and the accuracy rate can reach 85% - 90%. The tree species characteristic database contains image processing parameters of common tree species, such as typical parameter settings for coniferous trees (pine, spruce, etc.) and broad-leaved trees (oak, ash, etc.).

[0028] In step S103, a multi-strategy boundary detection method is used to identify the annual ring boundary, specifically including: Performing edge detection based on gradient change to obtain a first boundary detection result. In a preferred embodiment of the present invention, a method combining the Sobel operator and the Canny edge detector is used. First, calculate the image gradient: , , Where: and are the gradients in the horizontal and vertical directions respectively, representing the change rates of the image in the horizontal and vertical directions at position ; and are the Sobel operator kernels, which are the convolution kernels in the horizontal and vertical directions respectively; is the input image; is the pixel coordinate; represents the double summation for convolution operation within the neighborhood.

[0029] Then, calculate the gradient magnitude and direction: , , Where: is the gradient magnitude, representing the intensity of the gradient; is the gradient direction, representing the angle of the gradient; is the arctangent function, used to calculate the angle.

[0030] Next, non-maximum suppression and double-threshold processing are applied to obtain edge pixels. For the tree-ring image, the low threshold is usually set to the 20-30 percentile of the gradient magnitude histogram, and the high threshold is set to the 60-70 percentile.

[0031] Region segmentation is performed based on texture features to obtain the second boundary detection result. The method combines local binary pattern (LBP) texture features and the watershed algorithm. First, calculate the LBP features: , where: is the central point is the LBP value of, representing the local texture feature; is the number of sampling points, indicating the number of sampling points evenly distributed on the circumference with radius , usually taken as 8; is the sampling radius, representing the radius of the sampling circle, usually taken as 1 or 2; is the gray value of the central point; is the gray value of the th sampling point; is the step function, taking 1 when and 0 otherwise represents the summation over all sampling points; represents the weight for converting the binary result to a decimal value.

[0032] Then, based on the LBP feature map, the watershed algorithm is applied to obtain the region segmentation result. Finally, the region boundary is extracted as the second boundary detection result.

[0033] Periodicity detection is performed based on morphological features to obtain the third boundary detection result. An important feature of tree rings is their periodic distribution. This invention utilizes this characteristic for detection. The specific method is: First, perform a radial cumulative projection on the image: , where: is the projection value in the direction of angle , representing the cumulative sum of pixels along this direction; is the radial distance, indicating the distance from the origin (usually the pith position) to the pixel point; and are the minimum and maximum radial distances respectively; is the image value in the polar coordinate system; represents the summation operation within the radial range. Then, perform Fourier analysis on the projection curve to detect the periodic components. Finally, based on the detected periodicity, predict the possible positions of the tree rings to generate the third boundary detection result.

[0034] Perform weighted fusion on the first boundary detection result, the second boundary detection result, and the third boundary detection result to obtain a fused boundary. The present invention adopts a weighted fusion method based on local consistency, and the fusion formula is: , Where: is the fused boundary, representing the fusion result at position ; is the th boundary detection result, corresponding to the results of three detection methods respectively; is the local weight, representing the reliability of the th method at position ; represents the weighted sum of the results of the three methods. The calculation formula of , Where: is the global confidence coefficient, based on the overall reliability of each method, usually set to 0.3 - 0.5; is the local consistency influence factor, controlling the importance of local consistency, usually taking 0.5 - 1.0; is the local consistency metric, representing the degree of consistency of the result of the th method with other methods near point , and the value range is [0, 1].

[0035] Perform continuity optimization and anomaly detection on the fused boundary to obtain the annual ring boundary. Continuity optimization includes break point connection and isolated point removal. In a preferred embodiment of the present invention, the following method is adopted for break point connection: For break points with a distance less than a preset threshold (usually set to 10% of the average annual ring width), use a Bessel curve to connect them to maintain the smoothness of the boundary. For isolated points, if the distance from them to the nearest boundary is greater than the threshold (usually set to 50% of the average annual ring width), then remove them.

[0036] In a preferred embodiment of the present invention, in step S103, performing continuity optimization and anomaly detection on the fused boundary specifically includes: Detect break points and isolated points in the fused boundary. A break point refers to a gap or discontinuous area in the boundary, and an isolated point refers to an isolated pixel or small area far from the main boundary. The present invention adopts a connected component analysis method for detection, and the specific steps are as follows: First, perform binary processing on the fused boundary image; then, perform connected component labeling; finally, analyze the size and positional relationship of each connected component to identify break points and isolated points.

[0037] Based on the boundary continuity constraint, repair the breakpoints and remove the isolated points. For the breakpoints, the present invention uses a curve fitting method for repair. Preferably, a cubic Bezier curve is used to connect the two ends of the breakpoint to ensure the continuity and smoothness of the curve. The parametric expression of the Bezier curve is: , where: is a point on the curve, representing the parameter corresponding curve point coordinates; and are the known boundary points at both ends of the breakpoint, representing the starting point and the ending point of the curve; and are control points, determined by maintaining the continuity of the tangent direction; is a parameter, with a value range of [0,1]; , , and are Bezier basis functions, controlling the shape of the curve. For isolated points, if their area is less than the threshold (usually set to 5% of the average annual ring area) and the distance to the nearest boundary is greater than the threshold (usually set to 50% of the average annual ring width), then they are removed.

[0038] Identify the abnormal annual rings in the fusion boundary, including false rings, missing rings, and traumatic rings. The identification of abnormal annual rings is based on morphological characteristics and growth rules. The specific methods include: (1) False ring identification: False rings are usually narrow and discontinuous. The present invention adopts the following judgment criterion: If the width of the annual ring is less than 30% of the average width of normal annual rings and the circumferential continuity is lower than 70%, it is marked as a potential false ring.

[0039] (2) Missing ring identification: By analyzing the spacing between adjacent annual rings, if the spacing at a certain place exceeds 1.8 times the normal spacing, there may be a missing ring.

[0040] (3) Traumatic ring identification: Traumatic rings usually show local distortion. The present invention identifies traumatic rings by analyzing the local curvature anomaly value. If the local curvature exceeds 3 times the average curvature and the affected range is limited, it may be a traumatic ring.

[0041] According to the annual ring growth rule, mark or correct the abnormal annual rings. For the identified abnormal annual rings, the present invention adopts different treatment strategies: (1) False rings: Marked as false rings and excluded in subsequent measurements; (2) Missing rings: Estimate the possible position and width of the missing ring according to the width characteristics of adjacent annual rings, and mark it as an estimated value in the result; (3) Trauma wheel: Keep the trauma wheel, but adjust the measurement path in the affected area to avoid the distorted part.

[0042] Generate a boundary quality score to identify areas that require manual intervention. The present invention establishes a boundary quality scoring system based on the following indicators: boundary continuity, morphological regularity, consistency with the expected pattern, etc. The quality score uses a quantitative scale of 0 - 100, and areas with a score below 60 are marked as requiring manual intervention. In practical applications, the system will generate a boundary quality distribution map to visually display the quality status of each area, assisting researchers in quickly locating problem areas.

[0043] In step S104, determine the dynamic measurement path configuration, specifically including: Analyze the morphological characteristics of the annual ring boundary, including the overall shape, eccentricity, and complexity. In a preferred embodiment of the present invention, the morphological characteristic analysis uses the following methods: (1) Overall shape: Determine the overall shape by fitting an ellipse and calculating the main axis direction. The center , major axis a, minor axis b, and rotation angle θ of the fitted ellipse are calculated by the least squares method.

[0044] (2) Eccentricity: Calculate the eccentricity e using the parameters of the fitted ellipse: , where: e is the eccentricity, indicating the flatness of the ellipse, with a value range of [0, 1]; a is the length of the major axis; b is the length of the minor axis; represents the square root operation. The closer the eccentricity is to 1, the flatter the annual ring shape; the closer it is to 0, the rounder the shape.

[0045] (3) Complexity: Quantify it by calculating the fractal dimension of the boundary or the ratio of the boundary length to the area. The complexity index C can be expressed as: , where: C is the complexity index, indicating the complexity of the boundary; P is the boundary perimeter; A is the area of the closed region; π is the ratio of the circumference of a circle to its diameter, approximately equal to 3.14159; 4π is a normalization factor such that for a circle, C = 1. The larger the C value, the more complex the boundary.

[0046] According to the morphological characteristics and sample size, calculate the optimal number of measurement paths, ranging from 3 to 9. The present invention uses the following adaptive formula to determine the number of measurement paths N: , where: is the number of measurement paths, indicating the number of measurement paths to be set; represents the ceiling function to ensure an integer result; is the normalized value of the sample size, calculated as the sample diameter divided by the reference value of 10 mm; is the eccentricity; is the normalized value of the complexity index; and γ are weight coefficients that control the influence degree of each factor, and the empirical values are 0.3, 2.0, and 1.5 respectively. In this way, the larger the sample, the flatter the shape, and the more complex the boundary, the more measurement paths there are. For circular tree rings, a radial measurement path layout is configured. For an approximately circular tree ring with an eccentricity , the present invention adopts a radial layout with uniform angular distribution. If the number of measurement paths is , then the angle of the th path is: , where: is the angle of the th measurement path, representing the included angle between this path and the reference direction; is the initial angle, usually set to 0° (i.e., the horizontal direction); is the path index, and the value range is from 0 to is the angle of a complete circle; represents the angular interval between adjacent paths.

[0047] For elliptical tree rings, the measurement path density is increased along the long and short axis directions. For an elliptical tree ring with an eccentricity , the present invention adopts a non-uniform angular distribution and increases the path density in the long and short axis directions. The specific method is: adjust the angular distribution so that the angular density is inversely proportional to the elliptical radius to ensure that there are denser measurement paths in the region with larger curvature (usually the short axis direction).

[0048] For irregular tree rings, the measurement path distribution is dynamically adjusted according to the morphological complexity. For an irregular tree ring with an eccentricity or a complexity index , the present invention adopts an adaptive distribution strategy based on curvature. First, calculate the curvature distribution of the boundary; then, increase the measurement path density in the region with larger curvature. The specific algorithm is: divide the boundary into segments ( is much larger than , usually take ), calculate the average curvature of each segment, and then determine the positions of measurement paths based on the cumulative distribution of the curvature values to ensure that the path distribution adapts to the curvature distribution.

[0049] Evaluate the coverage uniformity of the measurement paths to ensure that the key areas are fully measured. The present invention introduces a coverage uniformity index U, and the calculation formula is: , Wherein: is the coverage uniformity index, representing the degree of uniformity of the measurement path distribution, with a value range of [0,1]; is the standard deviation of the distance between adjacent measurement paths, representing the degree of dispersion of the distance distribution; is the average distance; is the coefficient of variation, representing the relative degree of dispersion. The closer the value of is to 1, the more uniform the coverage. In practical applications, if , the measurement path distribution needs to be adjusted to improve the uniformity. In addition, the present invention also checks whether key regions (such as high-curvature regions) have sufficient measurement path coverage. If the measurement path density in a certain key region is lower than the threshold. The system will automatically add additional measurement paths.

[0050] In step S105, the intersection coordinates of the annual ring boundary and the multiple measurement paths are obtained. This step realizes the accurate intersection positioning of the annual ring boundary and the measurement paths. In a preferred embodiment of the present invention, the acquisition of the intersection coordinates adopts sub-pixel level precise positioning techniques, including gray interpolation method, gradient maximum positioning, and edge fitting method.

[0051] The gray interpolation method constructs a high-precision gray curve near the intersection, and then determines the true intersection position at a finer scale. The gradient maximum positioning improves the intersection positioning accuracy by finding the local gradient peak points. The edge fitting method performs local curve fitting on the edges near the intersection to further improve the positioning accuracy. By comprehensively using these techniques, the present invention can achieve sub-pixel level intersection positioning accuracy, significantly improving the accuracy of subsequent measurements.

[0052] In step S106, based on the intersection coordinates, the complete annual ring boundary curve is reconstructed by multi-dimensional curve fitting, specifically including: Based on the intersection coordinates, the preliminary contour of the annual ring boundary is determined. In a preferred embodiment of the present invention, the determination of the preliminary contour adopts a method combining intersection connection method and morphological feature analysis. First, the intersections on the same annual ring boundary are connected in sequence to form a preliminary contour; then, it is corrected in combination with morphological features (such as the expected shape of the annual ring) to remove obviously unreasonable parts.

[0053] According to the complexity of the annual ring boundary, the curve fitting method is adaptively selected. The present invention provides three curve fitting methods with different complexities, and automatically selects the most suitable method according to the complexity of the annual ring boundary: When the boundary is regular, piecewise polynomial fitting is adopted. For a regular boundary with a complexity index C < 1.2, piecewise polynomial fitting can not only ensure the fitting accuracy but also maintain the calculation efficiency. In the preferred embodiment of the present invention, a cubic polynomial is used for piecewise fitting, and each segment contains 3 to 5 intersection points. The first-order derivative continuity is maintained between adjacent segments to ensure the smoothness of the overall curve.

[0054] When the boundary is moderately complex, spline curve fitting is adopted. For a moderately complex boundary with a complexity index of 1.2 < C < 1.5, the spline curve can better adapt to local changes. The present invention adopts a cubic B-spline curve, and its parametric expression is: When the boundary is moderately complex, spline curve fitting is adopted. For a moderately complex boundary with a complexity index of, the spline curve can better adapt to local changes. The present invention adopts a cubic B-spline curve, and its parametric expression is: , where: is the point on the curve, representing the curve point coordinates corresponding to the parameter ; is the cubic B-spline basis function, which controls the shape of the curve; are the control points, which are determined by the intersection point coordinates; is the number of control points; and are the parameter ranges; represents the weighted sum of all control points. When the boundary is highly complex, non-parametric curve fitting is adopted. For a highly complex boundary with a complexity index of, non-parametric curve fitting can better capture the local details of the boundary. The present invention adopts a non-parametric fitting method based on radial basis function (RBF), and its expression is: , where: is the fitting function, representing the function value at the point ; are the known intersection point coordinates; is the radial basis function, usually the Gaussian function or the multiquadric function , where is the distance, and are the shape parameters; is the weight coefficient, which is determined by solving a system of linear equations; is the low-order polynomial term, usually a linear or quadratic polynomial; is the number of intersection points; represents the weighted sum of all intersection points; represents the point to the intersection point Euclidean distance

[0055] Optimize the fitting result to ensure that the boundary is smooth, continuous and conforms to the biological growth law. The present invention optimizes the fitting result by the following methods: (1) Smoothness optimization: By adding a smoothness penalty term, avoid the fluctuations caused by overfitting. The optimization objective function is: , where: is the optimization objective function, representing the overall optimization goal; is the data fitting error, representing the deviation between the fitting curve and the actual intersection point, usually using the mean square error; is the smoothness metric, usually using the square integral of the second derivative of the curve; is the smoothness coefficient, controlling the importance of smoothness, adaptively adjusted according to the data quality, usually in the range of 0.01 - 0.1.

[0056] (2) Biological rationality constraint: Based on the tree growth law, add morphological constraint conditions. For example, tree rings are usually approximately concentric circles or ellipses, and the adjacent tree rings usually maintain similar morphological characteristics, etc. During the fitting process, if there are situations that violate these laws, the system will automatically adjust the fitting parameters to ensure the biological rationality of the result.

[0057] Generate a complete mathematical model of the tree ring boundary, supporting the accurate coordinate calculation of any point. The final mathematical model of the tree ring boundary is represented parametrically, and can support the point calculation under any parameter value. For piecewise polynomials and spline curves, directly use their parametric expressions; for non - parametric curves, achieve the calculation of any point through interpolation methods. In addition, the present invention also provides a discrete representation of the boundary for subsequent analysis and visualization.

[0058] In step S107, according to the tree ring boundary curve, calculate the local inclination angle distribution, specifically including: Uniformly sample multiple points on the tree ring boundary curve. To ensure the uniformity of sampling, the present invention adopts an equal - arc - length sampling strategy instead of simple equal - parameter sampling. The specific method is: First, calculate the total arc length of the boundary curve, then determine the number of sampling points (usually taking 50 - 100), and finally uniformly sample in the arc - length parameter space to obtain a sampling point sequence. For the arc - length parameter , the arc - length parameter of the th sampling point is: , where: is the The arc length parameter of a sampling point, representing the distance along the curve from the starting point; is the total arc length of the boundary curve; is the number of sampling points; is the sampling point index, with a value range from 0 to represents the arc length interval between adjacent sampling points. Calculate the tangent direction at each sampling point to obtain the local tangent vector. For a parameterized curve , the tangent vector can be calculated through the derivative: , where: is the tangent vector, representing the tangent direction of the curve at the parameter ; and are respectively the derivatives of the curve parametric equations and with respect to the parameter , representing and the rate of change in the direction. For a discretely represented curve, the tangent direction can be approximately calculated by the difference between adjacent points.

[0059] Based on the local tangent vector, calculate the local tilt angle at each sampling point. The local tilt angle is defined as the angle between the tangent direction and the reference direction (usually the horizontal direction), and the calculation formula is: , where: is the local tilt angle, representing the angle between the tangent direction and the horizontal direction; and are respectively the and components of the tangent vector; is the arctangent function. To ensure the continuity of the angle, the present invention uses the four - quadrant arctangent function arctan2 , to obtain the angle value within the range.

[0060] Establish a local tilt angle distribution map to describe the tilt characteristics of the entire annual ring boundary. The local tilt angle distribution map is a functional relationship between the sampling position and the tilt angle, and can be used to visually display the tilt conditions of each part of the annual ring boundary. In the preferred embodiment of the present invention, a polar coordinate display method is adopted, where the angular coordinate represents the circumferential position of the sampling point, the radial coordinate represents the magnitude of the local tilt angle, and the color coding represents the rate of change of the tilt angle.

[0061] Identify the tilt angle mutation region, divide the tilt angle region, and provide a basis for the selection of the correction model. The present invention identifies the tilt angle mutation region by analyzing the change rate of the local tilt angle. The specific method is: calculate the change rate of the tilt angle between adjacent sampling points , if the change rate exceeds the threshold (usually set to 5° / mm), it is marked as a mutation region. Wherein: is the difference in tilt angle between adjacent sampling points, in degrees (°); is the physical arc length distance between adjacent sampling points, in millimeters (mm); is the change rate threshold, in degrees per millimeter (° / mm). Based on the distribution of the mutation region, the annual ring boundary is divided into multiple tilt angle regions, and the tilt angle change within each region is relatively gentle. This regional division provides an important basis for the subsequent selection of the adaptive correction model.

[0062] In step S108, based on the local tilt angle distribution, an area-adaptive correction model is used to correct the annual ring width, which specifically includes: dividing the annual ring boundary into multiple correction regions according to the local tilt angle distribution. The regional division is based on the change characteristics of the tilt angle to ensure that the tilt angle change within each region is relatively consistent. The present invention adopts an adaptive segmentation algorithm based on the change rate. The specific steps are: first, calculate the first derivative (change rate) of the tilt angle; then, identify the points where the change rate exceeds the threshold as segmentation points; finally, obtain multiple correction regions. Preferably, according to the complexity of the annual ring morphology, the number of regions is usually between 3 and 8. For regions with small tilt angle changes, a uniform tilt model is applied. For regions with a standard deviation of the tilt angle, it can be regarded as uniform tilt, and a simple cosine correction model is adopted: , Wherein: is the corrected true width, representing the actual growth width of the annual ring; is the measured tilted width, representing the width obtained by direct measurement; is the average tilt angle within the region; is the cosine correction factor, which projects the tilted measurement value onto a plane perpendicular to the growth direction.

[0063] For regions with a gradient change in tilt angle, a gradually changing tilt model is applied. For regions with a linear change trend in tilt angle and a standard deviation of the tilt angle, the present invention adopts a gradually changing tilt model: , Wherein: is the corrected true width; is the arc length parameter, representing the position along the boundary; and is the start and end positions of the region; is a function of the tilt angle with respect to the arc length, representing the tilt angle at position ; is the position ; the correction factor at the position represents the integral operation within the region range, corresponding to continuous correction. Usually, a linear model is adopted, and the parameters and are determined by linear regression, where is the intercept, representing the initial tilt angle, is the slope, representing the change rate of the tilt angle.

[0064] For the region with abrupt change in tilt angle, a piecewise tilt model is applied. For the region with drastic change in tilt angle and standard deviation , the present invention adopts a piecewise tilt model. The specific method is: the region is further divided into multiple sub-regions, and a uniform tilt model is applied to each sub-region, and then the results are combined. The division of sub-regions is based on the local extreme points of the tilt angle, ensuring that the tilt angle changes smoothly within each sub-region.

[0065] The correction parameters are calculated separately for each region to perform precise correction. According to the selected correction model, the corresponding correction parameters are calculated for each region. For the uniform tilt model, the correction parameter is the average tilt angle ; for the gradually changing tilt model, the correction parameters are the linear coefficients and ; for the piecewise tilt model, the correction parameters are the set of average tilt angles of each sub-region. These parameters are determined based on the local tilt angle distribution data through the least squares method or other optimization methods.

[0066] The true width between adjacent annual ring boundaries is calculated to generate the corrected annual ring width data. The present invention obtains the true width by calculating the shortest distance between adjacent annual ring boundaries. The specific method is: for each point on the annual ring boundary, calculate its shortest distance to the adjacent annual ring boundary, and then perform correction according to the local tilt angle. The final annual ring width is taken as the weighted average of the correction results of multiple measurement points, and the weight is related to the quality score of the measurement points. This method takes into account the complex shape of the annual ring boundary and can more accurately reflect the true growth situation.

[0067] In step S109, the corrected annual ring width data and the quality assessment report are generated, specifically including: Evaluating various sources of uncertainty in the measurement process, including uncertainties caused by image resolution, boundary recognition, tilt angle measurement, and correction model selection. The present invention adopts an uncertainty evaluation method based on the error propagation theory, specifically considering the following sources of uncertainty: Evaluate various sources of uncertainty in the measurement process, including uncertainties caused by image resolution, boundary recognition, tilt angle measurement, and calibration model selection. The present invention adopts an uncertainty evaluation method based on the error propagation theory, specifically considering the following sources of uncertainty: (1) Image resolution uncertainty : Related to the spatial resolution of the image, usually expressed as half of the pixel size. For an image with a resolution of (pixels / mm), mm, where is the number of pixels per millimeter, and 0.5 represents the uncertainty of half a pixel.

[0068] (2) Boundary recognition uncertainty : Related to the accuracy of the boundary detection algorithm and the image quality. It is determined by the present invention through multiple repeated tests and is generally 0.5 - 1.5 pixels, which is converted to the actual length of mm, where represents the pixel uncertainty range of boundary positioning, is the number of pixels per millimeter.

[0069] (3) Tilt angle measurement uncertainty : Related to the accuracy of tangent direction calculation. Through error propagation analysis, the standard uncertainty of tilt angle measurement is usually 1 - 3°, representing the angular error range of tilt angle measurement.

[0070] (4) Calibration model selection uncertainty : Related to the applicability of the calibration model. It is evaluated by the present invention by comparing the differences in calibration results of different models and is usually 3% - 8% of the calibration amount, representing the relative error range caused by model selection. Calculate the combined uncertainty of each annual ring width. Based on the error propagation theory, the formula for the combined standard uncertainty is: , where: is the combined standard uncertainty, with the unit of millimeter (mm) is the uncertainty caused by image resolution, with the unit of millimeter (mm) is the boundary recognition uncertainty, with the unit of millimeter (mm) is the measured tilt width, with the unit of millimeter (mm); is the tilt angle, with the unit of degree (°) is the radian representation of the tilt angle measurement uncertainty, obtained by converting the angular uncertainty : , with the unit of radian ( ) The relative uncertainty selected for the calibration model, which is a dimensionless quantity, usually takes values of 0.03 - 0.08 ( ) is the sine function; represents the square root operation, and the expanded uncertainty usually takes , corresponding to a confidence level of approximately 95%, where 2 is the coverage factor.

[0071] According to the magnitude of the uncertainty, the quality of the tree-ring width data is classified. The present invention establishes a quality classification standard based on the relative uncertainty (the ratio of the uncertainty to the measured value): Grade A: Relative uncertainty < 2%, high-precision data, which can be used for detailed research; Grade B: Relative uncertainty 2% - 5%, medium-high-precision data, suitable for most research needs; Grade C: Relative uncertainty 5% - 10%, medium-precision data, which can be used for trend analysis; Grade D: Relative uncertainty > 10%, low-precision data, for reference only, not recommended for quantitative research.

[0072] Generate a standard format output containing the original data, calibrated data, calibration parameters, and quality ratings. The present invention supports multiple data output formats, including: (1) Standard format for tree-ring research: such as Tucson, Belfast, etc. formats, which are convenient for integration with existing tree-ring research software; (2) General table format: such as CSV, Excel, etc., which are convenient for data sharing and further analysis; (3) Structured data format: such as JSON, XML, etc., which are convenient for programmatic processing and data exchange.

[0073] The output content includes complete information such as the original measurement data, calibrated data, calibration parameters, quality ratings, and abnormal tree-ring records, ensuring the traceability and repeatability of the data.

[0074] Provide data visualization results to visually display the measurement and calibration effects. The present invention provides rich visualization functions, including: (1) Visualization of tree-ring boundaries: Display the identified tree-ring boundaries, measurement paths, and intersection positions; (2) Comparison of calibration effects: Intuitively display the difference in tree-ring width before and after calibration; (3) Quality distribution map: Display the measurement quality of different regions through color coding; (4) Tree-ring sequence diagram: Display the calibrated tree-ring width sequence and its temporal change trend.

[0075] These visualization results help researchers intuitively understand the measurement process and results, improving research efficiency.

[0076] In a preferred embodiment of the present invention, a complete implementation process is provided: Obtain a high-resolution scanned image (1200 dpi) of a pine tree ring sample. The system automatically identifies it as a pine tree and performs multi-band adaptive enhancement processing using preprocessing parameters specific to conifers. In the enhanced image, the boundaries of the tree rings are clearer, and the contrast between early and late wood is increased by about 40%.

[0077] Based on the enhanced image, the system uses a multi-strategy boundary detection method to identify the boundaries of the tree rings. Through the result fusion of three methods: gradient detection, texture segmentation, and periodic analysis, the boundaries of 38 complete tree rings are successfully identified. The system automatically detects and processes 2 false rings and 1 traumatic ring, and the average boundary quality score is 85 points (out of 100).

[0078] According to the morphological analysis of the tree ring boundaries, the system determines 5 radial measurement path configurations. Since the sample shows a slight ellipse (eccentricity e = 0.35), the system appropriately increases the measurement path density in the long and short axis directions to ensure uniform coverage of the measurement (uniformity index U = 0.86).

[0079] The system accurately obtains the intersection coordinates of the tree ring boundaries and the measurement paths, and uses sub-pixel positioning technology with a positioning accuracy of 0.2 pixels (about 0.004 mm).

[0080] Based on the obtained intersection coordinates, the system reconstructs the complete tree ring boundaries by spline curve fitting. Since the complexity of the tree ring boundaries is medium (C = 1.3), the system automatically selects a cubic B-spline curve for fitting, and the fitting error is less than 0.5 pixels.

[0081] According to the reconstructed tree ring boundary curve, the system calculates the local inclination angle distribution. By uniformly sampling 60 points on each tree ring boundary, the calculated local inclination angle distribution is between 12° and 38°, and the average inclination angle is 25°.

[0082] Based on the local inclination angle distribution, the system divides the tree ring boundaries into 4 correction regions and applies different correction models: the first region (0° - 90°) applies a uniform inclination model; the second region (90° - 180°) and the third region (180° - 270°) apply a gradient inclination model; the fourth region (270° - 360°) applies a uniform inclination model. Compared with the uncorrected width, the corrected tree ring width is reduced by about 9% on average.

[0083] The system generates calibrated tree-ring width data and a quality assessment report. Comprehensive uncertainty analysis shows that the quality of the calibrated tree-ring width data is mainly at level B (85%) and level C (15%), and there is no data at level D. The quality assessment report details the contributions of various uncertainty sources, providing an important reference for the reliability of the data for researchers.

[0084] Another embodiment of the present invention is directed to the treatment of complex abnormal tree rings: Input an oak sample image (1600 dpi) containing multiple false rings and traumatic rings. The system automatically identifies it as a broad-leaved tree and applies multi-band adaptive enhancement processing using specific parameters for broad-leaved trees. After enhancement, even the blurred tree-ring boundaries become relatively clear and distinguishable.

[0085] The system successfully identifies the main tree-ring boundaries through a multi-strategy boundary detection method. Through morphological feature analysis, the system identifies 3 false rings and 1 traumatic ring. The characteristics of the false rings are narrow width (only 25% of the normal tree-ring width) and low circumferential continuity (about 65%); the traumatic ring shows local high curvature (3.5 times the average curvature).

[0086] For such complex samples, the system automatically plans 7 measurement paths, especially increasing the measurement density near the traumatic area and adjusting the path direction to avoid severely distorted parts.

[0087] The system marks the false rings and excludes them from the measurement sequence, and adopts a local adjustment strategy for the traumatic rings, using special calibration parameters in the affected area.

[0088] Due to the complex tree-ring morphology (complexity index C = 1.7), the system automatically selects a non-parametric curve fitting method to reconstruct the tree-ring boundaries and ensures that the fitting results conform to the laws of tree growth through biological rationality constraints.

[0089] The tilt angle analysis shows that the tilt angle of this sample has a wide distribution range (8° - 45°) and there are obvious mutation areas. Based on this, the system divides the tree-ring boundaries into 6 calibration regions and applies different calibration models respectively.

[0090] The final generated quality assessment report shows that the main tree-ring data reaches level B and the local traumatic area is level C. The system also provides a detailed record of abnormal processing to ensure the traceability of the data.

[0091] The above embodiments clearly demonstrate the application effects of the present invention in different situations. Compared with the prior art, the present invention not only significantly improves the measurement accuracy (an increase of 40% - 60%), but also significantly enhances the degree of automation, reduces manual intervention, and can effectively handle complex and abnormal situations, providing a more reliable data basis for dendrochronology research.

[0092] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, various modifications and variations can be made to the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for correcting the measurement error of tree ring width, characterized in that, Including: Obtain an image of the tree-ring sample; Perform multi-band adaptive enhancement processing on the tree-ring sample image to obtain an enhanced tree-ring image; Based on the enhanced tree-ring image, use a multi-strategy boundary detection method to identify the tree-ring boundary; According to the tree-ring boundary, determine a dynamic measurement path configuration, where the dynamic measurement path configuration includes multiple measurement paths; Obtain the intersection coordinates of the tree-ring boundary and the multiple measurement paths; Based on the intersection coordinates, reconstruct a complete tree-ring boundary curve through multi-dimensional curve fitting; According to the tree-ring boundary curve, calculate the local inclination angle distribution; Based on the local inclination angle distribution, use a regionalized adaptive correction model to correct the tree-ring width; Generate corrected tree-ring width data and a quality assessment report.

2. The method according to claim 1, wherein The multi-band adaptive enhancement processing of the tree-ring sample image specifically includes: Decompose the tree-ring sample image into three frequency bands: high frequency, medium frequency, and low frequency; Perform detail enhancement and noise suppression processing on the high-frequency band; Perform edge sharpening and texture enhancement processing on the medium-frequency band; Perform contrast optimization and illumination equalization processing on the low-frequency band; Based on tree species characteristic parameters, adaptively adjust the enhancement parameters of each frequency band; Fuse the processed three frequency bands to obtain the enhanced tree-ring image.

3. The method according to claim 1, characterized in that The multi-strategy boundary detection method for identifying the tree-ring boundary specifically includes: Perform edge detection based on gradient change to obtain a first boundary detection result; Perform region segmentation based on texture features to obtain a second boundary detection result; Perform periodic detection based on morphological features to obtain a third boundary detection result; Perform weighted fusion on the first boundary detection result, the second boundary detection result, and the third boundary detection result to obtain a fused boundary; Perform continuity optimization and anomaly detection on the fused boundary to obtain the tree-ring boundary.

4. The method according to claim 3, characterized in that, The continuity optimization and anomaly detection of the fused boundary specifically include: Detect breakpoints and isolated points in the fused boundary; Based on boundary continuity constraints, repair the breakpoints and remove the isolated points; Identify abnormal tree-rings in the fused boundary, including false rings, missing rings, and traumatic rings; According to the tree-ring growth rule, mark or correct the abnormal tree-rings; Generate a boundary quality score to identify areas that require manual intervention.

5. The method according to claim 1, characterized in that The determination of the dynamic measurement path configuration specifically includes: Analyze the morphological features of the tree-ring boundary, including overall shape, eccentricity, and complexity; According to the morphological features and sample size, calculate the optimal number of measurement paths, ranging from 3 to 9; For circular tree-rings, configure a radial measurement path layout; For elliptical tree-rings, increase the measurement path density along the major and minor axes; For irregular tree-rings, dynamically adjust the measurement path distribution according to the morphological complexity; Evaluate the coverage uniformity of the measurement paths to ensure that key areas are fully measured.

6. The method according to claim 1, wherein The reconstruction of the complete tree-ring boundary curve through multi-dimensional curve fitting specifically includes: Based on the intersection coordinates, determine a preliminary contour of the tree-ring boundary; According to the complexity of the tree-ring boundary, adaptively select a curve fitting method, including: When the boundary is relatively regular, use piecewise polynomial fitting; When the boundary is moderately complex, spline curve fitting is adopted; When the boundary is highly complex, non-parametric curve fitting is adopted; Optimize the fitting results to ensure that the boundary is smooth, continuous and conforms to the biological growth law; Generate a complete mathematical model of the annual ring boundary to support the accurate coordinate calculation of any point.

7. The method according to claim 1, characterized in that, The calculation of the local inclination angle distribution specifically includes: Uniformly sample multiple points on the annual ring boundary curve; Calculate the tangent direction at each sampling point to obtain the local tangent vector; Based on the local tangent vector, calculate the local inclination angle of each sampling point; Establish a local inclination angle distribution map to describe the inclination characteristics of the entire annual ring boundary; Identify the regions with abrupt changes in inclination angles, divide the inclination angle regions, and provide a basis for the selection of correction models.

8. The method according to claim 1, wherein The correction of the annual ring width using the regionalized adaptive correction model specifically includes: According to the local inclination angle distribution, divide the annual ring boundary into multiple correction regions; For the regions with small changes in inclination angles, apply the uniform inclination model; For the regions with gradient changes in inclination angles, apply the gradual inclination model; For the regions with abrupt changes in inclination angles, apply the segmented inclination model; Calculate the correction parameters for each region separately and perform precise correction; Calculate the true width between adjacent annual ring boundaries and generate the corrected annual ring width data.

9. The method according to claim 1, characterized in that, The generation of the corrected annual ring width data and the quality assessment report specifically includes: Evaluate various sources of uncertainty in the measurement process, including the uncertainties caused by image resolution, boundary recognition, inclination angle measurement and correction model selection; Calculate the combined uncertainty of each annual ring width; Grade the quality of the annual ring width data according to the magnitude of the uncertainty; Generate a standard format output containing the original data, corrected data, correction parameters and quality ratings; Provide the data visualization results to visually display the measurement and correction effects.

10. The method according to claim 1, wherein Before the multi-band adaptive enhancement processing, it also includes: Obtain the tree species characteristic information; According to the tree species characteristic information, extract the corresponding image processing parameters from the pre-established tree species characteristic database; The multi-band adaptive enhancement processing is configured with the image processing parameters; Among them, the tree species characteristic information includes tree species type, annual ring density, early and late wood contrast and boundary clarity.

Images