A visual precision positioning system applied to a log splitter

Abstract

The present invention belongs to the technical field of visual positioning, and specifically discloses a visual precision positioning system applied to a wood splitter. The system includes an image acquisition module, a wood species recognition module, a positioning correction module, and a pusher and wood splitting module. In this solution, a local quality index image is obtained by calculating the local quality index, the image quality of any region is quantified, the negative impacts of interference features such as saw marks, cracks, and dirt are reduced, the accuracy of the wood splitter operation is improved, the deep learning model pays more attention to high-quality regions, and selectively ignores low-quality regions, thereby enhancing the reliability of feature extraction and wood species recognition; according to the wood species, the shear strength of the wood is determined, a global shear strength image is obtained using a linear regression model, the cutting angle of the wood splitting unit is adjusted, the pushing speed of the pusher unit is optimized, and the wood splitter is assisted to reduce energy consumption and save costs.

Description

Technical Field

[0001] The present invention relates to the technical field of visual positioning, and specifically refers to a visual precision positioning system applied to a wood splitter. Background Art

[0002] A visual precision positioning system applied to a wood splitter refers to a system that optimizes the wood splitter by using computer vision technology.

[0003] In existing approximate solutions, for example, CN114820606B, a control system and method for a laser cutting device based on visual positioning. This solution addresses the technical problem in the prior art of lacking a method for detecting the cutting effect of the cut material and adjusting the parameters of the laser cutting device according to the detection results. By performing HSV color space conversion on the cross-section image, finding the main cutting seam, calculating the notch coefficient and the flatness of the cutting surface, and adjusting the parameters of the laser cutting device, it realizes the detection of the cutting effect of the cut material and the adjustment of the parameters of the laser cutting device according to the detection results, ensuring the technical effect of the cutting accuracy of subsequent materials. However, there is a technical problem that it does not consider the negative impact of interference features such as cracks and saw marks in the cutting material, which reduce the significance of important features.

[0004] Knots in wood reduce the strength of the wood, making it more prone to breakage or deformation. The internal texture structure of the wood also directly affects the required pressure tons of the pusher device and the cutting angle of the wood splitting device. There is a technical problem that the existing wood splitter system has not yet had a solution for adjusting the operation accuracy of the pusher device and the wood splitting device according to the wood strength. Summary of the Invention

[0005] In view of the above situation, to overcome the defects of the prior art, the present invention provides a visual precision positioning system applied to a wood splitter. Regarding the technical problem in the prior art that it does not consider the negative impact of interference features such as cracks and saw marks in the cutting material, which reduce the significance of important features, this solution obtains a local quality index image by calculating the local quality index, quantifies the image quality of any region, reduces the negative impact of interference features such as saw marks, cracks, and dirt, improves the accuracy of the wood splitter operation, enables the deep learning model to pay more attention to high-quality regions, and selectively ignores low-quality regions, thereby enhancing the reliability of feature extraction and wood species recognition. Regarding the technical problem that the existing wood splitter system has not yet had a solution for adjusting the operation accuracy of the pusher device and the wood splitting device according to the wood strength, this solution determines the shear strength of the wood according to the wood species, uses a linear regression model to obtain a global shear strength image, adjusts the cutting angle of the wood splitting unit, optimizes the pushing speed of the pusher unit, and assists the wood splitter in reducing energy consumption and saving costs.

[0006] The technical solution adopted by the present invention is as follows: The present invention provides a visual precision positioning system applied to a wood splitter. The visual precision positioning system applied to a wood splitter includes an image acquisition module, a wood species identification module, a positioning correction module, and a feeding and wood splitting module;

[0007] The image acquisition module uses an industrial camera to obtain a cross-sectional image of the wood;

[0008] The wood species identification module adopts a wood species identification method to analyze the cross-sectional texture according to the cross-sectional image of the wood and identify the wood species;

[0009] The positioning correction module adopts a positioning correction method to determine the global shear strength image of the wood according to the wood species:

[0010] The feeding and wood splitting module includes a wood splitting unit and a feeding unit; the feeding and wood splitting module adjusts the cutting angle of the wood splitting unit and optimizes the feeding speed of the feeding unit according to the global shear strength image.

[0011] The wood species identification module adopts a wood species identification method. The wood species identification method specifically includes the following steps:

[0012] Step S1: Cross-section optimization, which is used to reduce the influence of interference features and extract local structures. The interference features refer to the cracks and saw marks of the wood. The specific operations are as follows;

[0013] Step S11: Obtain a filtered image. Specifically, a group of filters with different directional characteristics are used. The filter contains a Gaussian function with zero mean. By changing the variances of the filter in the horizontal and vertical directions, the directional characteristics of the filter are adjusted. The group of filters are respectively applied to the cross-sectional image of the wood to obtain a group of filtered images. The filtered images are used to highlight the texture corresponding to the directional characteristics of the filter in the cross-sectional image of the wood;

[0014] Step S12: Reduce noise. Specifically, the majority voting method is adopted to smooth the filtered image;

[0015] Step S13: Calculate the response value. Specifically, all pixel points in the filtered image are traversed. The pixel values of the neighboring pixel points around each pixel point are respectively convolved with all filters to obtain the response value of each pixel point under the corresponding filter. The direction of the filter with the largest response value is recorded as the dominant direction of the pixel point. The dominant directions of all pixel points are recorded to obtain a direction map;

[0016] Step S2: Quality assessment, which is used to construct a local quality map. The specific operations are as follows;

[0017] Step S21: Eliminate discontinuity, which is used to eliminate the discontinuity caused by the direction angles of 0° and 180°. Specifically, calculate the local direction consistency between each pixel point in the direction map and its surrounding neighboring pixel points. The smaller the local direction consistency, the more consistent the direction of the pixel point with that of its surrounding neighboring pixel points, and the clearer the texture of the annual rings. The calculation formula for the local direction consistency is as follows:

[0018] ;

[0019] In the formula, represents the local direction consistency, represents the dominant direction of each pixel point in the direction map, represents the direction map rotated by 90°, represents the operation of calculating the standard deviation, represents the operation of taking the minimum value;

[0020] Step S22: Calculate the local direction matching degree, which is used to evaluate the wood texture. Specifically, use a computer vision algorithm to obtain the connection line from each pixel point to the pith of the wood, and calculate the local direction matching degree between the connection line and the dominant direction of the pixel point in the direction map. The smaller the local direction matching degree, the more the direction of the pixel point matches the texture direction of the annual rings. The calculation formula for the local direction matching degree is as follows:

[0021] ;

[0022] In the formula, represents the local direction matching degree, represents the direction of the connection line, represents the direction of the connection line rotated by 90°, represents the modulo operation;

[0023] Step S23: Calculate the local quality index, which is used to evaluate the visibility of the annual rings and the regional quality. Specifically, perform an element-wise multiplication operation on the local direction consistency and the local direction matching degree of all pixel points to obtain the local quality index. The higher the local quality index, the higher the weight of the pixel point for wood species identification. According to the local quality index, construct a local quality map;

[0024] Step S3: Species output, which is used to identify the wood species using a convolutional neural network. Specifically, use the local quality index in the local quality map as the weight of the wood cross-sectional image, and use the local quality map as an additional channel to be embedded into the wood cross-sectional image as input data for the convolutional neural network. The convolutional neural network outputs the wood species.

[0025] The positioning correction module adopts a positioning correction method, and the positioning correction method specifically includes the following steps:

[0026] Step M1: Determine the shear strength. Specifically, according to the type of wood, determine the wood density and the initial shear strength.

[0027] Step M2: Determine the internal structure. Specifically, use a laser scanner to obtain the point cloud data of the wood surface, and based on wood science knowledge, reconstruct the internal texture structure of the wood.

[0028] Step M3: Determine the stiffness coefficient. Specifically, use a stress wave propagation device to measure the acoustic wave propagation speed of the stress wave in the wood, and calculate the wood stiffness coefficient. The formula used is as follows:

[0029] ;

[0030] In the formula, represents the stiffness coefficient, represents the acoustic wave propagation speed, represents the wood density;

[0031] Step M4: Stiffness correction, which is used to correct the initial shear strength using a linear regression model. Specifically, the presence of knots will affect the shear strength of the wood. Using computer vision technology, identify the position and shape of the knots in the wood cross-sectional image, adjust the weight of the wood cross-sectional image according to the position and shape of the knots, and use the position and shape of the knots, the initial shear strength, and the local quality map as three additional channels, and embed them into the internal texture structure and wood cross-sectional image of the wood as the input data of the linear regression model. The linear regression model outputs a global shear strength image.

[0032] The beneficial effects achieved by the present invention using the above solution are as follows:

[0033] (1) Aiming at the technical problem that the existing technology does not consider the negative impact of interference features such as cracks and saw marks in the cutting material, which reduce the significance of important features, this solution obtains a local quality index image by calculating the local quality index, quantifies the image quality of any area, reduces the negative impact of interference features such as saw marks, cracks, and dirt, improves the accuracy of the wood splitter operation, enables the deep learning model to pay more attention to high-quality areas, and selectively ignores low-quality areas, thereby improving the reliability of feature extraction and wood species identification;

[0034] (2) Aiming at the technical problem that there is no solution in the existing wood splitter system to adjust the operation accuracy of the feeding device and the wood splitting device according to the wood strength, this solution determines the shear strength of the wood according to the wood type, uses a linear regression model to obtain a global shear strength image, adjusts the cutting angle of the wood splitting unit, optimizes the feeding speed of the feeding unit, and helps the wood splitter reduce energy consumption and save costs. Brief Description of the Drawings

[0035] Figure 1 Module connection diagram of a visual precision positioning system for a log splitter provided by the present invention;

[0036] Figure 2 Flow schematic diagram of a wood species identification method;

[0037] Figure 3 Schematic diagram of step S1;

[0038] Figure 4 Schematic diagram of step S2;

[0039] Figure 5 Flow schematic diagram of a positioning correction method.

[0040] The accompanying drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention, and do not constitute a limitation to the present invention. Specific embodiments

[0041] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work belong to the scope of protection of the present invention.

[0042] Embodiment 1: Refer to Figures 1 to 5 , this embodiment provides a visual precision positioning system for a log splitter. The visual precision positioning system for a log splitter includes an image acquisition module, a wood species identification module, a positioning correction module, and a feeding and splitting module;

[0043] The image acquisition module uses an industrial camera to obtain a cross-sectional image of the wood;

[0044] The wood species identification module adopts a wood species identification method to analyze the cross-sectional texture according to the cross-sectional image of the wood and identify the wood species;

[0045] The positioning correction module adopts a positioning correction method to determine the global shear strength image of the wood according to the wood species:

[0046] The feeding and splitting module includes a splitting unit and a feeding unit; the feeding and splitting module adjusts the cutting angle of the splitting unit and optimizes the feeding speed of the feeding unit according to the global shear strength image.

[0047] Embodiment 2: Refer to Figures 1 to 4, this embodiment is based on the above - mentioned embodiment. The wood species recognition module adopts a wood species recognition method, and the wood species recognition method specifically includes the following steps:

[0048] Step S1: Cross - section optimization, which is used to reduce the influence of interference features and extract local structures. The interference features refer to the cracks and saw marks on the wood. The specific operations are as follows;

[0049] Step S11: Obtain a filtered image. Specifically, a set of filters with different directional characteristics is used. The filter contains a Gaussian function with zero mean. By changing the variances of the filter in the horizontal and vertical directions, the directional characteristics of the filter are adjusted. The set of filters is respectively applied to the wood cross - section image to obtain a set of filtered images, and the filtered images are used to highlight the textures corresponding to the directional characteristics of the filters in the wood cross - section image;

[0050] Step S12: Reduce noise. Specifically, the majority voting method is used to smooth the filtered images;

[0051] Step S13: Calculate the response value. Specifically, all pixel points in the filtered image are traversed. The pixel values of the neighboring pixel points around each pixel point are respectively convolved with all filters to obtain the response value of each pixel point under the corresponding filter. The direction of the filter with the maximum response value is recorded as the dominant direction of the pixel point, and the dominant directions of all pixel points are recorded to obtain a direction map;

[0052] Step S2: Quality assessment, which is used to construct a local quality map. The specific operations are as follows;

[0053] Step S21: Eliminate discontinuities, which is used to eliminate the discontinuities caused by the direction angles of 0° and 180°. Specifically, the local direction consistency between each pixel point in the direction map and its neighboring pixel points is calculated. The smaller the local direction consistency, the more consistent the direction of the pixel point with its neighboring pixel points, and the clearer the texture of the annual ring. The calculation formula of the local direction consistency is as follows:

[0054] ;

[0055] In the formula, represents the local direction consistency, represents the dominant direction of each pixel point in the direction map, represents the direction map rotated by 90°, represents the operation of calculating the standard deviation, represents the operation of taking the minimum value;

[0056] Step S22: Calculate the local direction matching degree for evaluating the wood texture. Specifically, using a computer vision algorithm, obtain the connection line from each pixel point to the wood pith, and calculate the local direction matching degree between the connection line and the dominant direction of the pixel point in the direction map. The smaller the local direction matching degree, the more the direction of the pixel point matches the texture direction of the annual ring. The calculation formula of the local direction matching degree is as follows:

[0057] ;

[0058] In the formula, represents the local direction matching degree, represents the direction of the connection line, represents the direction of the connection line after rotating 90°, represents the modulo operation;

[0059] Step S23: Calculate the local quality index for evaluating the annual ring visibility and regional quality. Specifically, perform an element-wise multiplication operation on the local direction consistency and the local direction matching degree of all pixel points to obtain the local quality index. The higher the local quality index, the higher the weight of the pixel point for wood species identification. According to the local quality index, construct a local quality map;

[0060] Step S3: Species output, which is used to identify the wood species using a convolutional neural network. Specifically, use the local quality index in the local quality map as the weight of the wood cross-sectional image, and use the local quality map as an additional channel to be embedded into the wood cross-sectional image as input data in the convolutional neural network. The convolutional neural network outputs the wood species.

[0061] Embodiment 3: Refer to Figures 1 to 5 , this embodiment is based on the above embodiment. The positioning correction module uses a positioning correction method, and the positioning correction method specifically includes the following steps:

[0062] Step M1: Determine the shear strength. Specifically, according to the wood species, determine the wood density and the initial shear strength;

[0063] Step M2: Determine the internal structure. Specifically, use a laser scanner to obtain the wood surface point cloud data, and based on wood science knowledge, reconstruct the wood internal texture structure;

[0064] Step M3: Determine the stiffness coefficient. Specifically, use a stress wave propagation device to measure the acoustic wave propagation speed of the stress wave in the wood, and calculate the wood stiffness coefficient. The formula used is as follows:

[0065] ;

[0066] In the formula, represents the stiffness coefficient, represents the acoustic wave propagation speed, represents the density of the wood;

[0067] Step M4: Stiffness correction, which is used to correct the initial shear strength by using a linear regression model. Specifically, the presence of knots will affect the shear strength of the wood. By using computer vision technology, the position and shape of the knots in the cross-sectional image of the wood are identified, the weights of the cross-sectional image of the wood are adjusted according to the position and shape of the knots, and the position and shape of the knots, the initial shear strength, and the local quality map are used as three additional channels and embedded into the internal texture structure and the cross-sectional image of the wood that serve as the input data of the linear regression model. The linear regression model outputs the global shear strength image.

[0068] Example 4: Refer to Figures 1 to 5 , in this example, based on the above example, the wood splitting unit is a cross knife.

[0069] Example 5: Refer to Figures 1 to 5 , in this example, based on the above example, the wood splitting unit is a straight knife.

[0070] Example 6: Refer to Figures 1 to 5 , in this example, based on the above example, in step S11, the set of filters with different direction characteristics is 4 different filter sizes (5x5, 7x7, 9x9, and 11x11), and each size corresponds to 12 different directions (0°, 15°, 30°, 45°, 60°, 75°, 90°, 105°, 120°, 135°, 150°, and 165°), and a total of 48 direction-sensitive zero-mean Gaussian filters are used.

[0071] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0072] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

[0073] The above describes the present invention and its implementation manners. Such description is not restrictive. What is shown in the drawings is only one of the implementation manners of the present invention, and the actual structure is not limited thereto. Generally speaking, if those of ordinary skill in the art are inspired by it and, without departing from the gist of the present invention, design similar structural manners and embodiments to this technical solution without creative efforts, they shall fall within the protection scope of the present invention.

Claims

1. A visual precision positioning system applied to a wood splitter, characterized in that, It includes an image acquisition module, a wood species identification module, a positioning correction module, and a feeding and splitting module; The image acquisition module uses an industrial camera to obtain a cross-sectional image of the wood; The wood species identification module adopts a wood species identification method to analyze the cross-sectional texture based on the cross-sectional image of the wood and identify the wood species; The positioning correction module adopts a positioning correction method to determine the global shear strength image of the wood according to the wood species: The feeding and splitting module includes a splitting unit and a feeding unit; the feeding and splitting module adjusts the cutting angle of the splitting unit and optimizes the feeding speed of the feeding unit according to the global shear strength image; The wood species identification method specifically includes the following steps: Step S1: Cross-section optimization, which is used to extract local structures; Step S2: Quality assessment, which is used to construct a local quality map; Step S3: Species output, which is used to identify the wood species by using a convolutional neural network. Specifically, the local quality index in the local quality map is used as the weight of the cross-sectional image of the wood, and the local quality map is used as an additional channel and embedded into the cross-sectional image of the wood as the input data of the convolutional neural network. The convolutional neural network outputs the wood species; The positioning correction method specifically includes the following steps: Step M1: Determine the shear strength. Specifically, according to the wood species, determine the wood density and the initial shear strength; Step M2: Determine the internal structure. Specifically, use a laser scanner to obtain the surface point cloud data of the wood and reconstruct the internal texture structure of the wood based on wood science knowledge; Step M3: Determine the stiffness coefficient. Specifically, use a stress wave propagation device to measure the acoustic wave propagation speed of the stress wave in the wood and calculate the wood stiffness coefficient. The formula used is as follows: ； In the formula, represents the stiffness coefficient, represents the acoustic wave propagation velocity, represents the density of the wood; Step M4: Stiffness correction, which is used to correct the initial shear strength by using a linear regression model. Specifically, the presence of knots will affect the shear strength of the wood. Use computer vision technology to identify the position and shape of the knots in the cross-sectional image of the wood, adjust the weight of the cross-sectional image of the wood according to the position and shape of the knots, and use the position and shape of the knots, the initial shear strength, and the local quality map as three additional channels and embed them into the internal texture structure and cross-sectional image of the wood as the input data of the linear regression model. The linear regression model outputs the global shear strength image.

2. The visual accuracy positioning system applied to a wood splitter according to claim 1, wherein: In step S1, the cross-section optimization specifically includes the following steps: Step S11: Obtain a filtered image. Specifically, use a set of filters with different directional characteristics. The filter contains a Gaussian function with a zero mean. Adjust the directional characteristics of the filter by changing the variances of the filter in the horizontal and vertical directions. Apply this set of filters to the cross-sectional image of the wood respectively to obtain a set of filtered images. The filtered images are used to highlight the texture corresponding to the directional characteristics of the filter in the cross-sectional image of the wood; Step S12: Reduce noise. Specifically, use the majority voting method to smooth the filtered images; Step S13: Calculate the response value. Specifically, traverse all the pixel points in the filtered image, perform convolution operations on the pixel values of the neighboring pixel points around each pixel point with all the filters respectively, obtain the response value of each pixel point under the corresponding filter, record the direction of the filter with the maximum response value as the dominant direction of the pixel point, and record the dominant directions of all pixel points to obtain the direction map.

3. The visual precision positioning system applied to a log splitter according to claim 2, wherein: In step S2, the quality assessment specifically includes the following steps: Step S21: Eliminate discontinuity, which is used to eliminate the discontinuity caused by the direction angles of 0° and 180°. Specifically, calculate the local direction consistency between each pixel point in the direction map and the neighboring pixel points around it. The smaller the local direction consistency, the more consistent the direction of the pixel point is with the neighboring pixel points around it, and the clearer the texture of the tree rings. The calculation formula of the local direction consistency is as follows: ； In the formula, represents local direction consistency, represents the dominant direction of each pixel in the direction diagram, represents the direction diagram rotated by 90°, represents the operation of calculating the standard deviation, represents the operation of taking the minimum value; Step S22: Calculate the local direction matching degree, which is used to evaluate the wood texture. Specifically, use a computer vision algorithm to obtain the connection line from each pixel point to the pith of the wood, and calculate the local direction matching degree between the connection line and the dominant direction of the pixel point in the direction map. The smaller the local direction matching degree, the more consistent the direction of the pixel point is with the texture direction of the tree rings; Step S23: Calculate the local quality index, which is used to evaluate the visibility of the tree rings and the regional quality. Specifically, perform element-wise multiplication on the local direction consistency and the local direction matching degree of all pixel points to obtain the local quality index. The higher the local quality index, the higher the weight of the pixel point for identifying the wood species. Construct a local quality map according to the local quality index.

Images