Hyperspectral imaging acquisition device for recycling fiber identification
The fiber cross-sectional images are collected, pre-processed, feature extraction and comparison through hyperspectral image and image processing system, which solves the problem of fiber identification, achieves rapid and accurate fiber classification, and improves reuse efficiency.
Patent Information
- Application Number
- CN202510653568.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-21
- Publication Date
- 2025-09-02
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In the prior art, it is difficult to achieve rapid identification of fibers while acquiring images.
A hyperspectral image is used to collect cross-sectional images of fibers, and image preprocessing, feature extraction and comparison evaluation are performed through the main control computer and image processing system, and a discriminant model is used to achieve rapid identification of fiber components.
The rapid and accurate identification and classification of large amounts of fiber samples is achieved, and the efficiency of recycling fiber treatment is improved.
Smart Images
Figure CN120577236A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of image acquisition technology, and in particular to a hyperspectral imaging acquisition device for identifying recycled fibers. Background Art
[0002] Recycled fibers come from a wide range of sources and may contain different types of fiber raw materials. Accurately identifying their composition and purity helps ensure that products meet relevant standards and quality requirements.
[0003] Due to the differences in chemical and molecular structures of fibers made of different materials, they absorb, reflect or transmit spectra of different wavelengths in different ways, thus having their own spectral characteristics. The hyperspectral imaging acquisition device collects image information of fibers in different spectral bands to obtain the spectral fingerprint of the target. It can accurately identify and classify a large number of fiber samples in a short period of time, greatly improving the efficiency of recycled fiber processing and helping to quickly separate and classify different types of fibers for subsequent reprocessing and utilization. Summary of the Invention
[0004] (1) Technical issues
[0005] The present invention aims to at least solve the problem in the prior art of requiring rapid fiber identification while acquiring images.
[0006] (2) Technical content
[0007] This solution provides a hyperspectral imaging acquisition device for recycling fiber identification, which is achieved by the following specific technical means: including a hyperspectral imager, a main control computer and an image processing system;
[0008] The hyperspectral imager is used to collect images of cross-sections of recycled fibers;
[0009] The main control computer is electrically connected to the hyperspectral imager and is used to receive images collected by the hyperspectral imager, and the image processing system is loaded in the main control computer;
[0010] The image processing system is used to process, analyze and compare the collected images to identify the fiber components.
[0011] Preferred technical solution 1: The image processing system includes an image preprocessing module, a feature extraction module and a comparison and evaluation module;
[0012] The image preprocessing module is used to preprocess the images collected by the hyperspectral imager;
[0013] The feature extraction module is used to extract features from the preprocessed image;
[0014] The comparison and evaluation module is used to compare the extracted feature data and provide an identification result.
[0015] Preferred technical solution 2: The image preprocessing module preprocesses the collected fiber cross-section image in a manner including image correction and image enhancement;
[0016] Image correction: The original image is corrected by the radiation correction field to eliminate the radiation difference between different bands, so that the spectral information in the image can more accurately reflect the true characteristics of the fiber;
[0017] Image enhancement: Since fiber images may have noise, low contrast and other problems, it is necessary to enhance the corrected image. The histogram equalization method is used to improve the contrast of the image and make the fiber features more obvious;
[0018] Use filtering algorithms to remove noise points in the image and smooth the image, but be careful to avoid excessive filtering that may cause blurred fiber edges.
[0019] Preferred technical solution three: the feature extraction of the feature extraction module includes spectral feature extraction and spatial feature extraction;
[0020] The spectral feature extraction: extracting the spectral curve of each fiber pixel from the hyperspectral image and analyzing its spectral features, such as absorption peak, reflection peak, spectral slope, etc.;
[0021] Calculating the first-order and second-order derivatives of the spectral curve can highlight the changing trend and inflection point information of the spectral curve, which helps to distinguish different chemical fibers more accurately;
[0022] The spatial feature extraction includes texture feature extraction and shape feature extraction;
[0023] The texture feature extraction: using gray level co-occurrence matrix, local binary pattern and other methods to extract the spatial texture features of the fiber in the hyperspectral image;
[0024] The shape feature extraction is to obtain the shape features of the fiber, such as length, width, aspect ratio, circularity, etc., through edge detection, contour extraction and other algorithms.
[0025] Preferred technical solution four: There is a discriminant model in the comparison and evaluation module. The hyperspectral image of the chemical fiber to be identified is preprocessed and feature extracted, and then input into the trained discriminant model. The discriminant model will output the predicted category of the fiber to achieve the identification of the chemical fiber.
[0026] Preferred technical solution five: Establishment of the discrimination model: Training with chemical fiber sample data of known categories to establish a discrimination model based on a convolutional neural network.
[0027] Preferred technical solution six: the preprocessing method of the image preprocessing module also includes size normalization.
[0028] Preferred technical solution seven: the image correction also includes geometric correction.
[0029] (3) Technical effects
[0030] The above structure enables this solution to have the following beneficial effects:
[0031] 1. Use a hyperspectral camera to collect hyperspectral images of recycled fibers, and use an image processing system to quickly process the collected images, eliminating the need for manual comparison;
[0032] 2. It can organize the collected hyperspectral images, provide fast and accurate image identification through the established discrimination model, and automatically output the identification results;
[0033] 3. It can accurately identify and classify a large number of fiber samples in a short period of time, greatly improving the efficiency of recycled fiber processing for subsequent reprocessing and utilization. BRIEF DESCRIPTION OF THE DRAWINGS
[0034] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:
[0035] Figure 1 This is the structural diagram of this solution;
[0036] Figure 2 This is a schematic diagram of the device for this scheme.
[0037] Among them, 1. Hyperspectral imager, 2. Main control computer, 3. Image processing system, 31. Image preprocessing module, 311. Image correction, 312. Image enhancement, 313. Size normalization, 32. Feature extraction module, 321. Spectral feature extraction, 322. Spatial feature extraction, 33. Comparison and evaluation module, 331. Discriminant model. DETAILED DESCRIPTION
[0038] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.
[0039] See also Figure 1-Figure 2, a hyperspectral imaging acquisition device for recycling fiber identification, comprising a hyperspectral imager 1, a main control computer 2 and an image processing system 3;
[0040] The hyperspectral imager 1 is used to collect images of the cross-section of recycled fibers. The hyperspectral imager 1 uses a hyperspectral camera. When collecting hyperspectral images, it is necessary to ensure that the illumination is uniform, the intensity is moderate, there is no shadow and reflection interference, and the light source is stable and covers the spectral range required by the camera to ensure imaging quality and accuracy.
[0041] The main control computer 2 is electrically connected to the hyperspectral imager 1 and is used to receive images collected by the hyperspectral imager 1, and the image processing system 3 is loaded in the main control computer 2;
[0042] The image processing system 3 is used to process, analyze and compare the collected images to identify the fiber components;
[0043] The image processing system 3 includes an image preprocessing module 31, a feature extraction module 32 and a comparison and evaluation module 33;
[0044] The image preprocessing module 31 is used to preprocess the images collected by the hyperspectral imager 1;
[0045] The feature extraction module 32 is used to extract features from the pre-processed image;
[0046] The comparison and evaluation module 33 is used to compare the extracted feature data and provide an identification result.
[0047] See also Figure 1 , a hyperspectral imaging acquisition device for recycling fiber identification, wherein the image preprocessing module 31 preprocesses the collected fiber cross-section image in a manner including image correction 311 , image enhancement 312 , and size normalization 313 ;
[0048] Image Correction 311: Perform radiation correction on the original image through the radiation correction field to eliminate the radiation differences between different bands, so that the spectral information in the image more accurately reflects the true characteristics of the fiber;
[0049] The image correction 311 also includes geometric correction, that is, during the acquisition process, due to factors such as the movement of the device or sample, the image may be geometrically distorted. The image is geometrically corrected using geometric correction algorithms, such as affine transformation and polynomial fitting, to ensure that the fiber morphology and position in the image are accurate;
[0050] Image enhancement 312: Since fiber images may have problems such as noise and low contrast, it is necessary to enhance the corrected image. The histogram equalization method is used to improve the image contrast and make the fiber features more obvious.
[0051] Use filtering algorithms to remove noise points in the image and smooth the image, but be careful to avoid excessive filtering that may cause blurring of fiber edges;
[0052] Size Normalization 313: To facilitate subsequent processing and analysis, the image size is normalized. Depending on the input requirements of the model, the image can be resized to a uniform size, such as 256×256 pixels or 512×512 pixels.
[0053] See also Figure 1 , a hyperspectral imaging acquisition device for recycling fiber identification, wherein the feature extraction module 32 includes spectral feature extraction 321 and spatial feature extraction 322;
[0054] The spectral feature extraction 321 extracts the spectral curve of each fiber pixel from the hyperspectral image and analyzes its spectral features, such as absorption peak, reflection peak, spectral slope, etc. Different types of chemical fibers have unique spectral absorption and reflection characteristics within specific wavelength ranges. These features can serve as an important basis for fiber identification.
[0055] Calculating the first-order and second-order derivatives of the spectral curve can highlight the changing trend and inflection point information of the spectral curve, which helps to more accurately distinguish different chemical fibers. For example, some chemical fibers have obvious extreme points in the first-order derivative at a specific wavelength, which can be used for identification.
[0056] The spatial feature extraction 322 includes texture feature extraction and shape feature extraction;
[0057] The texture feature extraction method uses gray-level co-occurrence matrix, local binary pattern and other methods to extract the spatial texture features of fibers in hyperspectral images. The surface texture of chemical fibers is closely related to their type and production process. For example, the thickness, uniformity and weaving method of the fibers are all reflected in the texture features.
[0058] Shape feature extraction: The shape features of the fiber, such as length, width, aspect ratio, and circularity, are obtained through edge detection, contour extraction, and other algorithms. These shape features are important for identifying the type and quality of the fiber.
[0059] The comparison and evaluation module 33 includes a discriminant model 331;
[0060] Establishment of the discriminant model 331: By training with chemical fiber sample data of known categories, a discriminant model 331 based on a convolutional neural network is established, with the extracted features as input and the class labels of the samples as output. By learning the relationship between the features and the categories of the samples, the discriminant model 331 can accurately classify and identify unknown fibers;
[0061] Optimization of the discriminant model 331: using an independent test data set to verify the discriminant model 331 and evaluate its accuracy and reliability. If the performance of the model is not satisfactory, further optimization of the model parameters, feature selection or other improvement measures may be adopted to improve the discrimination ability of the discriminant model 331.
[0062] After preprocessing and feature extraction, the hyperspectral image of the chemical fiber to be identified is input into the trained discriminant model 331 , which outputs the predicted category of the fiber to achieve identification of the chemical fiber.
[0063] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.
Claims
1. A hyperspectral imaging acquisition device for recycling fiber identification, characterized by: It includes a hyperspectral imager (1), a main control computer (2) and an image processing system (3); The hyperspectral imager (1) is used to collect images of cross-sections of recycled fibers; The main control computer (2) is electrically connected to the hyperspectral imager (1) and is used to receive images collected by the hyperspectral imager (1), and the image processing system (3) is loaded in the main control computer (2); The image processing system (3) is used to process, analyze and compare the collected images to achieve identification of fiber components.
2. The hyperspectral imaging acquisition device for recycling fiber identification according to claim 1, characterized in that: The image processing system (3) includes an image preprocessing module (31), a feature extraction module (32) and a comparison and evaluation module (33); The image preprocessing module (31) is used to preprocess the image collected by the hyperspectral imager (1); The feature extraction module (32) is used to extract features from the pre-processed image; The comparison and evaluation module (33) is used to compare the extracted feature data and provide an identification result.
3. The hyperspectral imaging acquisition device for recycling fiber identification according to claim 2, characterized in that: The image preprocessing module (31) preprocesses the collected fiber cross-sectional image in a manner including image correction (311) and image enhancement (312); Image correction (311): performing radiometric correction on the original image using a radiometric correction field; Image enhancement (312): Improve the contrast of the image, remove noise points in the image, smooth the image, and make the fiber features more obvious.
4. The hyperspectral imaging acquisition device for recycling fiber identification according to claim 2, characterized in that: The feature extraction of the feature extraction module (32) includes spectral feature extraction (321) and spatial feature extraction (322); The spectral feature extraction (321) extracts the spectral curve of each fiber pixel from the hyperspectral image and analyzes its spectral features; Calculate the first-order and second-order derivatives of the spectral curve to highlight the changing trend and inflection point information of the spectral curve; The spatial feature extraction (322) includes texture feature extraction and shape feature extraction.
5. The hyperspectral imaging acquisition device for recycling fiber identification according to claim 2, characterized in that: The comparison evaluation module (33) includes a discriminant model (331).
6. The hyperspectral imaging acquisition device for recycling fiber identification according to claim 3, characterized in that: The preprocessing method of the image preprocessing module (31) also includes size normalization (313).
7. The hyperspectral imaging acquisition device for recycling fiber identification according to claim 3, characterized in that: The image correction (311) also includes geometric correction.
8. The hyperspectral imaging acquisition device for recycling fiber identification according to claim 4, characterized in that: The texture feature extraction is used to extract the spatial texture features of fibers in the hyperspectral image.
9. The hyperspectral imaging acquisition device for recycling fiber identification according to claim 4, characterized in that: The shape feature extraction is used to obtain the shape features of the fiber.
10. The hyperspectral imaging acquisition device for recycling fiber identification according to claim 1, characterized in that: The hyperspectral imager (1) adopts a hyperspectral camera.