Shaving board grading method based on improved decision tree algorithm

Abstract

The invention discloses a shaving board grading method based on an improved decision tree algorithm. The method comprises the steps that shaving board images are collected; manually grading the shaving board image; image data of only one defect on the surface of the shaving board is selected, and the divided shaving board grade is used as a severity degree label of the defect in the image; extracting evaluation indexes of defects in the image to form judgment attributes; respectively generating a decision tree for defect evaluation indexes for each defect; image data with multiple defects on the surface of the shaving board is selected, each image is substituted into all decision-making tree output results for defect evaluation indexes and shaving board grades divided by the image to construct a training set, and all defect types are constructed into an attribute set; generating a decision tree divided for defect types; constructing a grading method; the grading method is applied to the production and detection link of the shaving board. By means of the method, accurate grading of the quality of the shaving board can be achieved, the detection precision and the working efficiency are improved, and the calculation cost is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of particleboard processing quality detection, and specifically relates to a particleboard grading method based on an improved decision tree algorithm, aiming to achieve precise grading of its quality through the processing and analysis of particleboard image data. Background Art

[0002] Particleboard, as a widely used wood-based panel, has important applications in the fields of furniture manufacturing, building decoration, etc. During the production process of particleboard, its surface quality is directly related to the overall quality and market competitiveness of the product. However, at present, particleboard production enterprises still mainly rely on arranging specialized inspectors for manual visual sorting in surface defect detection.

[0003] With the continuous acceleration of the production line speed, the disadvantages of the manual visual sorting method have become increasingly prominent. The surface defects of particleboard are usually relatively small compared to the entire board surface. On a high-speed particleboard production line, it is difficult for the human eye to accurately and efficiently detect these surface defects. At the same time, it is also difficult to distinguish the types of particleboard surface defects in a short time, resulting in low recognition rate and high false detection and missed detection rates. In addition, due to the extremely strong subjectivity of the manual visual sorting method and the lack of a unified appearance quality grading standard for particleboard surface defects at present, different inspectors may make very different judgments on the same batch of particleboard with different surface defects, resulting in low accuracy and unity in the judgment of particleboard appearance surface defects.

[0004] In recent years, machine vision technology has been increasingly widely used in the fields of industrial inspection, etc. Multiple scientific research teams have conducted research on the machine vision detection system for particleboard surface defects and made some new progress. The machine vision detection system for particleboard surface defects usually collects images through an industrial camera, and in the corresponding software on the computer, specific detection algorithms are used to extract the characteristics of the particleboard surface defects in the images, and the detection and recognition of particleboard surface defects are realized based on these characteristic information. Compared with the manual visual sorting method, the machine vision detection system for particleboard surface defects has higher detection accuracy, higher work efficiency, and can realize digital statistical management.

[0005] However, this solution is still in the stage of application research and development, and there are still some problems to be further improved. For example, the current detection algorithms for surface defect detection of particleboard mainly focus on the object detection tasks of certain types of defects. However, in the actual production of particleboard, the size, style, quantity, type, etc. of surface defects are numerous, and there is a lack of a solution for further grading the particleboard according to these defects. After successfully extracting the surface defect features of the particleboard, in order to achieve the grading process of the particleboard quality, a corresponding grading algorithm needs to be designed. This algorithm should be able to comprehensively consider multiple dimensions such as the type, quantity, area, and severity of the defects, and set different quality grades for the particleboard. However, due to the randomness of the appearance of surface defects of the particleboard, the subsequent processing methods for different defects are also different. Therefore, the grading of the particleboard is not simple and mechanical, but requires comprehensive consideration in combination with multiple dimensions. Currently, those skilled in the art have tried to use the decision tree algorithm to evaluate the quality grade of the particleboard according to multiple features of the surface defects of the particleboard. However, there are many types of particleboard defects, such as sand leakage, dust spots, glue spots, indentations, etc., and the evaluation indexes for each type of defect are different, such as area, quantity, range, dispersion degree, etc. If the traditional decision tree algorithm is directly used, there will be an attribute set containing dozens of elements, and the generated decision tree will be very complex and redundant, with a high calculation cost, which is not conducive to the application in actual production. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a particleboard grading method based on an improved decision tree algorithm for the deficiencies of the above-mentioned prior art. This method aims to overcome the disadvantages of the traditional manual visual sorting method, such as low efficiency, poor accuracy, and strong subjectivity. At the same time, it solves the deficiencies of the existing machine vision detection system in the grading of particleboard, such as the lack of a grading algorithm that comprehensively considers multiple dimensions, and the complexity and redundancy of the decision tree algorithm. Through the method of the present invention, the accurate grading of the particleboard quality can be achieved, the detection accuracy and work efficiency can be improved, the calculation cost can be reduced, and an efficient, accurate, and unified grading solution can be provided for particleboard production enterprises, thereby enhancing the market competitiveness of products and the economic benefits of enterprises.

[0007] To achieve the above technical objectives, the technical solution adopted by the present invention is as follows:

[0008] A particleboard grading method based on an improved decision tree algorithm, comprising the following steps:

[0009] Step 1: Collect particleboard image data and construct a data set for generating a decision tree;

[0010] Step 2: Manually grade the collected particleboard image data into four grades;

[0011] Step 3, select the image data with only one defect on the particleboard surface, and use the particleboard grade divided in the above step 2 as the severity label of the defect in the image; use the image processing method to extract the evaluation index of the defect in the image, the evaluation index includes area, quantity, range and dispersion, which constitute the determination attribute; construct the result of the determination attribute in each image and the severity of the defect on the image as a training set, construct the evaluation index of the defect as an attribute set, and generate a decision tree for the defect evaluation index for each defect; when generating the decision tree, according to the characteristics of different defects, the defect evaluation index is prioritized, which is specifically reflected in the attribute priority weighting in the decision tree generation algorithm;

[0012] Step 4: Select image data with various defects on the particleboard surface, bring each image into the decision tree for all defect evaluation indicators generated in the above step 3, and save the output result of each defect; construct the result of each defect obtained for each image and the particleboard grade divided by the image in the above step 2 as a training set, and construct all defect types as an attribute set; generate a decision tree for defect type division;

[0013] Step 5, construct a grading method: determine whether the particleboard surface contains defects. If there are no defects, it is a first-class board; if there are defects, further determine the type of surface defects of the particleboard, and extract the specific value of the evaluation index of each defect. If it contains only a single defect, it is brought into the decision tree for the defect evaluation index, and the severity of the defect determined by the decision tree is used as the grading level of the particleboard; if it contains multiple defects, it is brought into the decision tree for the defect type classification. At each judgment node, the evaluation index of this defect is brought into the decision tree for the defect evaluation index generated in step 3, and its output result is used as the judgment result of the node, until the decision tree for the defect type classification is completed, and the output result is used as the grading level of the particleboard;

[0014] Step 6: Apply the grading method obtained in step 5 above to the particleboard production and inspection process: after collecting the particleboard image and processing the image, pass the processed data into the grading method obtained in step 5 above, and classify the particleboard according to the output result. At this point, the particleboard grading is completed.

[0015] As a further improved technical solution of the present invention, the step 1 is specifically: using an industrial camera to collect particleboard image data and constructing a data set for generating a decision tree.

[0016] As a further improved technical solution of the present invention, the four levels in step 2 are specifically a primary board, a secondary board, a tertiary board and a quaternary board.

[0017] As a further improved technical solution of the present invention, step 3 is specifically as follows:

[0018] Step 3.1: Select the image data with only one type of defect on the surface of the particle board in step 2;

[0019] Step 3.2: Take the grade of the particle board divided for the image data in step 2 as the severity of the defect in the corresponding image;

[0020] Step 3.3: For each type of defect, use an image processing method to extract the evaluation indexes of the defect in the image. The evaluation indexes include area, quantity, range, and dispersion degree;

[0021] Step 3.4: Construct a training set with the results of the defect evaluation indexes extracted from each image and the severity of the defect on the corresponding image. Each item in the training set includes the data of each evaluation index of a type of defect and the corresponding severity label;

[0022] Step 3.5: Construct an attribute set with the evaluation indexes of the defect;

[0023] Step 3.6: For each type of defect, use the training set constructed in step 3.4 and the attribute set constructed in step 3.5, and adopt an improved ID3 decision tree learning algorithm to generate a decision tree for the defect evaluation indexes. The improved ID3 decision tree learning algorithm introduces a weight Ka set according to different defect types and each evaluation index, which is used to correct the influence degree of different types of defects on a certain evaluation index. The formula for selecting the optimal partitioning attribute by multiplying the information gain by the weight of the corresponding feature is optimized as: a* = arg(Ka·max(Gain(D,a))); where Gain(D,a) is the information gain, and Ka is the weight corresponding to a certain evaluation index in a type of defect.

[0024] As a further improved technical solution of the present invention, step 4 is specifically as follows:

[0025] Step 4.1: Select the image data of the particle board with multiple defects on the surface from the particle board image dataset constructed in step 2;

[0026] Step 4.2: Substitute each selected image with multiple defects into the decision trees for the defect evaluation indexes generated for each type of defect in step 3 respectively to obtain the output results of each type of defect. The results include the severity of each type of defect;

[0027] Step 4.3: Construct a new training set with the integrated results of multiple defects of each image and the grade of the particle board corresponding to this image divided in step 2;

[0028] Step 4.4: Take all defect types in each image as the attribute set;

[0029] Step 4.5: Use the traditional ID3 decision tree learning algorithm, and utilize the training set constructed in Step 4.3 and the attribute set constructed in Step 4.4 to generate a decision tree for defect type classification.

[0030] As a further improved technical solution of the present invention, the defect types include scratches, glue spots, indentations, large shavings, dust spots, and sand leakage.

[0031] The beneficial effects of the present invention are as follows:

[0032] (1) Compared with the traditional manual visual sorting method, the present invention uses machine vision technology and an improved decision tree algorithm to more accurately detect and grade the surface defects of particleboards. Machine vision technology can efficiently collect and process image data, while the improved decision tree algorithm can comprehensively consider various defect features to achieve more accurate grading.

[0033] (2) The manual visual sorting method is not only inefficient but also prone to inconsistent judgment results due to strong subjectivity. The present invention reduces the dependence on manual detection through an automated method, thereby reducing labor costs and errors caused by subjective factors.

[0034] (3) The present invention provides a set of clear grading standards and methods, classifying particleboards into four grades, and each grade has specific definitions of defect features and severity. This helps to unify the judgment standards for the quality of particleboards in the industry, improve the standardization of the market, and the comparability of products.

[0035] (4) There are many types of surface defects on particleboards, and the impact of each defect on quality is different. The present invention can comprehensively consider the features and severity of different defects by generating decision trees for single defects and multiple defects respectively, achieving more comprehensive and reasonable grading.

[0036] (5) When the traditional decision tree algorithm processes particleboard defects, due to the complex and redundant attribute set, the calculation cost is relatively high. The present invention optimizes the decision tree generation process by introducing attribute priority weighting and an improved information gain calculation method, reducing the computational complexity and making it more suitable for practical production applications. Description of the Drawings

[0037] Figure 1 Schematic diagrams of surface defects of various types of particleboards.

[0038] Figure 2 Schematic diagram of the decision tree for sand leakage defects.

[0039] Figure 3It is a schematic diagram of a decision tree divided according to defect types. Specific implementation manners

[0040] The following further describes the specific implementation manners of the present invention with reference to the accompanying drawings:

[0041] A particleboard grading method based on an improved decision tree algorithm includes:

[0042] Step 1: Use an industrial camera to collect particleboard image data and construct a data set for generating a decision tree.

[0043] Step 2: Manually grade the collected particleboard data. Considering the specific conditions of surface defects of the particleboard and actual production experience, the particleboard is divided into four grades.

[0044] Step 3: Select the data with only one type of defect on the surface of the particleboard, and use the particleboard grade divided in Step 2 above as the severity label of the defect in this image. Adopt a general image processing method to extract evaluation indicators such as the area, quantity, range, and dispersion degree of the defect in the image to form determination attributes. Construct a training set with the results of the determination attributes in each image and the severity of the defect on this image, and construct an attribute set with the evaluation indicators of this type of defect. Generate a decision tree for the defect evaluation indicators (i.e., defect attributes) for each type of defect. When generating the decision tree, according to the characteristics of different defects, divide the defect evaluation indicators into priorities, which is specifically reflected as attribute priority weighting in the decision tree generation algorithm. The defect types include scratches, glue spots, indentations, large particles, dust spots, and sand leakage. Defect images are as Figure 1 shown.

[0045] Step 4: Select the data with multiple types of defects on the surface of the particleboard, input each image into all the decision trees for defect evaluation indicators generated in Step 3 above, and save the output results of each type of defect. Construct a training set with the results of each type of defect obtained from each image and the particleboard grade divided in Step 2 above for this image, and construct an attribute set with all defect types. Generate a decision tree for defect type division.

[0046] Step 5, construct a grading method. Determine whether the particleboard surface contains defects. If there are no defects, it is a first-class board; if there are defects, further determine the type of surface defects of the particleboard, and extract the specific value of the evaluation index of each defect. If it contains only a single defect, bring it into the decision tree for the defect evaluation index, and use the severity of the defect determined by the decision tree as the grading level of the particleboard; if it contains multiple defects, bring it into the decision tree for the defect type classification, and at each judgment node, bring the evaluation index of this defect into the decision tree for the defect evaluation index generated in step 3, and use its output result as the judgment result of the node, until the decision tree for the defect type classification is completed, and the output result is used as the grading level of the particleboard.

[0047] Step 6: Apply the specific grading method obtained in step 5 above to the particleboard production and testing process. After the camera captures the particleboard image and undergoes a series of necessary general image processing, the data is passed into the grading method, and the particleboard is graded according to the output results. At this point, the particleboard grading is completed.

[0048] The step 2 is specifically as follows:

[0049] Step 2.1, the four grades are as follows: Level 1 board, no defects or only very few minor defects, can be produced directly.

[0050] Step 2.2: The secondary board has a certain number of slight defects or a small number of medium defects, and can be produced after manual flipping.

[0051] Step 2.3: The third-level board has multiple medium defects or a small number of serious defects. It can be produced after manual flipping.

[0052] Step 2.4: The fourth-level board has multiple serious defects, which affect its use and is directly scrapped.

[0053] The step 3 is specifically as follows:

[0054] Step 3.1, select the data with only one defect on the particleboard surface in step 2. For example, select the image data with only sand leakage defect, or select the data with only other single defects such as dust spots, glue spots, indentation, etc.

[0055] Step 3.2: The particleboard grade classified for the image in step 2 is used as the severity of the defect in the image. For example, if a particleboard with a sand leakage defect is manually classified as a second-level board, then the severity of the sand leakage defect of the board is second-level.

[0056] Step 3.3: For each type of defect, use general image processing methods, including but not limited to region segmentation, edge detection, morphological operations, etc., to extract evaluation indicators such as the area, quantity, range, and dispersion degree of the defect in the image. These evaluation indicators are key parameters describing the defect characteristics and characterize the severity of the defect from different perspectives. For example, for the sand leakage defect, the image processing method can calculate the area of the sand leakage region, count the number of sand leakage points, measure the maximum expansion range of the sand leakage region, and analyze the dispersion degree between the sand leakage points, etc., providing quantitative data for the subsequent decision tree generation.

[0057] Step 3.4: Construct the results of the determination attributes (i.e., defect evaluation indicators) extracted from each image, and the severity of the defect on this image, into a training set. Each item in the training set includes the data of various evaluation indicators of the defect and the corresponding severity label. For example, for the training set of the sand leakage defect, D_lousha = {(x_1: ["area": 100, "quantity": 5, "range": 120, "dispersion degree": "disperse"], y_1: "level three"), (x_2: ["area": 20, "quantity": 1, "range": 15, "dispersion degree": "concentrate"], y_2: "level two"), … (x_n: […], y_n: …)}, where n is the total number of the sand leakage data set.

[0058] Step 3.5: Construct the evaluation indicators of this type of defect (such as area, quantity, range, dispersion degree, etc.) into an attribute set. The attribute set is the input feature set in the decision tree algorithm. When generating the decision tree, the subsequent algorithm will classify and make decisions on the samples according to these attributes. For example, for the sand leakage defect, the attribute set is A_lousha = {a_1: "area", a_2: "quantity", a_3: "range", a_4: "dispersion degree"}. These attributes will be used as the nodes of the decision tree and be split and selected in the algorithm to generate the optimal decision tree structure.

[0059] Step 3.6: For each type of defect, use the constructed training set and attribute set above to generate a decision tree for the defect evaluation index. This method is improved based on the traditional ID3 decision tree learning algorithm to better adapt to the particleboard classification and grading tasks. For the surface defects of particleboard, for the same index but different types of defects, the impact on grading varies greatly. For example, if the sand leakage area accounts for 20% of the board surface, it can be rated as a second-grade board, while the same area of large wood chips is rated as a fourth-grade board. Thus, it can be seen that for the same degree of the area index but different types, there will be a large difference in grading. This method optimizes the selection of the optimal partitioning attribute a* in ID3. The traditional ID3 selects the partitioning attribute based on the information gain criterion, a* = arg(max(Gain(D, a))), where Gain(D, a) is the information gain. This method introduces a weight Ka set according to different defect types and various evaluation indexes to correct the influence degree of different types of defects on a certain evaluation index. The specific weight value is set according to the influence degree of a certain index on the particleboard grading in actual production experience. Multiply the information gain by the weight of the corresponding feature, and the formula for selecting the optimal partitioning attribute is optimized to a* = arg(Ka·max(Gain(D, a))). For example, for the sand leakage defect, input the training set D_lousha and the attribute set A_lousha into the improved decision tree learning algorithm TreeGenerate_lousha for the sand leakage defect, and generate the decision tree for the sand leakage defect (i.e., the decision tree for the sand leakage defect evaluation index) Tree_lousha as shown in Figure 2 Figure Tree_lousha for the sand leakage defect (i.e., the decision tree for the sand leakage defect evaluation index) as shown, that is, Tree_lousha = TreeGenerate_lousha(D_lousha, A_lousha).

[0060] The specific content of Step 4 is as follows:

[0061] Step 4.1: From the particleboard image dataset constructed in Step 2, select the data with multiple surface defects. These data may have multiple defect types such as sand leakage, dust spots, and glue spots at the same time, ensuring that the selected data can cover various defect combination situations that may occur in actual production, providing a basis for generating a multi-defect decision tree with universality in the follow-up.

[0062] Step 4.2: Bring each image with multiple defects selected into all the decision trees for single-defect evaluation indexes generated in Step 3. For example, for an image with both sand leakage and glue spot defects at the same time, bring it into the sand leakage defect decision tree (i.e., the decision tree for the sand leakage defect evaluation index) and the glue spot defect decision tree (i.e., the decision tree for the glue spot defect evaluation index) respectively. Through these single-defect decision trees, obtain the output results of each defect, and these results include relevant information such as the severity determination of each defect.

[0063] Step 4.3: Construct a new training set with the multiple defect results of each integrated image and the corresponding particleboard grade divided in Step 2. For example, construct the training set D_multi = {(x_1: ["sand leakage": "Grade 2", "dust spot": "Grade 2", "glue spot": "Grade 1", "large shaving": "Grade 1"], y_1: "Grade 2"), (x_2: ["sand leakage": "Grade 2", "dust spot": "Grade 2", "glue spot": "Grade 2", "large shaving": "Grade 3"], y_1: "Grade 4"), … (x_m: […], y_m: …)}, where m is the total number of datasets containing multiple defects.

[0064] Step 4.4: Take all defect types as the attribute set. For example, if there are four defects of sand leakage, dust spot, glue spot, and large shaving, then the attribute set is A_multi = {a_1: "sand leakage", a_2: "dust spot", a_3: "glue spot", a_4: "large shaving"}. This attribute set is the key to generating a multi-defect decision tree, and the decision tree will split and make decisions based on these attributes to judge the quality grade of the particleboard.

[0065] Step 4.5: Use the traditional ID3 decision tree learning algorithm TreeGenerate, and use the above-constructed training set D_multi and attribute set A_multi to generate a Figure 3 decision tree Tree_multi for defect type division as shown, that is, Tree_multi = TreeGenerate(D_multi, A_multi). During the generation process, the algorithm will automatically learn and determine the importance and priority of different defect types when dividing the quality grade according to the correlation between the multiple defect feature information in the training set and the particleboard grade.

[0066] Compared with the traditional manual visual sorting method, the present invention utilizes machine vision technology and an improved decision tree algorithm to more accurately detect and grade the surface defects of particle boards. Machine vision technology can efficiently collect and process image data, while the improved decision tree algorithm can comprehensively consider various defect features to achieve more accurate grading. The manual visual sorting method is not only inefficient but also prone to inconsistent judgment results due to strong subjectivity. The present invention reduces the dependence on manual inspection through an automated method, thereby reducing labor costs and errors caused by subjective factors. The present invention provides a set of clear grading criteria and methods, classifying particle boards into four grades, each grade having specific definitions of defect features and severity. This helps to unify the quality judgment criteria for particle boards in the industry, improve the standardization of the market, and enhance the comparability of products. There are various types of surface defects in particle boards, and the impact of each defect on quality is different. The present invention can comprehensively consider the features and severity of different defects by generating decision trees for single defects and multiple defects respectively, achieving more comprehensive and reasonable grading. When dealing with particle board defects, the traditional decision tree algorithm has a high computational cost due to a complex and redundant attribute set. The present invention optimizes the decision tree generation process by introducing attribute priority weighting and an improved information gain calculation method, reducing the computational complexity and making it more suitable for application in actual production.

[0067] The protection scope of the present invention includes but is not limited to the above embodiments. The protection scope of the present invention shall be subject to the claims, and any substitutions, deformations, and improvements that are easily conceivable by those skilled in the art to this technology shall fall within the protection scope of the present invention.

Claims

1. A particleboard grading method based on an improved decision tree algorithm, characterized in that: The following steps are involved: Step 1, collecting particleboard image data and constructing a data set for generating a decision tree; Step 2: manually classify the collected particleboard image data into four levels; Step 3, select the image data with only one defect on the particleboard surface, and use the particleboard grade divided in the above step 2 as the severity label of the defect in the image; use the image processing method to extract the evaluation index of the defect in the image, the evaluation index includes area, quantity, range and dispersion, which constitute the determination attribute; construct the result of the determination attribute in each image and the severity of the defect on the image as a training set, construct the evaluation index of the defect as an attribute set, and generate a decision tree for the defect evaluation index for each defect; when generating the decision tree, according to the characteristics of different defects, the defect evaluation index is prioritized, which is specifically reflected in the attribute priority weighting in the decision tree generation algorithm; Step 4: Select image data with various defects on the particleboard surface, bring each image into the decision tree for all defect evaluation indicators generated in the above step 3, and save the output result of each defect; construct the result of each defect obtained for each image and the particleboard grade divided by the image in the above step 2 as a training set, and construct all defect types as an attribute set; generate a decision tree for defect type division; Step 5, construct a grading method: determine whether the particleboard surface contains defects. If there are no defects, it is a first-class board; if there are defects, further determine the type of surface defects of the particleboard, and extract the specific value of the evaluation index of each defect. If it contains only a single defect, it is brought into the decision tree for the defect evaluation index, and the severity of the defect determined by the decision tree is used as the grading level of the particleboard; if it contains multiple defects, it is brought into the decision tree for the defect type classification. At each judgment node, the evaluation index of this defect is brought into the decision tree for the defect evaluation index generated in step 3, and its output result is used as the judgment result of the node, until the decision tree for the defect type classification is completed, and the output result is used as the grading level of the particleboard; Step 6: Apply the grading method obtained in step 5 above to the particleboard production and inspection process: after collecting the particleboard image and processing the image, pass the processed data into the grading method obtained in step 5 above, and classify the particleboard according to the output result. At this point, the particleboard grading is completed.

2. The particleboard grading method based on the improved decision tree algorithm according to claim 1 is characterized in that: The step 1 is specifically as follows: using an industrial camera to collect particleboard image data and constructing a data set for generating a decision tree.

3. The particleboard grading method based on the improved decision tree algorithm according to claim 1 is characterized in that: The four levels in step 2 are specifically a primary board, a secondary board, a tertiary board and a quaternary board.

4. The particleboard grading method based on the improved decision tree algorithm according to claim 1 is characterized in that: The step 3 is specifically as follows: Step 3.1, selecting the image data in which only one defect exists on the surface of the particleboard in step 2; Step 3.2, using the particleboard grade divided by the image data in step 2 as the severity of the defect in the corresponding image; Step 3.3, for each defect, use image processing methods to extract evaluation indicators of defects in the image, including area, quantity, range and dispersion; Step 3.4, the results of the defect evaluation index extracted from each image and the severity of the defect on the corresponding image are constructed into a training set; each item in the training set includes various evaluation index data of a defect and the corresponding severity label; Step 3.5, construct the defect evaluation index as an attribute set; Step 3.6: For each defect, use the training set constructed in step 3.4 and the attribute set constructed in step 3.5, and adopt the improved ID3 decision tree learning algorithm to generate a decision tree for the defect evaluation index; the improved ID3 decision tree learning algorithm introduces the weight Ka set according to different defect types and various evaluation indicators, which is used to correct the influence of different types of defects on a certain evaluation indicator, and multiply the information gain by the weight of the corresponding feature to select the optimal partition attribute formula: a*=arg(Ka·max(Gain(D,a))); wherein Gain(D,a) is the information gain, and Ka is the weight corresponding to a certain evaluation indicator in a defect.

5. The particleboard grading method based on the improved decision tree algorithm according to claim 1 is characterized in that: The step 4 is specifically as follows: Step 4.1, selecting particleboard image data with multiple defects on the surface from the particleboard image data set constructed in step 2; Step 4.2, each selected image with multiple defects is respectively brought into the decision tree for defect evaluation index generated for each defect in step 3, and the output result of each defect is obtained, and the result includes the severity of each defect; Step 4.3, constructing a new training set with the integrated multiple defect results of each image and the particleboard grade corresponding to the image divided in step 2; Step 4.4, taking all defect types in each image as attribute sets; Step 4.5: Use the traditional ID3 decision tree learning algorithm to generate a decision tree for defect type classification using the training set constructed in step 4.3 and the attribute set constructed in step 4.

4.

6. The particleboard grading method based on the improved decision tree algorithm according to claim 1, characterized in that: The types of defects include scratches, glue spots, indentations, large shavings, dust spots, and sand leakage.

Images