Product quality inspection system for pultruded products

Abstract

The present invention relates to the technical field of product detection, and in particular to a product quality detection system for pultruded products. The system comprises: an information acquisition module, which is a box-type structure, and a camera device and a fill light device are arranged in the box body. The camera device is used to collect the outer surface information of a to-be-detected part, and the fill light device is used to provide the camera device with lighting that meets the requirements; a data processing module is used to receive the outer surface information collected by the camera device, and perform feature analysis to obtain feature data of the to-be-detected part, and analyze the feature data to obtain a quality detection report of the to-be-detected part; the present invention realizes automated quality detection of pultruded products through a visual detection system, improves production efficiency, reduces the work pressure of inspection and detection of the appearance, size, straightness, etc. of pultruded products, and reduces manpower consumption.

Description

Technical Field

[0001] The invention relates to the technical field of product detection, and in particular to a product quality detection system for pultruded products. Background Art

[0002] Compared with traditional materials, fiber-reinforced resin-based composite materials have many advantages such as high specific strength, light weight, high specific modulus, good fatigue resistance and good vibration reduction performance. The various component materials of the composite material play a synergistic role in performance and have superior comprehensive performance that cannot be matched by a single material.

[0003] There are various ways to form fiber-reinforced resin-based composite materials, among which pultrusion is an efficient production method that can achieve large-scale, continuous production. Compared with traditional processing methods, pultrusion has higher production efficiency and a pultrusion speed of 0.4-1.2m / min, which can meet the market demand for a large number of high-quality composite products. At the same time, it also brings tremendous pressure to the inspection and testing of the appearance, size, straightness, etc. of pultruded products. Summary of the invention

[0004] In view of this, the present invention aims to provide a product quality detection system for extruded and drawn products, in order to solve at least one of the above-mentioned technical problems.

[0005] To achieve the above object, the technical solution of the present invention is achieved as follows:

[0006] A product quality detection system for pultruded products, the system comprising:

[0007] An information acquisition module, which is a box-type structure, in which a camera device and a fill light device are arranged, the camera device is used to collect the outer surface information of the to-be-detected part, and the fill light device is used to provide the camera device with lighting that meets the requirements;

[0008] The data processing module receives the outer surface information collected by the camera device, and uses the neural network model to perform feature analysis to obtain feature data of the part to be tested, and analyzes the feature data to obtain a quality inspection report of the part to be tested.

[0009] Furthermore, the process of collecting the outer surface information of the to-be-detected part by means of a camera device includes:

[0010] The camera device continuously shoots the part to be inspected to obtain a partial image of the part to be inspected, and after removing noise from the partial image using a Gaussian filter function, a complete image of the part to be inspected is obtained through a splicing operation;

[0011] The image of the inspected part is enhanced and the processed image is edge segmented to obtain a center image and an edge image. Texture features are extracted from the center image and the edge image respectively, and the extracted results are used as the outer surface information.

[0012] Furthermore, the execution process of the splicing operation includes:

[0013] Obtain the key points of the local image, calculate the descriptors of the key points, perform a fast approximate nearest neighbor search on the descriptors, and obtain the matching points between each local image;

[0014] The affine transformation matrix between the images is calculated by matching points. By calculating the transformation matrix, a local image is transformed into the coordinate system of another local image, and the transformed local images are fused to obtain an image of the part to be inspected without splicing gaps.

[0015] Furthermore, the image fusion process includes:

[0016] Constructing Gaussian pyramids and Laplacian pyramids of each local image, fusing the Laplacian pyramids of the two images in the overlapping part of the two local images, upsampling and adding the fused Laplacian pyramids layer by layer to complete the image fusion; wherein the Gaussian pyramid is generated by downsampling the local images multiple times, and the Laplacian pyramid is generated by subtracting adjacent layers of the Gaussian pyramid.

[0017] Furthermore, the process of performing image enhancement processing on the image of the part to be inspected includes:

[0018] Construct a second-order derivative operator and convert the image of the object to be inspected into a grayscale image;

[0019] Traverse each pixel of the grayscale image, take each pixel as the center, take its neighborhood pixels and multiply them with the corresponding elements in the second-order derivative operator and sum them to get the Laplace value of the pixel, and superimpose the Laplace value to the corresponding position of the image of the part to be detected to complete the image enhancement processing.

[0020] Furthermore, in the process of superimposing the Laplace value to the corresponding position of the image of the part to be inspected, the value range of each pixel after superimposition is limited. When the superimposed value exceeds the value range, the value of the pixel is taken as the maximum value within the value range.

[0021] Furthermore, the execution process of the texture feature extraction includes:

[0022] Construct an empty matrix, traverse each pixel of the image of the object to be detected, find its neighboring pixel at the pixel in the specified direction and distance, record the pixel value of the current pixel as i, record the pixel value of the neighboring pixel as j, and add 1 to the i-th row and j-th column of the empty matrix;

[0023] For each element in the traversed matrix, the ratio of the element to the total number of elements in the matrix is ​​calculated to obtain a probability matrix; the texture features of the image of the object to be detected are calculated using the probability matrix.

[0024] Furthermore, the texture features include:

[0025] Contrast, ;entropy, ; angular second moment, ; uniformity, ; relevance, ;in, is the value of the element in the probability matrix, μi and μj are the means of the rows and columns, and σi and σj are the standard deviations of the rows and columns.

[0026] Furthermore, the neural network model includes an edge feature recognition model and a center feature recognition model, and the texture features of the center image and the edge image are respectively input into the edge feature recognition model and the center feature recognition model for training.

[0027] Furthermore, the process of providing the camera device with lighting that meets the requirements by using the fill light device includes:

[0028] The camera device is equipped with a light sensor, which monitors the current light intensity and color temperature and automatically adjusts the brightness of the fill light device according to the light intensity.

[0029] Compared with the prior art, the product quality inspection system for pultruded products of the present invention has the following beneficial effects:

[0030] Through the visual inspection system, automated quality inspection of pultruded products is achieved, which improves production efficiency, reduces the workload of inspection and testing of pultruded product appearance, size, straightness, etc., and reduces manpower consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 The figure is a schematic diagram of the working process of the visual detection system described in the embodiment of the present invention. DETAILED DESCRIPTION

[0032] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solution 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.

[0033] like Figure 1As shown, this embodiment provides a product quality detection system for pultruded products, the system comprising:

[0034] An information acquisition module, which is a box-type structure, in which a camera device and a fill light device are arranged, the camera device is used to collect the outer surface information of the to-be-detected part, and the fill light device is used to provide the camera device with lighting that meets the requirements;

[0035] The data processing module receives the outer surface information collected by the camera device, and uses the neural network model to perform feature analysis to obtain feature data of the part to be tested, and analyzes the feature data to obtain a quality inspection report of the part to be tested.

[0036] The process of collecting the outer surface information of the to-be-detected part by means of a camera device comprises:

[0037] The camera device continuously shoots the part to be inspected to obtain a partial image of the part to be inspected, and after removing noise from the partial image using a Gaussian filter function, a complete image of the part to be inspected is obtained through a splicing operation;

[0038] The image of the inspected part is enhanced and the processed image is edge segmented to obtain a center image and an edge image. Texture features are extracted from the center image and the edge image respectively, and the extracted results are used as the outer surface information.

[0039] The execution process of the splicing operation includes:

[0040] Obtain the key points of the local image, calculate the descriptors of the key points, perform a fast approximate nearest neighbor search on the descriptors, and obtain the matching points between each local image;

[0041] The affine transformation matrix between the images is calculated by matching points. By calculating the transformation matrix, a local image is transformed into the coordinate system of another local image, and the transformed local images are fused to obtain an image of the part to be inspected without splicing gaps.

[0042] The image fusion process includes:

[0043] Constructing Gaussian pyramids and Laplacian pyramids of each local image, fusing the Laplacian pyramids of the two images in the overlapping part of the two local images, upsampling and adding the fused Laplacian pyramids layer by layer to complete the image fusion; wherein the Gaussian pyramid is generated by downsampling the local images multiple times, and the Laplacian pyramid is generated by subtracting adjacent layers of the Gaussian pyramid.

[0044] The process of performing image enhancement processing on the image of the part to be inspected comprises:

[0045] Construct a second-order derivative operator and convert the image of the object to be inspected into a grayscale image;

[0046] Traverse each pixel of the grayscale image, take each pixel as the center, take its neighborhood pixels and multiply them with the corresponding elements in the second-order derivative operator and sum them to get the Laplace value of the pixel, and superimpose the Laplace value to the corresponding position of the image of the part to be detected to complete the image enhancement processing.

[0047] In the process of superimposing the Laplace value to the corresponding position of the image of the part to be inspected, the value range of each pixel after superimposition is limited. When the superimposed value exceeds the value range, the value of the pixel is taken as the maximum value within the value range.

[0048] The execution process of the texture feature extraction includes:

[0049] Construct an empty matrix, traverse each pixel of the image of the object to be detected, find its neighboring pixel at the pixel in the specified direction and distance, record the pixel value of the current pixel as i, record the pixel value of the neighboring pixel as j, and add 1 to the i-th row and j-th column of the empty matrix;

[0050] For each element in the traversed matrix, the ratio of the element to the total number of elements in the matrix is ​​calculated to obtain a probability matrix; the texture features of the image of the object to be detected are calculated using the probability matrix.

[0051] The texture features include:

[0052] Contrast, ;entropy, ; angular second moment, ; uniformity, ; relevance, ;in, is the value of the element in the probability matrix, μi and μj are the means of the rows and columns, and σi and σj are the standard deviations of the rows and columns.

[0053] The neural network model comprises an edge feature recognition model and a center feature recognition model, and the texture features of the center image and the edge image are respectively input into the edge feature recognition model and the center feature recognition model for training.

[0054] The process of providing the camera device with lighting that meets the requirements by using the fill light device includes:

[0055] The camera device is equipped with a light sensor, which monitors the current light intensity and color temperature and automatically adjusts the brightness of the fill light device according to the light intensity.

[0056] Obviously, the above embodiments are merely examples for the purpose of clear explanation, and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the scope of protection of the invention.

Claims

1. Product quality inspection system for pultruded products, characterized in that: The system comprises: An information acquisition module, which is a box-type structure, in which a camera device and a fill light device are arranged. The camera device continuously shoots the part to be detected to obtain a local image of the part to be detected. After the local image is denoised using a Gaussian filter function, a complete image of the part to be detected is obtained through a splicing operation, and image enhancement processing is performed on the image of the part to be detected; Perform edge segmentation on the processed image to obtain a center image and an edge image, extract texture features from the center image and the edge image respectively, and use the extraction results as external surface information to provide the camera device with lighting that meets the requirements through a fill light device; The data processing module receives the outer surface information collected by the camera device, and performs feature analysis on the center image and the edge image respectively using different neural network models to obtain feature data of the part to be inspected, and analyzes the feature data to obtain a quality inspection report of the part to be inspected; The execution process of the splicing operation includes: Obtain the key points of the local image, calculate the descriptors of the key points, perform a fast approximate nearest neighbor search on the descriptors, and obtain the matching points between each local image; The affine transformation matrix between the images is calculated by matching the points. A local image is transformed into the coordinate system of another local image by calculating the transformation matrix, and the transformed local images are fused to obtain an image of the part to be inspected without any joint gaps. The image fusion process includes: Constructing Gaussian pyramids and Laplacian pyramids of each local image, fusing the Laplacian pyramids of the two images in the overlapping part of the two local images, sampling up layer by layer from the fused Laplacian pyramid and adding them, and completing image fusion; wherein the Gaussian pyramid is generated by downsampling the local image multiple times, and the Laplacian pyramid is generated by subtracting adjacent layers of the Gaussian pyramid; The process of performing image enhancement processing on the image of the part to be inspected comprises: Construct a second-order derivative operator and convert the image of the object to be inspected into a grayscale image; Traverse each pixel of the grayscale image, take each pixel as the center, take its neighboring pixels and multiply and sum them with the corresponding elements in the second-order derivative operator to obtain the Laplace value of the pixel, and superimpose the Laplace value to the corresponding position of the image of the object to be detected to complete the image enhancement processing; In the process of superimposing the Laplace value to the corresponding position of the image of the part to be inspected, the value range of each pixel after superimposition is limited. When the superimposed value exceeds the value range, the value of the pixel is taken as the maximum value within the value range.

2. The product quality inspection system for pultruded products according to claim 1, characterized in that: The execution process of the texture feature extraction includes: Construct an empty matrix, traverse each pixel of the image of the object to be detected, find its neighboring pixel at the pixel in the specified direction and distance, record the pixel value of the current pixel as i, record the pixel value of the neighboring pixel as j, and add 1 to the i-th row and j-th column of the empty matrix; For each element in the traversed matrix, the ratio of the element to the total number of elements in the matrix is ​​calculated to obtain a probability matrix; the texture features of the image of the object to be detected are calculated using the probability matrix.

3. The product quality inspection system for pultruded products according to claim 2, characterized in that: The texture features include: Contrast, ;entropy, ; angular second moment, ; uniformity, ; relevance, ;in, is the value of the element in the probability matrix, μi and μj are the means of the rows and columns, and σi and σj are the standard deviations of the rows and columns.

4. The product quality inspection system for pultruded products according to claim 1, characterized in that: The neural network model comprises an edge feature recognition model and a center feature recognition model, and the texture features of the center image and the edge image are respectively input into the edge feature recognition model and the center feature recognition model for training.

5. The product quality inspection system for pultruded products according to claim 1, characterized in that: The process of providing the camera device with lighting that meets the requirements by using the fill light device includes: The camera device is equipped with a light sensor, which monitors the current light intensity and color temperature and automatically adjusts the brightness of the fill light device according to the light intensity.

Images