Search

CN-122016837-A - Device and method for detecting surface defects of wood-based plate

CN122016837ACN 122016837 ACN122016837 ACN 122016837ACN-122016837-A

Abstract

The invention discloses a device and a method for detecting surface defects of a wood-based plate, and relates to the technical field of plate defect detection. The invention tracks the plate speed in real time through the coding trigger unit and generates the synchronous trigger signal, so that the scanning line frequency of the camera is precisely matched with the plate movement speed, and the image movement blur is fundamentally eliminated. The differential double-line design allows for simultaneous acquisition of high resolution and standard resolution images, taking into account both detail and efficiency. The stroboscopic multispectral lighting unit alternately emits light with different wavelengths, and the contrast ratio of the defects to the background is enhanced by utilizing the reflection characteristic difference of the defects of different materials on the light with different wavelengths. The frequency domain detection branch carries out self-adaptive weighting on the defect characteristics in the transformation domain, so that the low-contrast defect and the complex background are effectively decoupled.

Inventors

  • LI JIANYONG
  • SUN ZHIQIANG
  • SUN HENGLI

Assignees

  • 湖北省松滋市航森木业有限公司

Dates

Publication Date
20260512
Application Date
20260319

Claims (10)

  1. 1. A wood-based panel surface defect detection device is characterized by comprising a conveying mechanism (1), a differential double-line scanning imaging assembly (2), a stroboscopic multispectral illumination unit (3), a coding trigger unit (4) and an image processing unit (5), wherein the conveying mechanism (1) is used for conveying a wood-based panel along a first direction, the differential double-line scanning imaging assembly (2) is arranged above the conveying mechanism (1) and comprises a first line scanning camera (21), a second line scanning camera (22), a first line light source (23) and a second line light source (24), the first line scanning camera (21) and the second line scanning camera (22) are spaced by a preset distance along the first direction, the scanning line frequency of the first line scanning camera (21) is higher than the scanning line frequency of the second line scanning camera (22), the first line light source (23) and the second line light source (24) are respectively arranged corresponding to the first line scanning camera (21) and the second line scanning camera (22), the stroboscopic multispectral illumination unit (3) is connected with the differential double-line scanning imaging assembly (2) and used for controlling the first line scanning camera (22) and the second line scanning camera (22) to emit light with different coding trigger light with the first line scanning unit (4) in an alternating mode, the system comprises an artificial board, an image processing unit (5) and a code triggering unit (4), wherein the artificial board is used for acquiring the instantaneous movement speed of the artificial board in real time and generating a trigger signal, and the image processing unit (5) is respectively connected with the differential double-line scanning imaging assembly (2) and the code triggering unit (4) and is used for synchronously controlling the image acquisition of the first line scanning camera (21) and the second line scanning camera (22) based on the trigger signal and processing the acquired images to identify defects.
  2. 2. The wood-based panel surface defect detection device according to claim 1, wherein the strobe multi-spectrum lighting unit (3) comprises a first light source driver (31), a second light source driver (32) and a synchronization controller (33), the first light source driver (31) is connected with the first line light source (23), the second light source driver (32) is connected with the second line light source (24), and the synchronization controller (33) is respectively connected with the first light source driver (31) and the second light source driver (32) and is used for controlling the first line light source (23) and the second line light source (24) to alternately emit light with different set wavelengths at preset time sequences.
  3. 3. The surface defect detection device for wood-based boards according to claim 2, wherein the first line light source (23) comprises at least a blue LED with a wavelength of 460nm-480nm and a green LED with a wavelength of 520nm-540nm, and the second line light source (24) comprises at least a red LED with a wavelength of 620nm-640nm and an infrared LED with a wavelength of 850nm-950 nm.
  4. 4. A wood-based panel surface defect detection device according to claim 3, characterized in that the scanning line frequency of the first line scanning camera (21) Line frequency with a second line scanning camera (22) The method meets the following conditions: wherein k is an integer greater than 1, the first line scan camera (21) is used to acquire high resolution detail images, and the second line scan camera (22) is used to acquire standard resolution images.
  5. 5. The wood-based panel surface defect detection device according to claim 4, wherein the coding triggering unit (4) comprises a rotary encoder (41) and a programmable logic controller (42), the rotary encoder (41) is coaxially connected with a driving roller of the conveying mechanism (1), and the programmable logic controller (42) is connected with the rotary encoder (41) and is used for calculating the real-time motion speed of the artificial panel according to the pulse signal of the rotary encoder (41) and generating triggering pulses of the first line scanning camera (21) and the second line scanning camera (22) according to preset scanning resolution.
  6. 6. The wood-based panel surface defect detection device according to claim 5, wherein the image processing unit (5) comprises an image acquisition card (51), an FPGA preprocessing module (52), an embedded processor (53), a display alarm module (54) and a rejection execution mechanism, the image acquisition card (51) is used for synchronously receiving image data output by the first line scanning camera (21) and the second line scanning camera (22), the FPGA preprocessing module (52) is connected with the image acquisition card (51) and used for correcting and splicing the received image data in real time, the embedded processor (53) is connected with the FPGA preprocessing module and used for running a defect detection algorithm, the display alarm module (54) is connected with the embedded processor (53) and used for displaying detection results and sending alarm signals when defects are detected, and the rejection execution mechanism is connected with the embedded processor (53) and arranged at the rear end of the conveying mechanism (1) and used for rejecting the defects from a production line according to the detection results.
  7. 7. The wood-based panel surface defect detection device according to claim 6, wherein the differential dual-line scanning imaging assembly (2) further comprises a pair of fine tuning holders (25), the pair of fine tuning holders (25) being configured to independently adjust the photographing angle and the focusing distance of the first line scanning camera (21) and the second line scanning camera (22); The included angle between the optical axis of the first line scanning camera (21) and the normal direction of the surface of the artificial board is 0-5 degrees, and the included angle between the optical axis of the second line scanning camera (22) and the normal direction of the surface of the artificial board is 10-20 degrees.
  8. 8. A method for detecting surface defects of a wood-based panel, characterized in that the wood-based panel is detected by using the device for detecting surface defects of a wood-based panel according to claim 7, comprising the steps of: Step 1, acquiring the motion speed of an artificial board in real time through a coding trigger unit (4), and generating a synchronous trigger signal according to the motion speed; Step 2, according to the synchronous trigger signal, controlling a first line scanning camera (21) in the differential speed double-line scanning imaging component (2) to acquire high-resolution images at a first line frequency, and simultaneously controlling a second line scanning camera (22) to acquire standard-resolution images at a second line frequency; Step 3, preprocessing the acquired high-resolution image and standard-resolution image, including image correction, denoising and stitching, to form an image pair to be detected; Step 4, inputting the image pair to be detected into a pre-trained double-domain parallel detection network, respectively extracting spatial domain features and frequency domain features, and fusing the spatial domain features and the frequency domain features to perform defect identification; and 5, classifying and positioning the defects according to the defect identification result, and generating a detection report.
  9. 9. The method for detecting surface defects of wood-based panels according to claim 8, wherein the dual-domain parallel detection network in step 4 comprises a spatial domain detection branch and a frequency domain detection branch; The airspace detection branch adopts an improved YOLOv framework, a coordinate attention mechanism is embedded in a main network, and a weighted bidirectional feature pyramid structure is adopted in a neck network to perform multi-scale feature fusion; The frequency domain detection branch comprises a fast Fourier transform module, a multi-axis frequency domain weighting representation module, a Gaussian filter module and an inverse fast Fourier transform module, wherein the fast Fourier transform module is used for converting an input image into a frequency domain, the multi-axis frequency domain weighting representation module is used for carrying out self-adaptive weighting on different frequency components in a transform domain so as to enhance defect characteristics and inhibit background textures; and carrying out feature fusion on the features extracted by the airspace detection branch and the enhancement feature map output by the frequency domain detection branch before a detection head.
  10. 10. The method for detecting surface defects of wood-based boards according to claim 9, wherein the multi-axis frequency domain weighting representation module weights the frequency domain coefficients by using a learnable weight matrix, and the weight matrix is generated by: W(u,v)=sigmoid(MLP(concat(P(u),P(v),|F(u,v)|))) Wherein u, v are frequency domain coordinates, P (u) and P (v) are position codes of u and v, |f (u, v) | is the amplitude of the frequency domain coefficient, MLP is a multi-layer perceptron, and sigmoid is an activation function; the frequency domain detection branch also comprises a self-adaptive dynamic downsampling module which is used for downsampling the feature map, and the self-adaptive dynamic downsampling module dynamically generates convolution kernel parameters according to the statistical characteristics of the input features so as to realize self-adaptive feature compression and enhancement of defects of different scales.

Description

Device and method for detecting surface defects of wood-based plate Technical Field The invention relates to the technical field of plate defect detection, in particular to a device and a method for detecting surface defects of a wood-based plate. Background Wood-based panels, such as particle board, medium density fiberboard, plywood, etc., are used as the main materials for furniture manufacture and interior decoration, and the surface quality thereof directly affects the grade and value of the final product. However, due to limitations of raw materials and production processes, many types of defects such as scratches, indentations, gum spots, oil stains, bubbles, cracks, etc. often occur on the surface of the artificial board. At present, quality inspection links of most of domestic artificial board production enterprises still depend on manual naked eye detection, and the problems of low detection efficiency, strong subjectivity, easy fatigue, incapability of realizing data tracing and the like exist. With the development of machine vision technology, an automatic detection method based on image processing is gradually applied to sheet surface defect detection. In the prior art, as disclosed in the patent with publication number CN210243521U, an artificial board surface defect visual detection device adopts a line scanning camera to collect images together with a line light source, and a cleaning mechanism and a rejecting mechanism are provided. As another example, patent publication No. CN115201111B discloses a device for detecting defects of an artificial board, and double-sided detection of a board is realized by clamping a flap component. In the algorithm level, the existing research mostly adopts a detection method based on deep learning, such as network structures of YOLO series, mobileNet and the like to carry out defect identification. However, the following technical bottlenecks still exist in the prior art: Motion blur problem-on high speed production lines, the relative motion between a conventional line scanning camera and a plate may cause image blur, affecting detection accuracy. Texture background interference, namely, the surface of the artificial board is provided with a complex wood grain or texture background, and defects with low contrast (such as shallow scratches and glue spots) are easily mixed with the background, so that the defects and the background are difficult to effectively separate by the traditional airspace processing method. The multi-scale defect detection is difficult, namely the defect size difference is large (from a small point-shaped defect in millimeter level to a large area defect in centimeter level), and the single-resolution imaging system and the single-scale detection algorithm are difficult to meet the detection requirements of all defect types. Uneven illumination and shadow, namely, shadow or reflective area is easily generated on the surface of the plate by single light source illumination, and the image quality is affected. Disclosure of Invention In order to solve the technical problems, the invention provides a device and a method for detecting surface defects of a wood-based panel. The technical scheme is as follows: The utility model provides a timber wood-based panel surface defect detection device, including conveying mechanism, differential double-line scanning imaging unit, stroboscopic multispectral lighting unit, code trigger unit and image processing unit, conveying mechanism is used for carrying the wood-based panel along first direction, differential double-line scanning imaging unit installs in conveying mechanism top, including first line scanning camera, second line scanning camera, first line light source and second line light source, first line scanning camera and second line scanning camera are along first direction interval preset distance, and the scanning line frequency of first line scanning camera is higher than the scanning line frequency of second line scanning camera, first line light source and second line light source correspond the setting with first line scanning camera and second line scanning camera respectively, stroboscopic multispectral lighting unit is connected with differential double-line scanning imaging unit for control first line light source and second line light source alternately send the illumination light of different settlement wavelength, the code trigger unit is connected with conveying mechanism's drive roller for obtain the instantaneous velocity of movement and generate the trigger signal in real time, image processing unit is connected with the trigger unit respectively with the double-line scanning imaging unit and is used for carrying out synchronous image acquisition to the first line scanning trigger unit and image acquisition to the image processing system defect. Optionally, the stroboscopic multispectral lighting unit includes first light source driver, second light source driver and synchr