A method and system for detecting wooden boards based on visual recognition

Abstract

The embodiment of the present invention relates to the technical field of board material detection, and discloses a method for detecting wooden boards based on visual recognition, including: acquiring a first detection image set at a first detection platform; performing image enhancement processing on the first grayscale image to obtain an enhanced detection image; determining contour image information of each wooden board in the enhanced detection image; performing image recognition processing on the contour image information of each wooden board through a contour recognition algorithm to determine contour feature parameters of the contour image information of each wooden board; inputting the contour image information of each wooden board in the enhanced detection image into a pre-built board defect recognition model for recognition to obtain infrared defect recognition results of each wooden board. The method for detecting wooden boards based on visual recognition disclosed in the embodiment of the present invention realizes efficient, accurate and automated detection of wooden boards through multiple steps such as infrared detection, image preprocessing, contour recognition and defect detection.

Description

Technical Field

[0001] The present invention relates to the technical field of board detection, and in particular to a method and system for detecting a board based on visual recognition. Background Art

[0002] At present, in the custom furniture industry, panel furniture is assembled using panels of various sizes. The size of the panels is one of the most critical quality indicators. If the size of the panels is too large, it will cause interference between the components and make them unable to be assembled. If the size is too small, the gaps between the components will be too large, affecting the appearance. The existing size detection equipment is mainly divided into two types: through-card size detection, the camera is fixed, and the panels are detected during stable operation. The through-type detection requires extremely high stability of the mechanical conveying system, the mechanical cost is very high and the area is large, and the conveying platform requires more than 2 times the length of the panel; fixed size detection, the panels are conveyed or placed in the equipment, and the size is detected by moving the camera to scan the panels, or moving the camera to the four corners of the panels to take pictures, and the detection speed is relatively slow. Therefore, designing a solution that can efficiently detect wooden boards has become a technical problem that technicians in this field need to solve urgently. Summary of the invention

[0003] In view of the above defects, an embodiment of the present invention discloses a wooden board detection method based on visual recognition, which can realize efficient detection of wooden boards and improve the overall detection efficiency.

[0004] A first aspect of an embodiment of the present invention discloses a method for detecting a wooden board based on visual recognition, comprising:

[0005] Acquire a first detection image disposed at a first detection platform by an infrared camera, wherein the first detection image includes a wood board image and a first image resolution; and perform a grayscale value change on the first detection image information to obtain a first grayscale image;

[0006] Performing image enhancement processing on the first grayscale image to obtain an enhanced detection image; and inputting the enhanced detection image into an image recognition model to perform image recognition processing to determine contour image information of each wood board in the enhanced detection image;

[0007] Performing image recognition processing on the contour image information of each wood board material by a contour recognition algorithm to determine contour feature parameters of the contour image information of each wood board material, wherein the contour feature parameters include contour size information;

[0008] The contour image information of each wooden board in the enhanced detection image is input into the pre-built board defect recognition model for recognition to obtain the infrared defect recognition results of each wooden board; and the corresponding infrared defect recognition results are output.

[0009] As an optional implementation, in the first aspect of the embodiment of the present invention, the wooden board detection direction is applied to a board detection system, the board detection system includes an infrared detection system and a visible light detection system; the infrared detection system includes an infrared camera, a first detection frame and a first detection platform, the first detection platform and the infrared camera are both arranged at the first detection frame, and the infrared camera is used to obtain infrared image information at the first detection platform; the visible light detection system includes a visible light camera, a second detection frame and a second detection platform; the second detection platform and the visible light camera are both arranged at the second detection frame, and the visible light camera is used to obtain visible light image information at the second detection platform;

[0010] The wooden board detection method further comprises:

[0011] Acquire a second detection image disposed at a second detection platform by a visible light camera, wherein the second detection image includes a wood board image and a second image resolution, and the first image resolution and the second image resolution match;

[0012] Input the second detection image into a pre-built appearance defect recognition model for recognition to obtain appearance defect recognition results of each wood board, wherein the appearance defect recognition results include defect type and defect position, wherein the defect position is the length information of the defect center point from each corner point;

[0013] The final defect result is determined according to the infrared defect recognition results and appearance defect recognition results of each wood board, and the final defect result is output.

[0014] As an optional implementation, in the first aspect of the embodiment of the present invention, the step of inputting the second detection image into a pre-built appearance defect recognition model for recognition to obtain appearance defect recognition results of each wood board material includes:

[0015] Inputting the second detection image into the constructed appearance defect recognition model to obtain a preliminary recognition result, wherein the preliminary recognition result includes the type of defect and the confidence level of the defect; determining all defects whose confidence level is greater than a preset confidence level threshold value of the corresponding defect type;

[0016] The final defect result is determined based on the infrared defect recognition result and the appearance defect recognition of each wood board, including:

[0017] Determine the prior probability of corresponding defect types in infrared defect recognition results and appearance defect recognition results;

[0018] Calculate the conditional probability of infrared defect recognition results and appearance defect recognition results;

[0019] The prior probabilities of corresponding defect types in the infrared defect recognition results and the appearance defect recognition results and the conditional probabilities of the infrared defect recognition results and the appearance defect recognition results are input into the Bayesian model to determine the posterior probability results of various defects, and the posterior probability results are output.

[0020] As an optional implementation manner, in the first aspect of the embodiment of the present invention, performing image enhancement processing on the first grayscale image to obtain an enhanced detection image includes:

[0021] Acquire a first grayscale image and a second grayscale image, and perform Gaussian filtering on the acquired first grayscale image and the second grayscale image through a Gaussian filter to obtain a filtered grayscale image;

[0022] Performing a downsampling operation on the filtered grayscale image according to a set downsampling rule to obtain a corresponding downsampled image;

[0023] The downsampled image obtained above is used as an image input to perform low-pass filtering and downsampling again until a set first sampling layer number is reached to obtain a first processed image set including multiple layers of resolution and gradually decreasing resolution; the first processed image set includes multiple layers of first processed images;

[0024] Performing upsampling processing on each layer in the first processed image set to obtain an upsampled image with its original resolution restored; the upsampling is processed by bilinear interpolation;

[0025] Subtracting pixels of the upsampled image from the next layer of images in the first processed image set to obtain a corresponding layer of a second processed image set; the second processed image set includes multiple layers of second processed images;

[0026] For each pixel position, compare the maximum absolute value of the parameters at the corresponding position in the first grayscale image and the second grayscale image as the fused parameter, and traverse each layer in the second processed image set to obtain a fused second processed image set;

[0027] Perform image reconstruction from the top layer of the fused second processed image set downwards layer by layer;

[0028] The fused Laplacian coefficients of the current layer are upsampled to match the resolution of the next layer of image; and the upsampled image is combined with the fused Gaussian pyramid image of the next layer; the above operation is performed on each layer until a complete fused image is reconstructed to obtain an enhanced detection image.

[0029] As an optional implementation, in the first aspect of the embodiment of the present invention, the image recognition processing of the contour image information of each wooden board material by using the contour recognition algorithm to determine the contour feature parameters of the contour image information of each wooden board material includes:

[0030] The gradient operator is used to calculate the gradient of the contour image information of each wooden board to obtain the gradient information of the image in the x direction and the y direction;

[0031] Construct the corner point response matrix of each pixel based on the obtained gradient information;

[0032] According to the corner point response matrix of each pixel point, its corner point response value is calculated, and the calculated corner point response value is thresholded, and non-maximum suppression is performed on the corner points after thresholding to eliminate redundant corner points;

[0033] If the corner point response value exceeds the first set threshold, the corresponding pixel point is determined as a corner point, and the corner points of the corresponding rectangular contour image are determined by calculating the angle between the connecting lines of adjacent corner points; the contour feature parameters of the contour image information of each wooden board are determined according to the corner points of the rectangular contour image;

[0034] Or, the gradient operator is used to calculate the gradient of the contour image information of each wooden board to obtain the gradient of the image in the x direction and the y direction;

[0035] Construct the corner point response matrix of each pixel based on the obtained gradient information;

[0036] The corner points are determined by calculating the difference between the two minimum eigenvalues ​​of the corner point response matrix, and the corner points of the corresponding rectangular contour image are determined by calculating the angle between the connecting lines of adjacent corner points; the contour feature parameters of the contour image information of each wooden board are determined based on the corner points of the rectangular contour image.

[0037] As an optional implementation, in the first aspect of the embodiment of the present invention, a heating system is further provided between the infrared detection system and the visible light detection system, and all wooden boards pass through the visible light system, the heating system and the infrared detection system in sequence;

[0038] Before acquiring the first detection image disposed at the first detection platform by the infrared camera, the method further includes:

[0039] The heating operation of the wood board is realized by the heating component in the heating system;

[0040] The number of visible light cameras in the visible light detection system is at least two, and the number of infrared cameras in the infrared detection system is at least two;

[0041] The method of acquiring a first detection image disposed on a first detection platform by using an infrared camera includes:

[0042] Acquire a first infrared image group disposed at a first detection platform by using a plurality of infrared cameras disposed at a first detection frame, wherein the first detection image group includes a plurality of first infrared images;

[0043] Acquire edge pixel data of both sides of a plurality of first infrared images and determine whether adjacent first infrared images have overlapping areas, and sequentially perform a splicing operation on the plurality of first infrared images according to the overlapping areas to obtain a first detection image;

[0044] And / or, the infrared detection system includes a feeding auxiliary mechanism, a backlight emitting platform, and a discharging auxiliary mechanism; wherein the feeding auxiliary mechanism includes a first driving motor and a first rubber-coated roller rod, the first driving motor is used to drive the first rubber-coated roller rod to roll, and the discharging auxiliary mechanism includes a second driving motor and a second rubber-coated roller rod; the second driving motor is used to drive the second rubber-coated roller rod to roll;

[0045] The backlight luminous platform includes tempered glass, a luminous lamp group and a light-distributing layer, wherein the light-distributing layer is located between the tempered glass and the luminous lamp group, and the luminous lamp group, the first driving motor and the second motor are all electrically connected to the industrial control machine.

[0046] As an optional implementation manner, in the first aspect of the embodiment of the present invention, the infrared camera and the visible light camera are calibrated by the following steps:

[0047] The calibration template is placed on the detection plane, and a plurality of template images are taken by the infrared camera or visible light camera to be calibrated, and the pixel points in the first direction and the second direction in the image are detected;

[0048] Determine a first detection parameter and a second detection parameter of a corresponding calibration template according to the association between the first direction pixel points and the second direction pixel points and the pixel-length mapping;

[0049] The first detection parameter is matched with the first calibration parameter of the calibration template to determine the first correction parameter, and the second detection parameter is matched with the second calibration parameter of the calibration template to determine the second correction parameter; and the first correction parameter and the second correction parameter are saved.

[0050] A second aspect of an embodiment of the present invention discloses a wooden board detection system based on visual recognition, comprising:

[0051] An acquisition module is used to acquire a first detection image provided at a first detection platform through an infrared camera, wherein the first detection image includes a wood board image and a first image resolution; and to perform a gray value change on the first detection image information to obtain a first grayscale image;

[0052] Enhancement processing module: used for performing image enhancement processing on the first grayscale image to obtain an enhanced detection image; and inputting the enhanced detection image into an image recognition model to perform image recognition processing to determine the contour image information of each wood board in the enhanced detection image;

[0053] Parameter determination module: used to perform image recognition processing on the contour image information of each wood board material through a contour recognition algorithm to determine the contour feature parameters of the contour image information of each wood board material, wherein the contour feature parameters include contour size information;

[0054] Defect recognition module: used to input the contour image information of each wooden board in the enhanced detection image into the pre-built board defect recognition model for recognition to obtain the infrared defect recognition results of each wooden board; and output the corresponding infrared defect recognition results.

[0055] The third aspect of an embodiment of the present invention discloses an electronic device, comprising: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the wooden board detection method based on visual recognition disclosed in the first aspect of the embodiment of the present invention.

[0056] A fourth aspect of an embodiment of the present invention discloses a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the wooden board detection method based on visual recognition disclosed in the first aspect of an embodiment of the present invention.

[0057] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0058] The wood board detection method based on visual recognition in the embodiment of the present invention obtains the image of the wood board through an infrared camera. This method can utilize the characteristics of infrared rays to effectively capture the temperature distribution differences on the surface of the wood board. This is particularly effective for detecting defects inside the wood board (such as cracks, decay, etc.), because these defects often lead to local temperature anomalies.

[0059] By processing the enhanced detection image with the image recognition model, the contour image information of each wood board can be accurately identified. Further using the contour recognition algorithm, contour feature parameters such as contour size information can be extracted, which is crucial for specification detection and quality control of wood board. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0061] Figure 1 is a flow chart of a method for detecting a wooden board based on visual recognition disclosed in an embodiment of the present invention;

[0062] Figure 2 is a schematic diagram of the process of appearance defect recognition disclosed in an embodiment of the present invention;

[0063] Figure 3 is a schematic diagram of the process of image enhancement disclosed in an embodiment of the present invention;

[0064] Figure 4 It is a structural schematic diagram of an infrared detection system disclosed in an embodiment of the present invention;

[0065] Figure 5 is a structural schematic diagram of a wooden board detection system based on visual recognition provided by an embodiment of the present invention;

[0066] Figure 6 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention.

[0067] Figure numerals: 1. infrared camera; 2. backlight emitting platform; 3. feeding auxiliary mechanism; 4. discharging auxiliary mechanism; 5. first detection frame. DETAILED DESCRIPTION

[0068] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0069] It should be noted that the terms "first", "second", "third", "fourth", etc. in the specification and claims of the present invention are used to distinguish different objects rather than to describe a specific order. The terms "including" and "having" in the embodiments of the present invention and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device including a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0070] In the custom furniture industry, panel furniture is assembled using panels of various sizes. The size of the panels is one of the most critical quality indicators. If the size of the panels is too large, interference between the components will occur, resulting in failure to assemble. If the size is too small, the gaps between the components will be too large, affecting the appearance. Existing size detection equipment is mainly divided into two types: through-card size detection, the camera is fixed, and the panels are detected during stable operation. The through-type detection requires extremely high stability of the mechanical conveying system, the mechanical cost is very high and the area is large, and the conveying platform requires more than 2 times the length of the panel; fixed size detection, the panel is conveyed or placed in the equipment, and the size is detected by moving the camera to scan the panel, or moving the camera to the four corners of the panel to take pictures, and the detection speed is relatively slow. Based on this, the embodiment of the present invention discloses a wooden board detection method, system, electronic device and storage medium based on visual recognition, and obtains the image of the wooden board through an infrared camera. This method can use the characteristics of infrared rays to effectively capture the temperature distribution difference on the surface of the wooden board, which is particularly effective for detecting defects inside the wooden board (such as cracks, decay, etc.), because these defects often lead to local temperature abnormalities.

[0071] By processing the enhanced detection image with the image recognition model, the contour image information of each wood board can be accurately identified. Further using the contour recognition algorithm, contour feature parameters such as contour size information can be extracted, which is crucial for specification detection and quality control of wood board.

[0072] Embodiment 1

[0073] See also Figure 1 , Figure 1 It is a flow chart of the method for detecting wooden boards based on visual recognition disclosed in the embodiment of the present invention. Among them, the execution subject of the method described in the embodiment of the present invention is an execution subject composed of software and / or hardware, and the execution subject can receive relevant information by wired or / and wireless means, and can send certain instructions. Of course, it can also have certain processing functions and storage functions. The execution subject can control multiple devices, such as a remote physical server or cloud server and related software, or it can be a local host or server and related software that performs related operations on a device placed somewhere. In some scenarios, multiple storage devices can also be controlled, and the storage devices can be placed in the same place or different places as the devices. For example Figure 1 As shown, the wooden board detection method based on visual recognition includes the following steps:

[0074] S101: acquiring a first detection image disposed at a first detection platform by an infrared camera, wherein the first detection image includes a wood board image and a first image resolution; and performing a grayscale value change on the first detection image information to obtain a first grayscale image;

[0075] S102: performing image enhancement processing on the first grayscale image to obtain an enhanced detection image; and inputting the enhanced detection image into an image recognition model to perform image recognition processing to determine contour image information of each wood board in the enhanced detection image;

[0076] S103: performing image recognition processing on the contour image information of each wood board material by a contour recognition algorithm to determine contour feature parameters of the contour image information of each wood board material, wherein the contour feature parameters include contour size information;

[0077] S104: inputting the contour image information of each wooden board in the enhanced detection image into a pre-built board defect recognition model for recognition to obtain infrared defect recognition results of each wooden board; and outputting the corresponding infrared defect recognition results.

[0078] The solution of the embodiment of the present invention performs grayscale value change processing on the acquired first detection image to obtain a first grayscale image, and then performs image enhancement processing on it, which can significantly improve the contrast between the wooden board and the background in the image, making subsequent image recognition processing more accurate.

[0079] By inputting the contour image information into the pre-built board defect recognition model, the infrared defects of each wooden board can be automatically identified. This method not only improves the detection efficiency, but also reduces the misjudgment and missed judgment caused by human factors, ensuring the accuracy and reliability of the detection results. The system can output the infrared defect recognition results of each wooden board, which provides an important reference for the subsequent classification, processing and use of wooden boards.

[0080] More preferably, the wood board detection direction is applied to a board detection system, and the board detection system includes an infrared detection system and a visible light detection system; the infrared detection system includes an infrared camera 1, a first detection frame 5 and a first detection platform, the first detection platform and the infrared camera 1 are both arranged at the first detection frame 5, and the infrared camera is used to obtain infrared image information at the first detection platform; the visible light detection system includes a visible light camera, a second detection frame and a second detection platform; the second detection platform and the visible light camera are both arranged at the second detection frame, and the visible light camera is used to obtain visible light image information at the second detection platform;

[0081] like Figure 2 As shown, the wooden board detection method further includes:

[0082] S100a: acquiring a second detection image disposed at a second detection platform by a visible light camera, wherein the second detection image includes a wooden board image and a second image resolution, and the first image resolution and the second image resolution match;

[0083] S100b: Input the second detection image into a pre-built appearance defect recognition model for recognition to obtain appearance defect recognition results of each wood board, wherein the appearance defect recognition results include defect type and defect position, wherein the defect position is the length information of the defect center point from each corner point;

[0084] S100c: Determine a final defect result based on the infrared defect recognition results and appearance defect recognition results of each wood board, and output the final defect result.

[0085] The above scheme realizes all-round and multi-angle detection of wooden boards by combining infrared detection system and visible light detection system. The infrared detection system can capture the defects inside the wooden board, while the visible light detection system can identify the appearance defects on the surface of the wooden board. The two complement each other and greatly improve the accuracy and comprehensiveness of the detection.

[0086] In the embodiment of the present invention, it is particularly pointed out that the first image resolution matches the second image resolution, which ensures the consistency and comparability of the infrared image and the visible light image in subsequent processing, and facilitates subsequent image fusion and defect analysis. In specific implementation, the resolutions can be completely the same or proportional, which can make the final detection result and the subsequent image fusion more convenient.

[0087] The second detection image obtained by the visible light camera is input into the pre-built appearance defect recognition model for recognition, which can accurately identify the appearance defects of the wood board, including the defect type and defect location. The accurate description of the defect location (such as the length information of the defect center point from each corner point) provides an important basis for subsequent defect repair and quality control. According to the infrared defect recognition results and the appearance defect recognition results, the final defect result of the wood board is comprehensively determined. This method not only considers the defects inside the wood board, but also the appearance defects on the surface, making the determination result more comprehensive and accurate. Through the automated detection system and recognition model, the detection time is greatly shortened and the detection efficiency is improved. At the same time, the misjudgment and missed judgment caused by human factors are reduced, ensuring the accuracy and reliability of the detection results.

[0088] More preferably, the step of inputting the second detection image into a pre-built appearance defect recognition model for recognition to obtain appearance defect recognition results of each wood board material includes:

[0089] Inputting the second detection image into the constructed appearance defect recognition model to obtain a preliminary recognition result, wherein the preliminary recognition result includes the type of defect and the confidence level of the defect; determining all defects whose confidence level is greater than a preset confidence level threshold value of the corresponding defect type;

[0090] By inputting the second inspection image into the appearance defect recognition model, a preliminary recognition result can be obtained, including the type of defect and the confidence level of the defect. This helps to screen out defects with higher confidence levels and reduce false positives.

[0091] The final defect result is determined based on the infrared defect recognition result and the appearance defect recognition of each wood board, including:

[0092] S100c1: determining the prior probability of corresponding defect types in the infrared defect recognition results and the appearance defect recognition results;

[0093] S100c2: Calculate the conditional probability of infrared defect recognition results and appearance defect recognition results;

[0094] S100c3: Input the prior probabilities of corresponding defect types in the infrared defect recognition results and the appearance defect recognition results and the conditional probabilities of the infrared defect recognition results and the appearance defect recognition results into the Bayesian model to determine the posterior probability results of each type of defect, and output the posterior probability results.

[0095] The Bayesian model is used to fuse the infrared defect recognition results and the appearance defect recognition results, and the posterior probability results of various defects are determined by calculating the prior probability and conditional probability. This method can comprehensively consider multiple information sources and improve the accuracy and reliability of defect recognition. According to the posterior probability results output by the Bayesian model, the final defect results of the wood board can be comprehensively determined. This method not only considers the information of the two detection methods of infrared and visible light, but also fuses and optimizes the information through Bayesian reasoning, making the determination results more comprehensive and accurate. The automated detection system and recognition model greatly improve the detection efficiency and reduce the degree of manual intervention. At the same time, by integrating multiple information sources and optimization algorithms, the accuracy and reliability of detection are improved.

[0096] More preferably, Figure 3 As shown, the performing image enhancement processing on the first grayscale image to obtain an enhanced detection image includes:

[0097] S1021: Obtain a first grayscale image and a second grayscale image, and perform Gaussian filtering on the obtained first grayscale image and the second grayscale image through a Gaussian filter to obtain a filtered grayscale image; first, obtain grayscale versions of two detection images, namely, the first detection image grayscale image and the second detection image grayscale image. Then, use a Gaussian filter to filter the two grayscale images, the purpose of which is to remove noise in the image while retaining important edge and feature information. The filtered image is called a filtered detection image grayscale image.

[0098] S1022: Downsampling the filtered grayscale image according to the set downsampling rule to obtain a corresponding downsampled image; Downsampling the filtered detection image grayscale image according to the set downsampling rule (such as sampling at regular intervals) to obtain an image sequence with gradually decreasing resolution, forming a first processed image set of the Gaussian pyramid. This set includes multiple layers of first processed images, and the resolution of each layer is lower than that of the previous layer.

[0099] S1023: using the downsampled image obtained above as an image input, and performing low-pass filtering and downsampling again until a set first sampling layer number is reached to obtain a first processed image set including multiple layers of resolution and gradually decreasing resolution; the first processed image set includes multiple layers of first processed images;

[0100] S1024: performing upsampling processing on each layer in the first processed image set to obtain an upsampled image with its original resolution restored; the upsampling is processed by bilinear interpolation;

[0101] S1025: Subtract the pixels of the upsampled image from the next layer of images in the first processed image set to obtain the corresponding layer of the second processed image set; the second processed image set includes multiple layers of second processed images; upsample each layer in the first processed image set (through bilinear interpolation) to restore its original resolution. Subtract the pixels of the upsampled image from the next layer of images in the first processed image set to obtain the corresponding layer of the second processed image set. The images in this set represent the detailed information of the image at different resolutions, i.e., the Laplacian pyramid.

[0102] The scheme of the embodiment of the present invention is based on the difference between the current layer image and the previous layer image after upsampling (i.e., resolution improvement) in the first processed image set; each layer of the Laplacian pyramid is the difference between the current layer image and the previous layer image in the Gaussian pyramid after upsampling (i.e., resolution improvement); each layer in the Gaussian pyramid (except the top layer) is upsampled to restore its original resolution; upsampling can be achieved by an interpolation method, such as bilinear interpolation or bicubic interpolation; the upsampled image is subtracted pixel by pixel from the next layer image in the Gaussian pyramid to obtain the corresponding layer of the Laplacian pyramid.

[0103] S1026: For each pixel position, compare the maximum absolute value of the parameter at the corresponding position in the first grayscale image and the second grayscale image as the fused parameter, traverse each layer in the second processed image set to obtain the fused second processed image set; for each layer in the second processed image set, compare the maximum absolute value of the parameter at the corresponding position in the grayscale image of the first detection image and the grayscale image of the second detection image, and use this as the fused parameter. In this way, each layer will obtain a fused image, forming a fused second processed image set.

[0104] S1027: reconstructing images layer by layer starting from the top layer of the fused second processed image set;

[0105] S1028: Up-sampling the fused Laplacian coefficients of the current layer to match the resolution of the image of the next layer; and combining the up-sampled image with the fused Gaussian pyramid image of the next layer; performing the above operation on each layer until a complete fused image is reconstructed to obtain an enhanced detection image. The Laplacian coefficient contains the high-frequency detail information of the image, so the images of different spectra or different perspectives can be decomposed into Laplacian pyramids, and then the fusion rules (such as the maximum absolute value fusion rule) are applied on each layer to obtain the fused Laplacian coefficients. Finally, by reconstructing the Laplacian pyramid layer by layer, a high-quality image that integrates multiple information can be obtained. Starting from the top layer (i.e., the layer with the highest resolution) of the fused second processed image set, image reconstruction is performed layer by layer.

[0106] Upsample the fused Laplacian coefficients of the current layer to match the resolution of the next layer. Merge the upsampled image with the fused Gaussian pyramid image of the next layer (actually obtained by adding the upsampling of the previous step and the Laplacian coefficients of the current layer). Perform the above operation for each layer until the complete fused image is reconstructed.

[0107] The solution of the embodiment of the present invention fuses the Laplacian pyramid images of the visible light image and the infrared image to obtain fused Laplacian pyramid images of each layer; the fused Laplacian images of each layer are iterated layer by layer from the top layer to the bottom layer to obtain the final fused image.

[0108] More preferably, the image recognition processing of the contour image information of each wood board material by using a contour recognition algorithm to determine the contour feature parameters of the contour image information of each wood board material includes:

[0109] The gradient operator is used to calculate the gradient of the contour image information of each wooden board to obtain the gradient information of the image in the x direction and the y direction;

[0110] Construct the corner point response matrix of each pixel based on the obtained gradient information;

[0111] According to the corner point response matrix of each pixel point, its corner point response value is calculated, and the calculated corner point response value is thresholded, and non-maximum suppression is performed on the corner points after thresholding to eliminate redundant corner points;

[0112] If the corner point response value exceeds the first set threshold, the corresponding pixel point is determined as a corner point, and the corner points of the corresponding rectangular contour image are determined by calculating the angle between the connecting lines of adjacent corner points; the contour feature parameters of the contour image information of each wooden board are determined according to the corner points of the rectangular contour image;

[0113] Or, the gradient operator is used to calculate the gradient of the contour image information of each wooden board to obtain the gradient of the image in the x direction and the y direction;

[0114] Construct the corner point response matrix of each pixel based on the obtained gradient information;

[0115] The corner points are determined by calculating the difference between the two minimum eigenvalues ​​of the corner point response matrix, and the corner points of the corresponding rectangular contour image are determined by calculating the angle between the connecting lines of adjacent corner points; the contour feature parameters of the contour image information of each wooden board are determined based on the corner points of the rectangular contour image.

[0116] The solution of the embodiment of the present invention calculates the gradient information of the image in the x-direction and the y-direction through the gradient operator, which can accurately capture the edge and contour changes in the image and provide accurate basic data for subsequent corner point detection and contour feature extraction.

[0117] Constructing the corner point response matrix of each pixel and calculating the corner point response value can effectively identify the corner points in the image. These corner points are usually the key feature points of the contour and are essential for determining the shape and size of the contour. Thresholding and non-maximum suppression of the corner point response values ​​can eliminate redundant corner points and improve the accuracy and efficiency of corner point detection.

[0118] The solution of the embodiment of the present invention can accurately determine the corner points of the rectangular contour image by calculating the angle between the connecting lines of adjacent corner points, and then determine the contour feature parameters of the wooden board material, such as length, width, angle, etc., based on these corner points. This method is not only applicable to wooden boards with regular shapes, but can also adapt to irregular shapes to a certain extent, thereby improving the versatility and accuracy of contour recognition.

[0119] This method provides two ways to determine corner points: one is based on thresholding and non-maximum suppression of corner point response values; the other is based on the difference between the two minimum eigenvalues ​​of the corner point response matrix. These two methods can be flexibly selected according to specific circumstances to adapt to different image quality and complexity. At the same time, this method is highly robust to noise and interference in the image, and can overcome problems such as poor image quality or uneven illumination to a certain extent, ensuring the accuracy and stability of contour recognition.

[0120] More preferably, a heating system is provided between the infrared detection system and the visible light detection system, and all the wood panels pass through the visible light system, the heating system and the infrared detection system in sequence;

[0121] Before acquiring the first detection image disposed at the first detection platform by the infrared camera, the method further includes:

[0122] The heating operation of the wood board is realized by the heating component in the heating system;

[0123] The number of visible light cameras in the visible light detection system is at least two, and the number of infrared cameras in the infrared detection system is at least two;

[0124] The method of acquiring a first detection image disposed on a first detection platform by using an infrared camera includes:

[0125] Acquire a first infrared image group disposed at a first detection platform by using a plurality of infrared cameras disposed at a first detection frame, wherein the first detection image group includes a plurality of first infrared images;

[0126] Acquire edge pixel data of both sides of a plurality of first infrared images and determine whether adjacent first infrared images have overlapping areas, and sequentially perform a splicing operation on the plurality of first infrared images according to the overlapping areas to obtain a first detection image;

[0127] And / or, the infrared detection system includes a feeding auxiliary mechanism 3, a backlight emitting platform 2, and a discharging auxiliary mechanism 4; wherein the feeding auxiliary mechanism 3 includes a first driving motor and a first rubber-coated roller rod, the first driving motor is used to drive the first rubber-coated roller rod to roll, and the discharging auxiliary mechanism 4 includes a second driving motor and a second rubber-coated roller rod; the second driving motor is used to drive the second rubber-coated roller rod to roll;

[0128] The backlight luminous platform includes tempered glass, a luminous lamp group and a light-distributing layer, wherein the light-distributing layer is located between the tempered glass and the luminous lamp group, and the luminous lamp group, the first driving motor and the second motor are all electrically connected to the industrial control machine.

[0129] The solution of the embodiment of the present invention can detect the wood board from two angles, visible light and infrared, respectively, by setting up a visible light detection system and an infrared detection system, so as to obtain more comprehensive board information. The visible light detection system can capture the details and texture of the board surface, while the infrared detection system can detect the temperature distribution and defects inside the board, such as cracks and cavities.

[0130] A heating system is set between the visible light detection system and the infrared detection system to heat the wood board, which can make the defects inside the board (such as cracks and voids) more obvious in the infrared image, thereby improving the accuracy and sensitivity of infrared detection. The main function of the heating system is to preheat or heat the wood board. By increasing the temperature of the board, the defects inside the board (such as cracks, voids, uneven moisture content, etc.) can change in physical or chemical properties, so that they show more obvious features in the infrared image. Infrared detection is based on the difference in temperature distribution on the surface or inside of an object. When the wood board is heated, the defective area inside it will have a temperature difference with the normal area. The infrared camera can capture these temperature differences and convert them into a visual infrared image. In the detection process, the wood board first passes through the visible light detection system to capture the details and texture information of its surface. Then, the board enters the heating system for preheating or heating. Finally, the heated board enters the infrared detection system, and the infrared camera captures the temperature distribution information inside it, thereby realizing the detection of internal defects of the board.

[0131] Through the heating process, the defects inside the plate appear more obvious in the infrared image, which enables the infrared detection system to more accurately identify the location and size of the defects. Compared with the visible light detection system, infrared detection can provide more in-depth and comprehensive information about the plate.

[0132] The heating system enables even tiny defects inside the plate to be captured by the infrared detection system. This increases the sensitivity of the detection, allowing the system to find smaller defects, thereby reducing the missed detection rate.

[0133] This solution is suitable for the detection of wood panels of different types and specifications. By adjusting the parameters of the heating system (such as heating temperature, heating time, etc.), it can adapt to the characteristics and detection requirements of different panels. The implementation of this solution has promoted the innovation and development of wood detection technology. By combining multiple technical means such as visible light detection, infrared detection and heating treatment, it provides the wood industry with a more advanced and reliable detection method.

[0134] At least two cameras are set in both the visible light detection system and the infrared detection system. By acquiring and stitching multiple images, a larger range and higher resolution detection image can be formed. This stitching technology not only increases the image coverage area, but also ensures the accuracy and continuity of the image through matching and correction of overlapping areas. The feeding auxiliary mechanism and the discharging auxiliary mechanism realize the automatic feeding and output of the wood board through the cooperation of the driving motor and the rubber-coated roller rod, thereby improving the degree of automation of the detection. At the same time, the design of the backlight luminous platform provides a stable and uniform light source for infrared detection through the combination of tempered glass, luminous lamp group and homogenizing layer, further improving the accuracy and stability of the detection.

[0135] More preferably, the infrared camera and the visible light camera are calibrated by the following steps:

[0136] The calibration template is placed on the detection plane, and a plurality of template images are taken by the infrared camera or visible light camera to be calibrated, and the pixel points in the first direction and the second direction in the image are detected;

[0137] Determine a first detection parameter and a second detection parameter of a corresponding calibration template according to the association between the first direction pixel points and the second direction pixel points and the pixel-length mapping;

[0138] The first detection parameter is matched with the first calibration parameter of the calibration template to determine the first correction parameter, and the second detection parameter is matched with the second calibration parameter of the calibration template to determine the second correction parameter; and the first correction parameter and the second correction parameter are saved.

[0139] Specifically, place a calibration template: First, place a calibration template on the detection plane. This template usually has a known size and shape and is used as a reference for camera calibration.

[0140] Shoot template images: Use the infrared camera or visible light camera to be calibrated to shoot several images of the calibration template. Shooting multiple images is to improve the accuracy and robustness of calibration. In these images, it is necessary to detect the pixels in the first direction (such as the horizontal direction) and the pixels in the second direction (such as the vertical direction). These pixels are usually the edges of the template or the locations of specific marks.

[0141] Determine detection parameters: Determine the first detection parameters and the second detection parameters of the calibration template based on the detected pixels in the first direction and the second direction, and the known pixel-length mapping association (which is usually determined by the physical parameters and optical characteristics of the camera). These detection parameters may include information such as the size, shape, and angle of the template, expressed in the pixel coordinate system.

[0142] Matching calibration parameters and determining correction parameters: The first detection parameters are matched with the first calibration parameters of the calibration template (known, standard parameters) to determine the first correction parameters. Similarly, the second detection parameters are matched with the second calibration parameters of the calibration template to determine the second correction parameters. These correction parameters are used to adjust the camera's image to more accurately reflect the size and shape of the real world.

[0143] Save the correction parameters: Finally, save the determined first correction parameters and second correction parameters. These parameters can be used to correct and calibrate the image when the camera is used for image acquisition and processing in the future.

[0144] The calibration process corrects the camera distortion, which improves the accuracy of the measurement. This is critical for applications that require accurate measurement of the size and shape of wood panels. The calibrated camera is able to capture details and features in the image more accurately, thereby enhancing the quality of the image. This allows the inspection system to have higher performance in identifying board defects, textures, etc. The calibration process ensures that the camera maintains consistent performance under different conditions such as lighting and temperature. This helps to improve the stability and reliability of the inspection system. Once the camera is calibrated and the correction parameters are saved, these parameters can be used directly in subsequent inspection processes without repeating the calibration. This simplifies the operation process and improves the inspection efficiency.

[0145] The wood board detection method based on visual recognition in the embodiment of the present invention obtains the image of the wood board through an infrared camera. This method can utilize the characteristics of infrared rays to effectively capture the temperature distribution differences on the surface of the wood board. This is particularly effective for detecting defects inside the wood board (such as cracks, decay, etc.), because these defects often lead to local temperature anomalies.

[0146] By processing the enhanced detection image with the image recognition model, the contour image information of each wood board can be accurately identified. Further using the contour recognition algorithm, contour feature parameters such as contour size information can be extracted, which is crucial for specification detection and quality control of wood board.

[0147] Embodiment 2

[0148] See also Figure 5 , Figure 5Schematic diagram of the structure of the wooden board detection system based on visual recognition disclosed in the embodiment of the present invention. Figure 5 As shown, the wooden board detection system based on visual recognition may include:

[0149] The acquisition module 21 is used to acquire a first detection image set at the first detection platform through an infrared camera, wherein the first detection image includes a wood board image and a first image resolution; and perform a gray value change on the first detection image information to obtain a first gray image;

[0150] Enhancement processing module 22: used for performing image enhancement processing on the first grayscale image to obtain an enhanced detection image; and inputting the enhanced detection image into an image recognition model to perform image recognition processing to determine the contour image information of each wood board in the enhanced detection image;

[0151] Parameter determination module 23: used to perform image recognition processing on the contour image information of each wood board material through a contour recognition algorithm to determine contour feature parameters of the contour image information of each wood board material, wherein the contour feature parameters include contour size information;

[0152] Defect recognition module 24: used to input the contour image information of each wooden board in the enhanced detection image into the pre-built board defect recognition model for recognition to obtain the infrared defect recognition result of each wooden board; and output the corresponding infrared defect recognition result.

[0153] The wood board detection method based on visual recognition in the embodiment of the present invention obtains the image of the wood board through an infrared camera. This method can utilize the characteristics of infrared rays to effectively capture the temperature distribution differences on the surface of the wood board. This is particularly effective for detecting defects inside the wood board (such as cracks, decay, etc.), because these defects often lead to local temperature anomalies.

[0154] By processing the enhanced detection image with the image recognition model, the contour image information of each wood board can be accurately identified. Further using the contour recognition algorithm, contour feature parameters such as contour size information can be extracted, which is crucial for specification detection and quality control of wood board.

[0155] Embodiment 3

[0156] See also Figure 6 , Figure 6 Schematic diagram of the structure of an electronic device disclosed in an embodiment of the present invention. The electronic device may be a computer, a server, etc. Of course, in certain circumstances, it may also be a smart device such as a mobile phone, a tablet computer, a monitoring terminal, and an image acquisition device with processing functions. Figure 6 As shown, the electronic device may include:

[0157] A memory 510 storing executable program codes;

[0158] a processor 520 coupled to the memory 510;

[0159] The processor 520 calls the executable program code stored in the memory 510 to execute part or all of the steps in the wooden board detection method based on visual recognition in the first embodiment.

[0160] An embodiment of the present invention discloses a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute part or all of the steps in the method for detecting a wooden board based on visual recognition in the first embodiment.

[0161] The embodiment of the present invention further discloses a computer program product, wherein when the computer program product is run on a computer, the computer is enabled to execute part or all of the steps in the method for detecting a wooden board based on visual recognition in the first embodiment.

[0162] An embodiment of the present invention also discloses an application publishing platform, wherein the application publishing platform is used to publish a computer program product, wherein when the computer program product runs on a computer, the computer executes part or all of the steps in the wooden board detection method based on visual recognition in embodiment one.

[0163] In various embodiments of the present invention, it should be understood that the size of the serial numbers of the processes does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0164] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, i.e., they may be located in one place or distributed over multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0165] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0166] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-accessible memory. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a memory and includes several requests to enable a computer device (which can be a personal computer, server or network device, etc., specifically a processor in a computer device) to perform some or all steps of the method described in each embodiment of the present invention.

[0167] In the embodiments provided by the present invention, it should be understood that "B corresponding to A" means that B is associated with A, and B can be determined according to A. However, it should also be understood that determining B according to A does not mean determining B only according to A, and B can also be determined according to A and / or other information.

[0168] A person of ordinary skill in the art may understand that part or all of the steps in the various methods of the embodiments may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, the storage medium including a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically-erasable programmable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0169] The above is a detailed introduction to the wood board detection method, system, electronic device and storage medium based on visual recognition disclosed in the embodiments of the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for general technical personnel in this field, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A method for detecting wooden boards based on visual recognition, characterized in that: The wooden board detection method is applied to a board detection system, which includes an infrared detection system and a visible light detection system; the infrared detection system includes an infrared camera, a first detection frame and a first detection platform, the first detection platform and the infrared camera are both arranged at the first detection frame, and the infrared camera is used to obtain infrared image information at the first detection platform; the visible light detection system includes a visible light camera, a second detection frame and a second detection platform; The second detection platform and the visible light camera are both arranged at the second detection frame, and the visible light camera is used to obtain visible light image information at the second detection platform; including: Acquire a first detection image disposed at a first detection platform by an infrared camera, wherein the first detection image includes a wood board image and a first image resolution; and perform a grayscale value change on the first detection image information to obtain a first grayscale image; Performing image enhancement processing on the first grayscale image to obtain an enhanced detection image; and inputting the enhanced detection image into an image recognition model to perform image recognition processing to determine contour image information of each wood board in the enhanced detection image; The image recognition processing is performed on the contour image information of each wood board material by a contour recognition algorithm to determine the contour feature parameters of the contour image information of each wood board material, and the contour feature parameters include contour size information; the image recognition processing is performed on the contour image information of each wood board material by a contour recognition algorithm to determine the contour feature parameters of the contour image information of each wood board material, including: The gradient operator is used to calculate the gradient of the contour image information of each wooden board to obtain the gradient information of the image in the x direction and the y direction; Construct the corner point response matrix of each pixel based on the obtained gradient information; According to the corner point response matrix of each pixel point, its corner point response value is calculated, and the calculated corner point response value is thresholded, and non-maximum suppression is performed on the corner points after thresholding to eliminate redundant corner points; If the corner point response value exceeds the first set threshold, the corresponding pixel point is determined as a corner point, and the corner points of the corresponding rectangular contour image are determined by calculating the angle between the connecting lines of adjacent corner points; the contour feature parameters of the contour image information of each wooden board are determined according to the corner points of the rectangular contour image; Or, the gradient operator is used to calculate the gradient of the contour image information of each wooden board to obtain the gradient of the image in the x direction and the y direction; Construct the corner point response matrix of each pixel based on the obtained gradient information; The corner points are determined by calculating the difference between the two minimum eigenvalues ​​of the corner point response matrix, and the corner points of the corresponding rectangular contour image are determined by calculating the angle between the connecting lines of the adjacent corner points; the contour feature parameters of the contour image information of each wood board are determined according to the corner points of the rectangular contour image; Input the contour image information of each wood board in the enhanced detection image into the pre-built board defect recognition model for recognition to obtain the infrared defect recognition result of each wood board; and output the corresponding infrared defect recognition result; The wooden board detection method further comprises: Acquire a second detection image disposed at a second detection platform by a visible light camera, wherein the second detection image includes a wood board image and a second image resolution, and the first image resolution and the second image resolution match; Input the second detection image into a pre-built appearance defect recognition model for recognition to obtain appearance defect recognition results of each wood board, wherein the appearance defect recognition results include defect type and defect position, wherein the defect position is the length information of the defect center point from each corner point; The final defect result is determined according to the infrared defect recognition results and appearance defect recognition results of each wood board, and the final defect result is output.

2. The method for detecting a wooden board based on visual recognition according to claim 1, characterized in that: The step of inputting the second detection image into a pre-built appearance defect recognition model for recognition to obtain appearance defect recognition results of each wood board includes: Inputting the second detection image into the constructed appearance defect recognition model to obtain a preliminary recognition result, wherein the preliminary recognition result includes the type of defect and the confidence level of the defect; determining all defects whose confidence level is greater than a preset confidence level threshold value of the corresponding defect type; The final defect result is determined based on the infrared defect recognition result and the appearance defect recognition of each wood board, including: Determine the prior probability of corresponding defect types in infrared defect recognition results and appearance defect recognition results; Calculate the conditional probability of infrared defect recognition results and appearance defect recognition results; The prior probabilities of corresponding defect types in the infrared defect recognition results and the appearance defect recognition results and the conditional probabilities of the infrared defect recognition results and the appearance defect recognition results are input into the Bayesian model to determine the posterior probability results of various defects, and the posterior probability results are output.

3. The method for detecting a wooden board based on visual recognition according to claim 1, characterized in that: The performing image enhancement processing on the first grayscale image to obtain an enhanced detection image includes: Acquire a first grayscale image and a second grayscale image, and perform Gaussian filtering on the acquired first grayscale image and the second grayscale image through a Gaussian filter to obtain a filtered grayscale image; Performing a downsampling operation on the filtered grayscale image according to a set downsampling rule to obtain a corresponding downsampled image; The downsampled image obtained above is used as an image input to perform low-pass filtering and downsampling again until a set first sampling layer number is reached to obtain a first processed image set including multiple layers of resolution and gradually decreasing resolution; the first processed image set includes multiple layers of first processed images; Performing upsampling processing on each layer in the first processed image set to obtain an upsampled image with its original resolution restored; the upsampling is processed by bilinear interpolation; Subtracting pixels of the upsampled image from the next layer of images in the first processed image set to obtain a corresponding layer of a second processed image set; the second processed image set includes multiple layers of second processed images; For each pixel position, compare the maximum absolute value of the parameters at the corresponding position in the first grayscale image and the second grayscale image as the fused parameter, and traverse each layer in the second processed image set to obtain a fused second processed image set; Perform image reconstruction from the top layer of the fused second processed image set downwards layer by layer; The fused Laplacian coefficients of the current layer are upsampled to match the resolution of the next layer of image; and the upsampled image is combined with the fused Gaussian pyramid image of the next layer; the above operation is performed on each layer until a complete fused image is reconstructed to obtain an enhanced detection image.

4. The method for detecting a wooden board based on visual recognition according to claim 1, characterized in that: A heating system is also provided between the infrared detection system and the visible light detection system, and all wood panels pass through the visible light system, the heating system and the infrared detection system in sequence; Before acquiring the first detection image disposed at the first detection platform by the infrared camera, the method further includes: The heating operation of the wood board is realized by the heating component in the heating system; The number of visible light cameras in the visible light detection system is at least two, and the number of infrared cameras in the infrared detection system is at least two; The method of acquiring a first detection image disposed on a first detection platform by using an infrared camera includes: Acquire a first infrared image group disposed at a first detection platform by using a plurality of infrared cameras disposed at a first detection frame, wherein the first infrared image group includes a plurality of first infrared images; Acquire edge pixel data of both sides of a plurality of first infrared images and determine whether adjacent first infrared images have overlapping areas, and sequentially perform a splicing operation on the plurality of first infrared images according to the overlapping areas to obtain a first detection image; And / or, the infrared detection system includes a feeding auxiliary mechanism, a backlight emitting platform, and a discharging auxiliary mechanism; wherein the feeding auxiliary mechanism includes a first driving motor and a first rubber-coated roller rod, the first driving motor is used to drive the first rubber-coated roller rod to roll, and the discharging auxiliary mechanism includes a second driving motor and a second rubber-coated roller rod; the second driving motor is used to drive the second rubber-coated roller rod to roll; The backlight luminous platform includes tempered glass, a luminous lamp group and a light-distributing layer, wherein the light-distributing layer is located between the tempered glass and the luminous lamp group, and the luminous lamp group, the first driving motor and the second motor are all electrically connected to the industrial control machine.

5. The method for detecting a wooden board based on visual recognition according to claim 1, characterized in that: The infrared camera and the visible light camera are calibrated by the following steps: The calibration template is placed on the detection plane, and a plurality of template images are taken by the infrared camera or visible light camera to be calibrated, and the pixel points in the first direction and the second direction in the image are detected; Determine a first detection parameter and a second detection parameter of a corresponding calibration template according to the association between the first direction pixel points and the second direction pixel points and the pixel-length mapping; The first detection parameter is matched with the first calibration parameter of the calibration template to determine the first correction parameter, and the second detection parameter is matched with the second calibration parameter of the calibration template to determine the second correction parameter; and the first correction parameter and the second correction parameter are saved.

6. A wood board detection system based on visual recognition, characterized in that: The wooden board detection system includes an infrared detection system and a visible light detection system; the infrared detection system includes an infrared camera, a first detection frame and a first detection platform, the first detection platform and the infrared camera are both arranged at the first detection frame, and the infrared camera is used to obtain infrared image information at the first detection platform; the visible light detection system includes a visible light camera, a second detection frame and a second detection platform; The second detection platform and the visible light camera are both arranged at the second detection frame, and the visible light camera is used to obtain visible light image information at the second detection platform; including: An acquisition module is used to acquire a first detection image provided at a first detection platform through an infrared camera, wherein the first detection image includes a wood board image and a first image resolution; and to perform a gray value change on the first detection image information to obtain a first grayscale image; Enhancement processing module: used for performing image enhancement processing on the first grayscale image to obtain an enhanced detection image; and inputting the enhanced detection image into an image recognition model to perform image recognition processing to determine the contour image information of each wood board in the enhanced detection image; Parameter determination module: used to perform image recognition processing on the contour image information of each wooden board material through a contour recognition algorithm to determine the contour feature parameters of the contour image information of each wooden board material, and the contour feature parameters include contour size information; the image recognition processing on the contour image information of each wooden board material through a contour recognition algorithm to determine the contour feature parameters of the contour image information of each wooden board material includes: The gradient operator is used to calculate the gradient of the contour image information of each wooden board to obtain the gradient information of the image in the x direction and the y direction; Construct the corner point response matrix of each pixel based on the obtained gradient information; According to the corner point response matrix of each pixel point, its corner point response value is calculated, and the calculated corner point response value is thresholded, and non-maximum suppression is performed on the corner points after thresholding to eliminate redundant corner points; If the corner point response value exceeds the first set threshold, the corresponding pixel point is determined as a corner point, and the corner points of the corresponding rectangular contour image are determined by calculating the angle between the connecting lines of adjacent corner points; the contour feature parameters of the contour image information of each wooden board are determined according to the corner points of the rectangular contour image; Or, the gradient operator is used to calculate the gradient of the contour image information of each wooden board to obtain the gradient of the image in the x direction and the y direction; Construct the corner point response matrix of each pixel based on the obtained gradient information; The corner points are determined by calculating the difference between the two minimum eigenvalues ​​of the corner point response matrix, and the corner points of the corresponding rectangular contour image are determined by calculating the angle between the connecting lines of the adjacent corner points; the contour feature parameters of the contour image information of each wood board are determined according to the corner points of the rectangular contour image; Defect recognition module: used to input the contour image information of each wood board in the enhanced detection image into the pre-built board defect recognition model for recognition to obtain the infrared defect recognition results of each wood board; and output the corresponding infrared defect recognition results; The wooden board detection system further comprises: Acquire a second detection image disposed at a second detection platform by a visible light camera, wherein the second detection image includes a wood board image and a second image resolution, and the first image resolution and the second image resolution match; Input the second detection image into a pre-built appearance defect recognition model for recognition to obtain appearance defect recognition results of each wood board, wherein the appearance defect recognition results include defect type and defect position, wherein the defect position is the length information of the defect center point from each corner point; The final defect result is determined according to the infrared defect recognition results and appearance defect recognition results of each wood board, and the final defect result is output.

7. An electronic device, characterized in that: include: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the wood board detection method based on visual recognition as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program enables a computer to execute the method for detecting a wooden board based on visual recognition according to any one of claims 1 to 5.

Images