Rapid evaluation method for wood property of cedarwood standing wood based on near infrared spectrum

Abstract

The invention relates to the technical field of rapid evaluation methods for the standing wood property of cunninghamia lanceolata, in particular to a rapid evaluation method for the standing wood property of the cunninghamia lanceolata based on a near infrared spectrum, and the method comprises the following steps: 1, dividing a trunk of a to-be-evaluated standing wood of the cunninghamia lanceolata into a plurality of sections; 2, taking the near infrared spectrum information at the tree segment as input, and taking the MVF or MVS at the tree segment as output; and 3, carrying out spectrum acquisition of multiple frequency bands on multiple directions of the optimal tree segment through a portable near-infrared spectrometer, processing the spectrum of each direction, establishing a correction model, and obtaining a measured and predicted MVS or MVF value in each direction. The method provided by the invention realizes real nondestructive evaluation. By using the portable near-infrared spectrometer, people can obtain abundant wood property information under the condition of not causing any physical damage to trees. Due to the characteristic, the method is particularly suitable for protective research and long-term monitoring projects of rare tree species.

Description

Technical Field

[0001] The invention relates to the technical field of methods for rapid evaluation of properties of standing Chinese fir timber, and in particular to a method for rapid evaluation of properties of standing Chinese fir timber based on near infrared spectroscopy. Background Art

[0002] With the rapid development of the forestry industry and the popularization of the concept of sustainable development, the demand for accurate assessment of wood quality is becoming increasingly urgent. As an important artificial afforestation tree species in southern my country, the accurate assessment of the standing timber properties of Chinese fir is of great significance for forest breeding, forest stand management and wood processing. However, traditional standing timber property assessment methods face many challenges and are difficult to meet the needs of modern forestry development.

[0003] For a long time, standing timber property assessment has mainly relied on destructive sampling methods, such as growth cone coring or felling analysis. Although these methods can provide relatively accurate data, they have significant limitations. First, these methods will inevitably cause damage to trees and increase the risk of invasion by pests and diseases, especially for rare tree species or plots that require long-term monitoring. This impact is more serious. Secondly, traditional methods are time-consuming and labor-intensive, making it difficult to conduct large-scale rapid assessments. In today's information age, this inefficient assessment method can no longer meet the needs of modern forestry management.

[0004] In addition, traditional methods can only obtain sample data from a single point or a limited location, which makes it difficult to fully reflect the wood properties of the entire tree. The growth of trees is a complex process, and their wood properties may vary significantly at different heights and orientations. It is difficult to accurately grasp the overall characteristics by relying solely on local sampling, and this one-sidedness may lead to a large deviation between the evaluation results and the actual situation.

[0005] In recent years, some non-destructive assessment technologies, such as sonication and X-ray tomography, have improved the non-destructive nature of assessment to a certain extent. However, these methods are either inaccurate or the equipment is expensive and bulky, making them difficult to apply on a large scale in the field. Therefore, the forestry industry has been seeking a standing timber property assessment method that can balance accuracy, efficiency and non-destructiveness. Summary of the invention

[0006] The present invention aims to solve the problems existing in the existing standing timber property assessment methods and proposes a method for rapid assessment of standing timber properties of Chinese fir based on near-infrared spectroscopy. The method achieves rapid, accurate and non-destructive assessment of standing timber properties of Chinese fir by comprehensively utilizing near-infrared spectroscopy technology, multi-point sampling strategy and advanced mathematical modeling.

[0007] The present invention provides a method for rapid evaluation of standing timber properties of Chinese fir based on near infrared spectroscopy, the method comprising the following steps:

[0008] Step 1: Divide the trunk of the standing tree to be evaluated into several sections, collect at least four near-infrared spectral sample points on each section, obtain the spectral curve at each tree section, and determine the most suitable tree section for collection; use near-infrared spectroscopy technology to obtain the spectral characteristics of each tree section of the standing tree to be evaluated, and use Xcalibur software to establish a corresponding calibration model for each tree section; use the density MVS (mg / cm 3 ) as the dependent variable and the number of spectra as the independent variable, a density quantitative model was constructed at each tree segment; at the breast height of the standing tree, the compressed wood fiber length MVF (mg / cm 3 ) is the dependent variable, the spectrum number is the independent variable, and a quantitative model of compressed wood fiber length is established; for the determined evaluation component, the measurement object is determined to be the content of the component, and the quantitative model of the evaluation component is determined and established with the corresponding database;

[0009] Step 2: Using the near-infrared spectrum information at the tree segment as input and the MVF or MVS at the tree segment as output, establish an objective function containing multiple evaluation component contents; solve the objective function at each tree segment through the optimization algorithm, and optimize to obtain the optimal tree segment objective function result with the least near-infrared spectrum; select the tree segment combination in the optimization algorithm as the standing tree segment combination, use it for tree standing test, and finally determine the optimal standing tree segment of the Chinese fir to be evaluated;

[0010] Step three: Use a portable near-infrared spectrometer to collect spectra in multiple frequency bands at multiple orientations of the optimal tree segment, process the spectrum at each orientation, establish a correction model, and obtain the measured predicted MVS or MVF value at each orientation; perform weighted combination of the measured predicted MVS or MVF values ​​at multiple orientations to obtain the predicted MVS or MVF value at breast height of the standing tree; calculate the content of the evaluation components at breast height based on the quantitative model of the evaluation components in the database.

[0011] Preferably, in step one, after determining the evaluation components, repeatedly obtain the density MVS and compressed wood fiber length MVF values ​​at several tree segments, use the near-infrared spectrum sample points at the tree segments as the input training set, establish a quantitative analysis model, incorporate the tree segment structure, tree segment diameter, and environmental factors as prediction factors into the calibration model, and determine the final calibration model; finally, bring the unknown samples into the final calibration model to calculate the MVS or Mv F values ​​at the tree segments.

[0012] Preferably, the method for measuring density and compressed wood fiber length specifically includes the following steps: according to the volume of the tree segment determined by near-infrared spectroscopy, compressing the wood in the tree segment under preset normal pressure and high temperature conditions to obtain compressed wood, slicing and drying the compressed wood; taking the compressed wood at the tree segment as the sampling object, measuring the density by weight method; taking the compressed wood at the tree segment as the sampling object, immersing it in a sodium chlorite solution for treatment, taking it out and washing it until there is no red color, and measuring the total weight of compressed wood fibers in the compressed wood by weight method; using an optical microscope to prepare slices for the sampling object, setting a corresponding magnification to obtain compressed wood slices; taking slice images, and calculating the compressed wood fiber length according to the image area; constructing a quantitative model for compressed wood fiber length includes the following steps: using Image-Pro Plus software to process the slice images of the compressed wood slices after preparation, and obtaining the fiber area of ​​the compressed wood and the outline of the outer edge of the fiber; according to the corresponding relationship between each pixel in the slice and the actual area and the fiber area, obtaining the actual fiber area, and calculating the logarithm of the number of fibers, using the calculation formula of the logarithmic fiber length, obtaining the value of the logarithmic fiber length, and then establishing a quantitative model for the compressed wood fiber length of wood composed of different components of standing trees.

[0013] Preferably, the calculation formula of the logarithmic fiber length is:

[0014] L=log(S / S0) / q0

[0015] Wherein, L is the logarithmic fiber length, S is the fiber area, S0 and q0 are constants; the formula of the logarithmic fiber area is: S=A×P, wherein S is the fiber area, A is the actual area corresponding to each pixel, and P is the number of pixels occupied by the fiber in the slice image, representing the actual area of ​​the fiber slice.

[0016] Preferably, in step one, the specific process of constructing the MVS or MVF quantitative model at each tree segment is: according to the wood centroid profile at the breast height of the standing tree, the tree segment at the wood centroid is equally spaced, so that the spectral information collected at the tree segment corresponds one to one with the density MVS or compressed wood fiber length MVF data, and the processed spectrum is used as the independent variable, and the MVS or MVF at the tree segment is used as the dependent variable to construct a quantitative analysis model to obtain the density MVS (or compressed wood fiber length MVF) quantitative model at the tree segment.

[0017] Preferably, in step 2, the objective function comprising multiple evaluation component contents is:

[0018]

[0019] Where n is the number of tree segments, M i is the predicted value of the corrected MVS (or corrected MVF) at the tree segment, T iis the MVS (or MVF) value measured at the tree segment, w i is the tree segment weight, Cov(M i M j ) is the cross-covariance value of MNS (or MVF) between tree segments, Cov(M i S i ) is the covariance between MVS (or MVF) and position at the tree segment, var(S i ) is the variance of MVS (or MVF) at the tree segment, and λ1, λ2 and λ3 are balance coefficients.

[0020] Preferably, in step 2, an optimization algorithm can be implemented to obtain the optimal tree segment objective function result with the least near infrared spectrum, and the selection of the tree segment combination is completed by solving the objective function of the optimization algorithm, and the optimal tree segment combination is determined; the tree segment combination is adjusted by solving the coefficients in the objective function, and the objective function coefficient solution formula is:

[0021]

[0022] Among them, a, b, c and d are the coefficients to be solved, f is the objective function; xij represents the eigenvalue of the jth tree segment of the ith sample, i represents the number of samples, and j represents the type of tree segment; di is the tree segment distance of the ith sample, yi is the breast diameter of the ith sample, m is the total number of samples, and n is the total number of tree segments.

[0023] Preferably, the rapid evaluation process of the component content includes: collecting spectra of multiple frequency bands at multiple positions of the optimal tree segment: fusing the spectral information of different frequency bands through the correction model obtained at each tree segment, and calculating the average value of the spectra of multiple frequency bands according to the weighted formula; using the fused spectral information to replace the original spectral information, measuring the model, and obtaining the MVF (or MVS) prediction value at the tree segment, repeating the above operation, and obtaining the MVS (or MVF) prediction value at different positions; obtaining the MVS or MVF value at the breast height of the standing tree according to the density quantitative model at each tree segment; determining the evaluation component quantitative model according to the density quantitative model, and substituting the MVS (or MVF) prediction value into In the quantitative model of evaluation components, the content result of the target component is obtained, and the estimation of the evaluation components of Chinese fir is completed; the fused spectral information replaces the original spectral information specifically including, for the preferred combination of tree segments containing the weight term of the distance between the tree segment and the breast height in the objective function, when performing standing tree structure analysis, calculating the entire standing tree segment, and optimizing by optimizing the distance between the tree segment and the breast height; when fusing information of multiple frequency band spectra, using the fused spectral information to replace the original spectral information, specifically including replacing the original spectrum by calculating the average value of multiple frequency band spectra; determining the optimal frequency band through the combination of tree segments, and for the spectrum at the distance between the shortest spectrum and the breast height, fusing the spectrum of the shortest tree segment with the average value of multiple spectrum segments.

[0024] Preferably, after the component to be evaluated is determined, a quantitative model of the component is obtained through a density quantitative model to obtain an estimation result of the component to be evaluated.

[0025] Preferably, the method of obtaining the quantitative model of the component to be evaluated through the density quantitative model means that after obtaining the measured MVS or MVF value of the determined component, the component is associated with the MVS (or MVF), and the density or compressed wood fiber length quantitative model is selected to establish a quantitative model of the density or compressed wood fiber length and the component to be evaluated; if there is no relationship between the component to be evaluated and the density or compressed wood fiber length, the quantitative model of the component to be evaluated is directly obtained; after the objective function result containing the contents of multiple evaluation components is optimized by calculating the tree segment through the near-infrared spectrum measurement point count, the near-infrared spectrum acquisition method of the selected acquisition frequency band to determine the evaluation component is: information fusion of multiple frequency band spectra, calculating the average value of multiple frequency band spectra instead of original data for modeling; using the fused spectral information as input, after the objective function is known, determining the optimal tree segment, and establishing a quantitative model of density and the component to be evaluated through the density quantitative model; selecting density as output, bringing in wavelength, acquiring tree segment and spectral information, outputting density, and obtaining the estimated result of the component.

[0026] Advantages and beneficial effects of the present invention:

[0027] The core of the present invention is to establish a complete evaluation process, including reasonable tree segmentation, multi-point spectral acquisition, optimal tree segment selection, spectral data fusion and the construction of a material property index prediction model. This method not only overcomes the limitations of traditional methods, but also shows significant advantages in many aspects.

[0028] First, the method of the present invention realizes a truly non-destructive assessment. By using a portable near-infrared spectrometer, we can obtain rich wood property information without causing any physical damage to the tree. This feature makes the method particularly suitable for conservation research and long-term monitoring projects of rare tree species.

[0029] Secondly, this method greatly improves the evaluation efficiency. Compared with the traditional method that takes several hours to evaluate a single tree, our method only takes about 15 minutes to complete a comprehensive evaluation of a tree. This high efficiency makes large-scale forest resource surveys possible and provides more timely and comprehensive data support for forestry management decisions.

[0030] Furthermore, the method of the present invention provides more comprehensive and accurate wood property evaluation results through multi-point sampling and data fusion technology. We not only consider the characteristics of trees at different heights, but also take into account the differences in different directions, so as to obtain an evaluation result that is more representative of the characteristics of the entire tree. Experimental data show that this method is comparable to the traditional coring method in terms of accuracy, and even better in terms of stability and repeatability.

[0031] In addition, this method has good adaptability and scalability. By adjusting the model parameters, we can quickly adapt to different tree species or different growing environments of Chinese fir, and even have the potential to be expanded to the assessment of other tree species. This flexibility opens the way for this method to be applied in a wider range of forestry.

[0032] Finally, it is worth mentioning that the method of the present invention provides new possibilities for forestry research. Since measurements can be made frequently and non-destructively, we can better study the dynamic changes in wood properties during tree growth, providing important support for the optimization of forest growth models and precise forestry management.

[0033] In general, the method for rapid evaluation of standing timber properties of Chinese fir based on near-infrared spectroscopy proposed in the present invention not only solves the problems of low efficiency, high destructiveness, and strong locality of traditional methods, but also shows significant advantages in terms of accuracy, comprehensiveness, and adaptability. The application of this method is expected to promote the innovation of forestry evaluation technology, provide strong support for the sustainable management and efficient utilization of forestry resources, and has important practical significance and long-term value for promoting the modernization of my country's forestry industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 The present invention is a flow chart of the method.

[0035] Figure 2 The present invention is a flow chart of dividing the trunks of standing Chinese fir trees.

[0036] Figure 3 It is a logic block diagram of near infrared spectrum acquisition and data preprocessing of the present invention.

[0037] Figure 4 A logical block diagram for the optimal tree segment selection of the present invention.

[0038] Figure 5 Logic block diagram established for the density, fiber length and evaluation component quantitative model of the present invention.

[0039] Figure 6 Rapid evaluation prediction for the present invention. DETAILED DESCRIPTION

[0040] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects thereof are described in detail below in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0041] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0042] See also Figure 1-6 The present invention relates to a method for rapid evaluation of standing timber properties of Chinese fir based on near infrared spectroscopy. The method aims to improve the efficiency and accuracy of standing timber property evaluation of Chinese fir and provide an important reference for forestry production and management.

[0043] First of all, the proposed method includes three main steps. The first step is data collection and model building, the second step is to optimize tree segment selection, and the third step is to perform rapid evaluation. Let's understand the specific content of each step in detail.

[0044] In the first step, we divide the trunks of the standing Chinese fir trees to be evaluated into several sections. Usually, we divide the trunks into 5-10 equal sections, so that we can fully consider the differences in wood properties of trees at different heights. On each tree section, we collect at least four near-infrared spectral sample points. For example, we can collect a sample point in each of the four directions of the east, south, west, and north of the trunk. The purpose of this is to fully reflect the characteristics of the tree section, because the wood properties of trees may be different in different directions.

[0045] Next, we used near-infrared spectroscopy to obtain the spectral characteristics of each tree segment. Near-infrared spectroscopy was chosen because it is fast, non-destructive, and portable, making it very suitable for field operations. We used Xcalibur software to establish the corresponding calibration model. Xcalibur software is a professional spectral analysis software that can help us process complex spectral data and establish accurate quantitative models.

[0046] In each tree segment, we establish two important quantitative models: density model and compressed wood fiber length model. The density model is based on the density MVS (mg / cm 3 ) as the dependent variable, and the compressed wood fiber length model uses the compressed wood fiber length MVF (mg / cm 3 ) are the dependent variables. These two indicators are important because they directly reflect the quality and properties of wood. For example, denser wood is generally stronger, while fiber length affects wood processing properties and pulp quality.

[0047] For the determined evaluation components, we determine the measurement object as the content of the component and establish the corresponding quantitative model and database. For example, if we are concerned about the cellulose content of wood, we will establish a quantitative model for the cellulose content.

[0048] In the second step, we use the near infrared spectral information at the tree segment as input and the MVF or MVS at the tree segment as output to establish an objective function containing multiple evaluation component contents. The purpose of this objective function is to help us find the optimal tree segment combination so that we can obtain the most accurate evaluation results with the least near infrared spectral measurements.

[0049] By solving the objective function at each tree segment with an optimization algorithm, we can obtain a solution that can achieve the optimal tree segment objective function result with the least near-infrared spectrum. This process is like solving a complex puzzle. We need to find several key tree segments that best represent the characteristics of the entire tree. For example, we may find that the tree segments at the diameter at breast height, the middle of the trunk, and the lower part of the crown are the most representative.

[0050] Then, we selected the tree segment combination in the optimization algorithm as the standing tree segment combination, used it in the tree standing test, and finally determined the optimal standing tree segment of the Chinese fir to be evaluated. The significance of this step is that we can greatly reduce the number of tree segments that need to be measured in actual operation, thereby improving work efficiency.

[0051] In the third step, we use a portable near-infrared spectrometer to collect spectra in multiple frequency bands at multiple locations of the optimal tree segment. For example, we may collect spectra in three different frequency bands at four locations (east, south, west, and north) of each optimal tree segment. The purpose of this is to obtain more comprehensive and accurate spectral information.

[0052] Next, we process the spectrum of each direction and build a correction model. This process is like putting a pair of "glasses" on the spectral data, allowing it to "see" the characteristics of wood more accurately. Through this model, we can get the measured predicted MVS or MVF value in each direction.

[0053] We then weight the predicted MVS or MVF values ​​from the measurements in multiple directions to get the predicted MVS or MVF value at breast height of the tree. It's like we are taking into account the characteristics of the tree in different directions to get an overall evaluation result. For example, we may give a higher weight to the measurement results of the direct sunlight side because the growth in this direction is usually more representative.

[0054] Finally, according to the quantitative model of the evaluation components in the database, we calculated the content of the evaluation components at breast height. In this way, we obtained the material property evaluation results of the standing Chinese fir trees.

[0055] Preferably, after determining the evaluation components, we will repeatedly obtain the density MVS and compressed wood fiber length MVF values ​​at several tree segments. The purpose of this is to increase the reliability and representativeness of the data. We use the near-infrared spectral sample points at the tree segment as the input training set to establish a quantitative analysis model. In this process, we not only consider the spectral information, but also incorporate the tree segment structure, tree segment diameter, environmental factors, etc. as predictive factors into the calibration model.

[0056] For example, we might consider the diameter of the tree segment, because different diameters may have different density and fiber length in the wood. We might also consider environmental factors, such as altitude and average annual temperature, because these factors may affect the growth of trees and, in turn, the properties of the wood. By including these additional predictors, we can improve the accuracy and applicability of the model.

[0057] Finally, we bring the unknown samples into the final calibration model to calculate the MVS or MVF value at the tree segment. This step is like we use the trained model to "predict" the characteristics of new trees.

[0058] In one embodiment of the present invention, we describe in detail a method for measuring density and compressed wood fiber length. First, we compress the wood in the tree segment under preset normal pressure and high temperature conditions based on the volume of the tree segment measured by near infrared spectroscopy. Usually, we set the temperature between 120℃-150℃, which can effectively soften the wood and facilitate compression. The compressed wood is sliced ​​and dried.

[0059] We measure density gravimetrically, which is a simple and accurate method. Specifically, we measure the mass and volume of the sample, and then divide the mass by the volume to get the density. For the measurement of compressed wood fiber length, we soak the sample in a sodium chlorite solution. The concentration of sodium chlorite solution is usually between 1% and 3%, which can effectively remove lignin and facilitate the observation of fibers.

[0060] The processed samples are sliced ​​and we observe and take pictures of the slices using an optical microscope, and then calculate the length of the compressed wood fibers based on the image area. Usually, we use a magnification of 400x or 600x, so that the morphology of individual fibers can be clearly observed.

[0061] In another embodiment, we describe in detail the construction process of the quantitative model of compressed wood fiber length. First, we use Image-Pro Plus software to process the compressed wood slice image after filming. This software can accurately measure the fiber area and contour. Through software processing, we can obtain the fiber area of ​​the compressed wood and the contour of the outer edge of the fiber.

[0062] We then use the correspondence between each pixel in the slice and the actual area and the fiber area to get the actual fiber area. For example, if we are using a microscope with 400x magnification, each pixel may correspond to 0.25 μm 2 If a fiber occupies 1000 pixels in the image, its actual area is 250 μm 2 .

[0063] Next, we calculated the logarithm of the number of fibers and used the logarithmic fiber length calculation formula to obtain the value of the logarithmic fiber length. The significance of this step is that fiber length is usually log-normally distributed, and using logarithmic values ​​can make the data closer to normal distribution, which is convenient for subsequent statistical analysis.

[0064] Finally, we established a quantitative model for the length of compressed wood fibers in different components of standing trees. This model can help us predict the fiber length of wood based on near-infrared spectroscopy data.

[0065] In another embodiment of the present invention, we describe in detail the calculation formula of logarithmic fiber length. The formula is:

[0066] L = log(S) / S0) / q0

[0067] Where L is the logarithmic fiber length, S is the fiber area, and S0 and q0 are constants. Typically, we set S0 to 1 μm. 2 , while q0 varies according to the wood species. For fir, q0 is usually between 0.3-0.5.

[0068] At the same time, we also give the calculation formula of logarithmic fiber area:

[0069] S=A×P

[0070] Among them, S is the fiber area, A is the actual area corresponding to each pixel, and P is the number of pixels occupied by the fiber in the slice image. This formula helps us convert the number of pixels in the image into the actual fiber area.

[0071] When constructing the MVS or MVF quantitative model at each tree segment, we first divide the tree segment at equal distances at the wood centroid according to the wood centroid profile at the breast height of the tree. This ensures that our sampling points can represent the characteristics of the entire tree segment. For example, we may take a sample point at each of the three positions of the inner, middle, and outer positions of the tree segment.

[0072] Then, we make the spectral information collected at the tree segment correspond to the density MVS or compressed wood fiber length MVF data. The purpose of this step is to establish the relationship between spectral data and wood property indicators.

[0073] Next, we use the processed spectrum as the independent variable and the MVS or MVF at the tree segment as the dependent variable to build a quantitative analysis model. This process is like teaching a computer how to "read" spectral data and infer the density or fiber length of wood from it.

[0074] Finally, we obtain a quantitative model of density MVS (or compressed wood fiber length MVF) at the tree section. This model is the "tool" we use to predict the properties of unknown samples.

[0075] In another embodiment of the present invention, we describe in detail the process of constructing the objective function. This objective function contains multiple evaluation component contents, and its form is:

[0076]

[0077] Among them, n is the number of tree segments, usually we will choose 5-10 tree segments, M i is the predicted value of the corrected MVS (or corrected MVF) at the tree segment, T i is the MVS (or MVF) value measured at the tree segment, w i is the tree segment weight, Cov(M i M j ) represents the cross-covariance of MVS (or MVF) between different tree segments, which helps us consider the correlation between tree segments. Cov(M i S i ) represents the covariance between MVS (or MVF) and the position of the tree segment, which takes into account the relationship between wood properties and the position of the tree segment. i ) is the variance of MVS (or MVF) at the tree segment, reflecting the degree of variation of wood properties within the tree segment. λ1, λ2, and λ3 are balance coefficients used to adjust the importance of each item. Usually, we use cross-validation to determine the optimal values ​​of these coefficients. We may give a higher weight to the tree segment at the DBH because this position is usually more representative of the characteristics of the entire tree.

[0078] In another embodiment of the present invention, we describe in detail how to optimize the tree segment selection. We use the following formula to adjust the tree segment combination:

[0079]

[0080] Among them, a, b, c and d are the coefficients to be solved, f is the objective function; x i j represents the characteristic value of the jth tree segment of the i-th sample, i represents the number of samples, and j represents the type of tree segment; d i is the tree segment distance of the i-th sample, y i is the DBH of the ith sample. m is the total number of samples, usually 100-200 trees are selected as samples. n is the total number of tree segments, as mentioned above, usually 5-10.

[0081] By solving this equation, we can find the optimal combination of tree segments. For example, we may find that for Chinese fir, selecting tree segments at DBH, 1 / 3 tree height, and 2 / 3 tree height can give the best evaluation results.

[0082] When conducting a rapid assessment of the content of the evaluation components, we first collect spectra in multiple frequency bands from multiple directions of the optimal tree segment. Usually, we will select 4-6 directions (such as east, south, west, north, up and down), and collect spectra in 3-5 frequency bands in each direction. This can fully reflect the characteristics of the tree. We fuse this spectral information, usually using a weighted average method. For example, we may give a higher weight to the spectrum on the south side of the trunk because the south side receives more light and may better reflect the growth status of the tree.

[0083] Then, we use the correction model obtained at each tree segment to fuse the spectral information of different frequency bands. This process is like translating information in different "languages" into the same "language", so that we can better understand and use this information. We calculate the average value of multiple frequency band spectra according to the weighted formula, which can reduce the error that may be caused by a single frequency band.

[0084] Next, we use the fused spectral information to replace the original spectral information and measure the model. This can improve the stability and accuracy of the model. Through this process, we get the MVF (or MVS) prediction value at the tree segment. We will repeat this operation to obtain the MVS (or MVF) prediction value at different orientations.

[0085] Based on the density quantitative model at each tree segment, we can get the MVS or MVF value at the breast height of the standing tree. This step is like we are using local information to infer the overall characteristics. For example, if we find that the density at breast height is higher, we may infer that the overall material of the tree is better.

[0086] Finally, we determined the evaluation component quantitative model based on the density quantitative model, substituted the MVS (or MVF) predicted value into the evaluation component quantitative model, and obtained the content result of the target component. In this way, we completed the estimation of the evaluation components of Chinese fir.

[0087] In another embodiment of the present invention, we describe in detail how to fuse and replace spectral information. For the preferred combination of tree segments that includes a weight term for the distance between the tree segment and the breast height in the objective function, we will calculate the entire standing tree segment when performing standing tree structure analysis. For example, we may divide the entire tree into 10 segments of equal length, and then calculate the distance between each segment and the breast height.

[0088] By optimizing the distance between the tree segment and the chest height, we can find several key tree segments that best represent the characteristics of the whole tree. This process is like we are looking for the "key points" of the tree, which can best reflect the characteristics of the whole tree.

[0089] When multiple frequency band spectra are fused, we use the fused spectrum information to replace the original spectrum information. Specifically, we use the method of calculating the average value of multiple frequency band spectra to replace the original spectrum. This can reduce the error that may be caused by a single frequency band and improve the stability of the model.

[0090] By combining tree segments, we can determine the optimal frequency band. For the spectrum at the shortest spectrum and chest height, we average the spectrum of the shortest tree segment to obtain more representative spectral information.

[0091] In another embodiment of the present invention, we explain how to determine the quantitative model of the class component to be evaluated. First, we obtain the quantitative model of the class component through the density quantitative model. This process is like we are building a bridge to connect the near infrared spectroscopy data and the wood properties we are interested in.

[0092] Specifically, we first obtain the measured MVS or MVF value of a certain class component. Then, we correlate that component with the MVS (or MVF). For example, if we find that wood density has a strong correlation with cellulose content, we can use this relationship to predict cellulose content.

[0093] Next, we select the density or compressed wood fiber length quantitative model and establish a quantitative model of density or compressed wood fiber length and the component to be evaluated. This step is like building a "translator" that can "translate" the information of density or fiber length into the content of the component we are interested in.

[0094] However, it is important to note that not all components have a direct relationship with density or fiber length. If the component to be evaluated does not have an obvious relationship with density or compressed wood fiber length, we need to directly obtain a quantitative model of the component to be evaluated. This may require us to perform more laboratory analysis to establish a direct relationship between the spectral data and the target component.

[0095] Finally, in an important embodiment of the present invention, we describe in detail how to use the near-infrared spectrum measurement points to calculate the tree segment optimization and determine the near-infrared spectrum collection method for the evaluation class component. First, we fuse the information of multiple frequency band spectra and calculate the average value of multiple frequency band spectra to replace the original data for modeling. This can reduce measurement errors and improve the stability of the model.

[0096] Then, we use the fused spectral information as input. After knowing the objective function, we can determine the optimal tree segment. This process is like solving a complex equation to find the solution that can best balance measurement efficiency and accuracy.

[0097] Through the density quantitative model, we establish a quantitative model of density and the components to be measured. The significance of this step is that we can use the easily measured density to infer the content of other components that may be more difficult to measure directly.

[0098] Finally, we select density as output, input wavelength, collect tree segment and spectral information, output density, and get the estimated result of the component. This process is like we are using a series of "clues" (wavelength, tree segment position, spectral information) to infer the "result" (component content) we are interested in.

[0099] In general, the method proposed in this paper combines near-infrared spectroscopy technology with advanced mathematical models to achieve rapid and accurate evaluation of standing timber properties of Chinese fir. This method not only improves evaluation efficiency and reduces human errors, but also obtains rich information without harming the trees.

[0100] For example, traditional methods for assessing wood properties may require felling trees or taking samples, which is not only time-consuming and labor-intensive, but also destructive to the forest. However, our method only requires scanning trees with a portable near-infrared spectrometer to quickly obtain a large amount of information. This is of great significance for large-scale forest resource surveys and management.

[0101] In addition, our method is highly adaptable and scalable. By adjusting the model parameters and selecting different evaluation components, this method can be applied to different species of trees and even other types of plants. This provides new possibilities for precision management in forestry and agriculture.

[0102] Finally, it is worth emphasizing that this invention is not only a technological innovation, but also an innovation of ideas. It embodies how to combine advanced spectral technology, mathematical modeling and traditional forestry knowledge to create more efficient and environmentally friendly solutions. This interdisciplinary approach provides inspiration for solving complex problems in other fields.

[0103] In order to verify the superiority of the present invention, we conducted a series of experiments to compare the performance of the present invention method with the traditional method in evaluating the properties of standing timber of Chinese fir. The following is a detailed description and analysis of the embodiments and comparative examples.

[0104] We selected a 50-year-old fir forest in Sanming City, Fujian Province as the experimental site. Using the method of the present invention, we conducted a rapid assessment of the standing timber properties of 100 fir trees. First, we divided each tree into 10 equal-length tree segments, and collected near-infrared spectral data in four directions (east, south, west, and north) on each tree segment. Then, we used the algorithm of the present invention to determine the optimal tree segment combination, and finally selected the three key tree segments at the diameter at breast height, 1 / 3 of the tree height, and 2 / 3 of the tree height for detailed analysis.

[0105] We used a portable near-infrared spectrometer (model: LabSpec 4Standard-Res) to collect spectra in three different wavelength ranges (350-1000nm, 1000-1800nm, 1800-2500nm) at four locations in each key tree section. Then, we applied the model and algorithm of the present invention to calculate the wood density (MVS), compressed wood fiber length (MVF) and cellulose content of each tree.

[0106] Comparative Example 1: In the same forest, we randomly selected another 100 fir trees and used the traditional coring method to evaluate the wood properties. We drilled a 5mm diameter wood core sample at the DBH of each tree and then measured the density, fiber length and cellulose content in the laboratory.

[0107] Testing standards and methods:

[0108] 1. Wood density (MVS): Determined in accordance with GB / T 1933-2009 “Determination of wood density”.

[0109] 2. Compressed wood fiber length (MVF): measured using TK-8003 fiber quality analyzer.

[0110] 3. Cellulose content: Determined in accordance with GB / T 744-1989 “Determination of cellulose content in wood – Nitrate alcohol method”.

[0111] The following is a comparison table of the test results of Example 1 and Comparative Example 1:

[0112] index Method of the present invention Comparative Example 1 (Traditional Coring Method) Average wood density (MVS) <![CDATA[0.428g / cm 3 ]]> <![CDATA[0.425g / cm 3 ]]> Wood density (MVS) standard deviation <![CDATA[0.021g / cm 3 ]]> <![CDATA[0.035g / cm 3 ]]> Average value of compressed wood fiber length (MVF) 2.85mm 2.82mm Standard deviation of compressed wood fiber length (MVF) 0.12mm 0.18mm Average cellulose content 45.2％ 44.8％ Standard deviation of cellulose content 1.5％ 2.3％ Single plant evaluation time 15 minutes 2 hours Tree damage none slight

[0113] It can be seen from the above test results that the method of the present invention (Example 1) shows obvious advantages in many aspects:

[0114] 1. Accuracy: The average values ​​of wood density, compressed wood fiber length and cellulose content obtained by the method of the present invention are very close to those of the traditional coring method, which proves the accuracy of the method. For example, the average value of wood density differs by only 0.003 g / cm 3 The relative error is less than 1%. This shows that our near-infrared spectroscopy method can accurately reflect the wood properties of Chinese fir.

[0115] 2. Stability: The method of the present invention shows smaller standard deviations in all measurement indicators. For example, the standard deviation of wood density is 40% smaller than that of the traditional method (0.021 vs 0.035 g / cm 3 ). This means that our method has higher stability and repeatability. This advantage may come from our comprehensive analysis of multiple tree segments and multiple locations, which reduces the error that may be caused by single-point sampling.

[0116] 3. Efficiency: The method of the present invention takes only 15 minutes to evaluate a single tree, while the traditional coring method takes 2 hours. This means that in the same amount of time, our method can evaluate 8 times the sample size of the traditional method. This high efficiency makes large-scale forest resource surveys possible and can provide more comprehensive and timely data support for forestry management decisions.

[0117] 4. Non-destructive evaluation: The method of the present invention does not require any physical damage to the tree, while the traditional coring method will leave small holes in the trunk, which may increase the risk of pests and diseases. Our non-destructive evaluation method is particularly suitable for rare tree species or forest stands that require long-term monitoring.

[0118] 5. Comprehensiveness: Although not directly reflected in the table, our method can obtain data from multiple heights and directions at the same time, providing more comprehensive information on tree wood properties. This is of great value for studying the growth characteristics and environmental adaptability of trees.

[0119] 6. Adaptability: The method of the present invention can quickly adapt to different tree species or different growing environments of Chinese fir by adjusting the model parameters. This flexibility makes the method have a wide range of application prospects.

[0120] In general, the method for rapid evaluation of standing timber properties of Chinese fir based on near-infrared spectroscopy proposed in this invention greatly improves the evaluation efficiency, reduces damage to trees, and provides more comprehensive material property information while ensuring accuracy. These advantages make this method not only suitable for conventional forest resource surveys, but also support advanced applications such as refined forestry management and tree breeding selection. With the accumulation of data and optimization of models, we believe that the advantages of this method will become more obvious and make important contributions to the sustainable development of modern forestry.

[0121] It should be noted that 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 principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for rapid evaluation of standing timber properties of Chinese fir based on near infrared spectroscopy, characterized in that : The method comprises the following steps: Step 1: Divide the trunk of the standing tree to be evaluated into several sections, collect at least four near-infrared spectral sample points on each section, obtain the spectral curve at each tree section, and determine the most suitable tree section for collection; use near-infrared spectroscopy technology to obtain the spectral characteristics of each tree section of the standing tree to be evaluated, and use Xcalibur software to establish a corresponding calibration model for each tree section; use the density MVS (mg / cm 3 ) as the dependent variable and the number of spectra as the independent variable, a density quantitative model was constructed at each tree segment; at the breast height of the standing tree, the compressed wood fiber length MVF (mg / cm 3 ) is the dependent variable, the spectrum number is the independent variable, and a quantitative model of compressed wood fiber length is established; for the determined evaluation component, the measurement object is determined to be the content of the component, and the quantitative model of the evaluation component is determined and established with the corresponding database; Step 2: Using the near-infrared spectrum information at the tree segment as input and the MVF or MVS at the tree segment as output, establish an objective function containing multiple evaluation component contents; solve the objective function at each tree segment through the optimization algorithm, and optimize to obtain the optimal tree segment objective function result with the least near-infrared spectrum; select the tree segment combination in the optimization algorithm as the standing tree segment combination, use it for tree standing test, and finally determine the optimal standing tree segment of the Chinese fir to be evaluated; Step three: Use a portable near-infrared spectrometer to collect spectra in multiple frequency bands at multiple orientations of the optimal tree segment, process the spectrum at each orientation, establish a correction model, and obtain the measured predicted MVS or MVF value at each orientation; perform weighted combination of the measured predicted MVS or MVF values ​​at multiple orientations to obtain the predicted MVS or MVF value at breast height of the standing tree; calculate the content of the evaluation components at breast height based on the quantitative model of the evaluation components in the database.

2. The method for rapid evaluation of standing timber properties of Chinese fir based on near infrared spectroscopy according to claim 1, characterized in that : In step one, after determining the evaluation components, repeatedly obtain the density MVS and compressed wood fiber length MVF values ​​at several tree segments, use the near-infrared spectrum sample points at the tree segments as the input training set, establish a quantitative analysis model, and incorporate the tree segment structure, tree segment diameter, and environmental factors as prediction factors into the calibration model to determine the final calibration model; finally, bring the unknown samples into the final calibration model to calculate the MVS or MVF values ​​at the tree segments.

3. The method for rapid evaluation of standing timber properties of Chinese fir based on near infrared spectroscopy according to claim 2, characterized in that The method for measuring density and compressed wood fiber length specifically includes the following steps: according to the volume of the tree segment determined by near-infrared spectroscopy, compressing the wood in the tree segment under preset normal pressure and high temperature conditions to obtain compressed wood, slicing and drying the compressed wood; taking the compressed wood at the tree segment as the sampling object, measuring the density by weight method; taking the compressed wood at the tree segment as the sampling object, immersing it in a sodium chlorite solution for treatment, taking it out and washing it until there is no red color, and measuring the total weight of compressed wood fibers in the compressed wood by weight method; using an optical microscope to prepare slices for the sampling object, setting the corresponding magnification to obtain compressed wood slices; taking slice images, and calculating the compressed wood fiber length according to the image area; the construction of a quantitative model for compressed wood fiber length includes the following steps: using Image-Pro Plus software to process the slice images of compressed wood after preparation, and obtaining the fiber area of ​​the compressed wood and the outline of the outer edge of the fiber; according to the corresponding relationship between each pixel in the slice and the actual area and the fiber area, obtaining the actual fiber area, and calculating the logarithm of the number of fibers, using the calculation formula of the logarithmic fiber length, obtaining the value of the logarithmic fiber length, and then establishing a quantitative model for the compressed wood fiber length of wood composed of different components of standing trees.

4. The method for rapid evaluation of standing timber properties of Chinese fir based on near infrared spectroscopy as claimed in claim 3, characterized in that The calculation formula of the logarithmic fiber length is: L=log(S / S0) / q0 Wherein, L is the logarithmic fiber length, S is the fiber area, S0 and q0 are constants; the formula of the logarithmic fiber area is: S=A×P, wherein S is the fiber area, A is the actual area corresponding to each pixel, and P is the number of pixels occupied by the fiber in the slice image, representing the actual area of ​​the fiber slice.

5. The method for rapid evaluation of standing timber properties of Chinese fir based on near infrared spectroscopy according to claim 4, characterized in that In step one, the specific process of constructing the MVS or MVF quantitative model at each tree segment is as follows: according to the wood centroid profile at the breast height of the standing tree, the tree segments at the wood centroid are equally spaced, so that the spectral information collected at the tree segment corresponds one to one with the density MVS or compressed wood fiber length MVF data, and the processed spectrum is used as the independent variable, and the MVS or MVF at the tree segment is used as the dependent variable to construct a quantitative analysis model to obtain the density MVS (or compressed wood fiber length MVF) quantitative model at the tree segment.

6. The method for rapid evaluation of standing timber properties of Chinese fir based on near infrared spectroscopy according to claim 5, characterized in that In step 2, the objective function including the content of multiple evaluation components is: Where n is the number of tree segments, M i is the predicted value of the corrected MVS (or corrected MVF) at the tree segment, T i is the MVS (or MVF) value measured at the tree segment, w i is the tree segment weight, Cov(M i M j ) is the cross-covariance value of MNS (or MVF) between tree segments, Cov(M i S i ) is the covariance between MVS (or MVF) and position at the tree segment, var(S i ) is the variance of MVS (or MVF) at the tree segment, and λ1, λ2 and λ3 are balance coefficients.

7. The method for rapid evaluation of standing timber properties of Chinese fir based on near infrared spectroscopy according to claim 6, characterized in that : In step 2, the optimization algorithm for obtaining the optimal tree segment objective function result with the least near-infrared spectrum can be solved. The selection of the tree segment combination is completed by solving the objective function of the optimization algorithm, and the optimal tree segment combination is determined; the tree segment combination is adjusted by solving the coefficients in the objective function. The formula for solving the objective function coefficients is: Among them, a, b, c and d are the coefficients to be solved, f is the objective function; xij represents the eigenvalue of the jth tree segment of the ith sample, i represents the number of samples, and j represents the type of tree segment; di is the tree segment distance of the ith sample, yi is the breast diameter of the ith sample, m is the total number of samples, and n is the total number of tree segments.

8. The method for rapid evaluation of standing timber properties of Chinese fir based on near infrared spectroscopy according to claim 7, characterized in that The rapid evaluation process of evaluating the component content includes: collecting spectra of multiple frequency bands at multiple positions of the optimal tree segment: fusing the spectral information of different frequency bands through the correction model obtained at each tree segment, and calculating the average value of the spectra of multiple frequency bands according to the weighted formula; using the fused spectral information to replace the original spectral information, measuring the model, and obtaining the MVF (or MVS) prediction value at the tree segment, repeating the above operation to obtain the MVS (or MVF) prediction value at different positions; obtaining the MVS or MVF value at the breast height of the standing tree according to the density quantitative model at each tree segment; The quantitative model of the evaluation component is determined according to the density quantitative model, and the MVS (or MVF) prediction value is substituted into the quantitative model of the evaluation component to obtain the content result of the target component, thereby completing the estimation of the evaluation component of Chinese fir; the fused spectral information replaces the original spectral information specifically including, for the preferred combination of tree segments containing the weight term of the distance between the tree segment and the breast height in the objective function, when performing the standing tree structure analysis, calculating the entire standing tree segment, and optimizing by optimizing the distance between the tree segment and the breast height; when fusing information of multiple frequency band spectra, using the fused spectral information to replace the original spectral information, specifically including replacing the original spectrum by calculating the average value of multiple frequency band spectra; determining the optimal frequency band through the combination of tree segments, and fusing the spectrum of the shortest tree segment for the spectrum at the distance between the shortest spectrum and the breast height by the average value of multiple spectrum segments.

9. The method for rapid evaluation of standing timber properties of Chinese fir based on near infrared spectroscopy according to claim 8, characterized in that : After the component to be evaluated is determined, the quantitative model of the component is obtained through the density quantitative model to obtain the estimated result of the component to be evaluated.

10. The method for rapid evaluation of standing timber properties of Chinese fir based on near infrared spectroscopy according to claim 19, characterized in that : The method of obtaining the quantitative model of the component to be evaluated through the density quantitative model means that after obtaining the measured MVS or MVF value of the determined component, the component is associated with the MVS (or MVF), and the density or compressed wood fiber length quantitative model is selected to establish a quantitative model of the density or compressed wood fiber length and the component to be evaluated; if there is no relationship between the component to be evaluated and the density or compressed wood fiber length, the quantitative model of the component to be evaluated is directly obtained; after the objective function result containing the contents of multiple evaluation components is optimized by calculating the tree segment through the near-infrared spectrum measurement point number, the near-infrared spectrum acquisition method of the selected acquisition frequency band to determine the evaluation component is: information fusion of multiple frequency band spectra, calculating the average value of multiple frequency band spectra instead of original data for modeling; using the fused spectral information as input, after the objective function is known, determining the optimal tree segment, and establishing a quantitative model of density and the component to be measured through the density quantitative model; selecting density as output, bringing in wavelength, acquiring tree segment and spectral information, outputting density, and obtaining the estimated result of the component.

Images