Textile blending ratio flaw detection method

Abstract

The invention provides a textile blending proportion defect detection method, which comprises the following steps: carrying out continuous image acquisition on a blended textile in a vibration state, and adjusting the exposure time and frame rate of camera equipment according to vibration frequency and amplitude characteristics to obtain a textile surface image sequence; the obtained textile surface image sequence is preprocessed, and warp and weft areas are segmented according to the surface texture features of the blended textile; based on the abnormal degree evaluation model, constructing a blending proportion flaw classifier, and training the classifier to identify different types of blending proportion flaws by using the warp and weft abnormal features, the fiber distribution features and the texture features as input; according to a blending proportion defect classifier, in combination with defect positioning requirements in a vibration environment, performing real-time tracking on detected different types of defect proportion areas, compensating image displacement caused by vibration, and ensuring that a defect detection result corresponds to an actual textile position.

Description

Technical Field

[0001] The invention relates to the field of information technology, and in particular to a method for detecting blending ratio defects of textiles. Background Art

[0002] Textile machines put textiles in a vibrating environment when weaving, and there is a complex technical problem when detecting textile blending ratio defects. Vibration can cause slight deformation and displacement on the surface of textiles, resulting in subtle changes in the arrangement and interweaving structure of the warp and weft. This change will affect the recognition accuracy of warp and weft anomalies by image processing technology, especially when detecting the breakage, dislocation and missing of the warp and weft. At the same time, the fiber composition of blended textiles is complex, and the deformation characteristics of different fibers under vibration conditions are different, which further increases the difficulty of detection. In addition, the vibration frequency and amplitude on high-speed production lines are often unstable, and this uncertainty makes it difficult to establish a unified detection model. Vibration may also cause jitter of the detection equipment itself, affecting the stability and continuity of image acquisition. In this case, how to accurately capture the microstructure of the textile surface in a vibrating environment and extract effective feature information from it has become a technical problem that needs to be solved urgently. Summary of the invention

[0003] The present invention provides a method for detecting textile blending ratio defects, which mainly includes:

[0004] Continuously collect images of the blended textile under vibration, adjust the exposure time and frame rate of the camera device according to the vibration frequency and amplitude characteristics, and obtain a sequence of textile surface images;

[0005] Preprocessing the obtained textile surface image sequence, segmenting the longitude and latitude areas according to the surface texture characteristics of the mixed textile;

[0006] Perform frequency domain analysis on the segmented textile warp and weft images, extract the arrangement direction and spacing information of the textile warp and weft by calculating the spectrum energy distribution characteristics, and perform matching and comparison with the preset normal textile warp and weft arrangement template to determine whether the warp and weft are abnormal;

[0007] The detected abnormal areas of the textile warp and weft are subjected to edge extraction and connected area marking segmentation, the area and perimeter geometric features of the abnormal areas are calculated, and an abnormality evaluation model is established in combination with the textile production process parameters to quantitatively analyze the defect type and severity.

[0008] Based on the abnormality evaluation model, a blending ratio defect classifier was constructed. The warp and weft abnormality features, fiber distribution features and texture features were used as input to train the classifier to identify different types of blending ratio defects.

[0009] According to the blended proportion defect classifier and the defect positioning requirements under vibration environment, different types of defect proportion areas are detected and tracked in real time to compensate for the image displacement caused by vibration and ensure that the defect detection results correspond to the actual textile position.

[0010] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0011] The present invention discloses a method for detecting blended proportion defects of textiles. The method uses a high-speed camera to continuously capture images of blended textiles in a vibrating state, and extracts the arrangement information of the warp and weft lines by performing frequency domain analysis using Fourier transform through image preprocessing and warp and weft area segmentation. For the detected abnormal area, a local affine transformation is used to correct the deviation caused by wrinkles, and fine segmentation is performed in combination with morphological operations and connected region marking algorithms. By extracting geometric and texture features, an abnormality degree assessment model based on a support vector machine and a blended proportion defect classifier are constructed. Finally, different types of defect areas are tracked in real time, the displacement caused by vibration is compensated, and the defect detection results are accurately corresponded to the actual textile position. The present invention can accurately detect the proportion defects of blended textiles in a vibrating environment, thereby improving the detection accuracy and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 The present invention is a flow chart of a method for detecting textile blending ratio defects.

[0013] Figure 2 It is a schematic diagram of a method for detecting textile blending ratio defects of the present invention.

[0014] Figure 3 It is another schematic diagram of a method for detecting textile blending ratio defects of the present invention. DETAILED DESCRIPTION

[0015] In order to further understand the content of the present invention, the present invention is described in detail in conjunction with the accompanying drawings and embodiments. The present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are only used to explain the relevant inventions, rather than to limit the invention. It is also necessary to explain that, for ease of description, only the parts related to the invention are shown in the accompanying drawings.

[0016] like Figure 1 -3. In this embodiment, a method for detecting blending ratio defects of textiles may specifically include:

[0017] S101, continuously collecting images of a blended textile in a vibrating state, adjusting the exposure time and frame rate of a camera device according to the vibration frequency and amplitude characteristics, and obtaining a textile surface image sequence.

[0018] A camera device is used to obtain a continuous image sequence of the mixed textile surface under a vibrating state, and the exposure time and acquisition frame rate of the camera device are adjusted according to the vibration frequency and vibration amplitude. For the continuous image sequence, the image is preprocessed by adaptive histogram equalization to obtain an enhanced textile surface image sequence; an optical flow estimation algorithm is used to calculate the displacement field between adjacent frames, and the image is corrected according to the displacement field to obtain a stable textile surface image sequence.

[0019] Specifically, the Phantomv2512 high-speed camera is used to continuously capture images of blended textiles according to the vibration frequency range of 10-1000Hz and the vibration amplitude range of 0.1-10mm. The exposure time of the camera is adjusted to 0.1-1ms and the acquisition frame rate is adjusted to 1000-10000fps to adapt to the vibration characteristics. The surface of the blended textile in the vibrating state is continuously captured to obtain a clear textile surface image sequence. For the acquired textile surface image sequence, the image is preprocessed by adaptive histogram equalization. By adjusting the contrast, brightness and saturation of the image, the clarity and detail visibility of the image are improved to obtain an enhanced textile surface image sequence. The enhanced textile surface image sequence is motion compensated, and the displacement field between adjacent frames is calculated by the optical flow estimation algorithm. The image is corrected according to the displacement field to eliminate the influence of vibration on the image quality, and a stable textile surface image sequence is obtained. The Phantomv2512 high-speed camera was used to continuously capture images of blended textiles. The exposure time of the camera was set to 0.5ms and the acquisition frame rate was set to 5000fps to adapt to the vibration frequency of 50Hz and the vibration amplitude of 5mm. The acquired textile surface image sequence was processed by adaptive histogram equalization to expand the image grayscale from 256 to 1024 to improve the image contrast. The enhanced image sequence was motion compensated, and the Lucas-Kanade optical flow algorithm was used to calculate the displacement field between adjacent frames. The average value of the displacement field was (2.5pixel, 1.8pixel). Based on this, the image was affinely transformed and corrected to eliminate the influence of vibration.

[0020] S102 , preprocessing the obtained textile surface image sequence, and segmenting the longitude and latitude regions according to the surface texture features of the mixed textile.

[0021] A textile surface image sequence is acquired, and the image sequence is filtered to obtain a denoised textile surface image; a contrast enhancement process is performed on the denoised textile surface image to obtain a contrast enhanced textile surface image; a Gabor filter group is used to extract texture features of the contrast enhanced textile surface image to obtain a filter response; a principal component analysis is performed on the filter response, and a preset number of principal components are selected as texture feature representations to obtain a texture feature map after dimensionality reduction; if the texture feature map after dimensionality reduction exists, a binarization process is performed on the texture feature map; a morphological opening operation is performed on the binarized feature map to obtain a longitude and latitude area segmentation result.

[0022] Specifically, for the collected textile surface image sequence, a Gaussian filter is used to filter the image. By setting the Gaussian kernel size to 5×5 and the standard deviation to 1.0, the image is convolved to obtain the denoised textile surface image. The denoised textile surface image is subjected to contrast-limited adaptive histogram equalization processing, the block size is set to 8×8 pixels, and the contrast limit threshold is set to 0.01, to obtain the contrast-enhanced textile surface image. For the contrast-enhanced textile surface image, a Gabor filter group is used to extract texture features. The orientation angles of the Gabor filter group are set to 0°, 22.5°, 45°, 67.5°, 90°, 112.5°, 135°, and 157.5°, the scales are 1, 2, 4, 8, and 16 pixels, and the center frequencies are 0.1, 0.2, 0.3, 0.4, and 0.5, to obtain 40 filter responses. The principal component analysis is performed on the Gabor filter response, and the first three principal components are selected as the texture feature representation to obtain the texture feature map after dimensionality reduction. Apply Otsu threshold segmentation algorithm to the texture feature map after dimensionality reduction, calculate the optimal threshold, and divide the feature Figure 2The image was then quantized, and then a 3×3 pixel rectangular structure element was used for morphological opening operation to remove the connected areas with an area less than 10 pixels, and the longitude and latitude area segmentation result map was obtained. The collected textile surface image sequence was processed by Gaussian filtering, using a 5×5 kernel and a standard deviation of 1.0, which reduced the image noise level and the average noise intensity from 15 to 3. The filtered image was enhanced by contrast-limited adaptive histogram equalization, with a block size of 8×8 pixels and a limit threshold of 0.01, which increased the image contrast by 40% and made the grayscale distribution more uniform. The enhanced image was applied with a Gabor filter bank to extract texture features, setting 8 direction angles, from 0° to 157.5°, with an interval of 22.5°, 5 scales, 1, 2, 4, 8, and 16 pixels, and 5 center frequencies, including 0.1 to 0.5, with an interval of 0.1, to obtain 40 response maps. The principal component analysis was performed on these 40 response maps, and the first three principal components were selected, which explained 95% of the variance and compressed the 40-dimensional features into 3 dimensions. The feature map after dimensionality reduction is segmented by threshold using Otsu algorithm, and the best threshold is calculated to be 127. Figure 2 The binary image is morphologically opened using a 3×3 rectangular structural element, and connected areas smaller than 10 pixels are removed, ultimately obtaining the warp and weft area segmentation results. The segmentation results show that the warp accounts for 45% of the image area, the weft accounts for 35%, the accuracy rate reaches 92%, and the average time consumed is 1.2 seconds per frame, meeting the real-time requirements. This method is suitable for blended textiles of different materials, such as cotton-polyester blends, wool-acrylic blends, etc., and has a good effect on fabrics with a warp and weft density of 30-120 threads per inch.

[0023] S103, performing frequency domain analysis on the segmented textile warp and weft images, extracting arrangement direction and spacing information of the textile warp and weft by calculating spectrum energy distribution characteristics, performing matching and comparison according to a preset normal textile warp and weft arrangement template, and determining whether the warp and weft are abnormal.

[0024] A two-dimensional fast Fourier transform is performed on the segmented textile warp and weft line image to obtain a frequency domain representation representing the structural features of the textile warp and weft lines. The amplitude spectrum of the frequency spectrum is calculated based on the frequency domain representation, and the amplitude spectrum is represented by polar coordinates. For the amplitude spectrum represented by the polar coordinates, multiple directions are taken at preset angle intervals within a preset angle range, and the energy distribution in each direction is calculated to obtain a feature vector representing the arrangement direction and spacing information of the textile warp and weft lines. The cosine similarity is used to calculate the similarity between the feature vector and the samples in the pre-established normal textile warp and weft line arrangement template library. If the similarity is less than a preset threshold, the warp and weft line arrangement is judged to be abnormal. For the warp and weft line area judged to be abnormal, the Euclidean distance between the feature vector and the template sample is calculated, the abnormal type is judged based on the Euclidean distance, and the specific type information of the abnormal warp and weft lines is determined.

[0025] Specifically, for the segmented textile warp and weft images, a 512×512-point two-dimensional fast Fourier transform is used to perform frequency domain conversion on the image. By calculating the complex form spectrum after Fourier transform, the frequency domain representation representing the warp and weft structural characteristics of the textile is obtained. The spectrum after Fourier transform is analyzed for energy distribution, and the amplitude spectrum of the spectrum is calculated. Polar coordinates are used to represent it. A direction is taken every 5° in the range of 0° to 180°, and the energy distribution in this direction is calculated to obtain a 36-dimensional feature vector to characterize the arrangement direction and spacing information of the textile warp and weft. According to the pre-established normal textile warp and weft arrangement template library, the cosine similarity is used to calculate the similarity between the feature vector and the sample in the template library, and the warp and weft arrangement is judged to be abnormal by setting the similarity threshold of 0.95. For the warp and weft areas judged to be abnormal, the Euclidean distance between the feature vector and the template is calculated, and the abnormal type is judged according to the distance, such as missing warp and uneven weft spacing, to obtain the specific type information of the abnormal warp and weft. A 512×512-point two-dimensional fast Fourier transform is applied to the segmented 1024×1024 pixel textile warp and weft line image to obtain a complex form spectrum map. The amplitude spectrum of the spectrum map is calculated and represented by polar coordinates. A direction is taken every 5° in the range of 0° to 180°, and the energy distribution in this direction is calculated to obtain a 36-dimensional feature vector. In the specific process, the amplitude spectrum is divided into 36 fan-shaped areas, each covering a 5° angle range, and the average energy of the pixel values ​​in each area is calculated to form a feature vector. For example, for cotton textiles, the energy values ​​of the feature vector of a normal sample in the 0° and 90° directions are 0.8 and 0.75 respectively, while the energy values ​​in other directions are all less than 0.2. The obtained feature vector is compared with the pre-established 100 normal sample template library, and the cosine similarity calculation method is used, and the similarity threshold is set to 0.95. If the calculated similarity is lower than the threshold, it is judged as abnormal. For the sample judged to be abnormal, the Euclidean distance between its feature vector and each sample in the template library is calculated, and the template with the smallest distance is the most similar normal sample. If the component of the Euclidean distance in the 5° direction exceeds 0.3, it is judged that the warp is missing; if the component in the 85°-95° ​​direction exceeds 0.2, it is judged that the weft spacing is uneven. In this way, it can detect abnormal situations where the warp density deviation exceeds 5% or the coefficient of variation of the weft spacing exceeds 10%, meeting the quality control requirements of textiles.

[0026] The abnormality detection threshold is dynamically adjusted according to the textile type and production process parameters to determine whether the current warp and weft arrangement is abnormal. The local affine transformation model is used to perform geometric correction on the detected local deformation area to eliminate the spectrum analysis deviation caused by the wrinkle factor.

[0027] Obtain the reference threshold corresponding to the textile type and production process parameters in the preset parameter library; perform real-time statistical analysis on the current batch of samples, calculate the mean and standard deviation of the density and spacing of the warp and weft lines of the samples, and generate the dynamic anomaly detection threshold for the current batch based on the mean, standard deviation and reference threshold. Perform regional division on the textile image, and obtain the feature vector representing the local texture structure by calculating the statistics of the gray-level co-occurrence matrix of each sub-region and the feature vector of the local binary mode. Use the feature vector to construct a local affine transformation model, map the feature vector to the parameters of the affine transformation, estimate the affine transformation matrix parameters, and perform geometric correction on the detected local deformation area. Perform Fourier transformation on the corrected image, calculate the spectral energy distribution characteristics of the image, normalize the features, and compare them element by element with the dynamic anomaly detection threshold. If the number of elements exceeding the threshold exceeds the preset percentage, it is determined that the current warp and weft arrangement is abnormal.

[0028] Specifically, the corresponding benchmark threshold is extracted from the preset parameter library according to the type of textile and the production process parameters. The parameter library contains the benchmark thresholds for different textile types such as cotton, polyester, and blended textiles, and the combinations of process parameters such as weaving speed and tension. The current batch of samples is counted in real time by the sliding window method, and the mean and standard deviation of the density and spacing of the warp and weft lines are calculated. The abnormality detection threshold suitable for the current batch is dynamically generated in combination with the benchmark threshold. The K-means clustering algorithm is used to divide the textile image into regions. The feature vector representing the local texture structure is obtained by calculating the four statistics of the gray level co-occurrence matrix of each sub-region, including energy, contrast, correlation, entropy, and the 59-dimensional feature vector of the local binary pattern. The local affine transformation model is constructed using the feature vector, and the feature vector is mapped to the six parameters of the affine transformation. The overdetermined equations are solved by the least squares method, and the affine transformation matrix parameters are estimated. The detected local deformation area is geometrically corrected to eliminate the image distortion caused by the wrinkle factor. The corrected image is transformed by Fourier, and the spectral energy distribution characteristics are calculated. The characteristics are normalized and compared element by element with the dynamically generated threshold. If the number of elements exceeding the threshold exceeds the preset percentage, the current warp and weft arrangement is judged to be abnormal, and the corrected and optimized abnormal detection results are obtained. For a batch of cotton-polyester blended fabrics, the benchmark thresholds are extracted from the parameter library, such as the warp and weft density deviation threshold of ±5%, and the spacing coefficient of variation threshold of 10%. Using a sliding window of 100×100 pixels and a step size of 50 pixels, the 1000×1000 pixel image is counted in real time, and the mean warp density is 120 strands / inch, the standard deviation is 3 strands / inch, and the mean weft density is 80 strands / inch, and the standard deviation is 2 strands / inch. Combined with the benchmark threshold, the abnormal detection thresholds for the current batch are dynamically generated as 114-126 strands / inch for warp density and 76-84 strands / inch for weft density. The K-means clustering algorithm is applied, and the number of clusters k is set to 5 to divide the image into regions. For each sub-region, the energy, contrast, correlation, and entropy statistics of the gray-level co-occurrence matrix and the 59-dimensional eigenvector of the local binary pattern are calculated to obtain a 63-dimensional eigenvector. The 63-dimensional eigenvector is converted into six parameters of the affine transformation (a11, a12, a21, a22, tx, ty) through linear mapping, and the equation system Ax=b is constructed, where A is the feature matrix, x is the affine parameter to be solved, and b is the ideal coordinate. The overdetermined equation system is solved using the least squares method to obtain the optimal affine transformation matrix. The matrix is ​​applied to the detected local deformation area for geometric correction to eliminate the distortion caused by wrinkles. The corrected image is subjected to a 512×512-point two-dimensional fast Fourier transform to calculate the spectral energy distribution characteristics. After the features are normalized, they are compared element by element with the dynamic threshold, and the threshold for the number of elements exceeding the threshold is set to 10%. If the threshold is exceeded, the current longitude and latitude arrangement is judged to be abnormal.

[0029] S104, extracting edges and marking connected regions on the detected abnormal areas of the warp and weft lines of the textile, calculating the area and perimeter geometric features of the abnormal areas, establishing an abnormality assessment model in combination with textile production process parameters, and conducting quantitative analysis on the types and severity of defects.

[0030] According to the image of the abnormal area of ​​the warp and weft lines of the textile, edge extraction is performed to obtain an edge binary image of the abnormal area; the connected area of ​​the edge binary image is segmented to obtain a marked abnormal area image; for the marked abnormal area image, the area and the perimeter of the abnormal area are calculated; according to the area and the perimeter of the abnormal area, the circularity and rectangularity shape factors are calculated to determine the geometric feature vector of the abnormal area; an abnormality degree evaluation model is constructed, and the type and severity of the detected defects are judged through the geometric feature vector and the weaving speed and yarn tension process parameters to obtain the abnormality evaluation result.

[0031] Specifically, for the detected abnormal areas of the warp and weft of the textile, the Canny edge detection algorithm is used for edge extraction. By setting the low threshold to 50, the high threshold to 150, and the standard deviation of the Gaussian filter to 1.4, the edge contour of the abnormal area is extracted to obtain the edge binary image. The eight-connected region labeling algorithm is applied to the edge binary image for connected region segmentation. The two-pass scanning method is used. The first pass is scanned from top to bottom and from left to right, and the second pass is scanned in reverse to solve the equivalent labeling problem and obtain the labeled abnormal area image. According to the labeled abnormal area image, the pixel counting method is used to calculate the area of ​​each abnormal area, and the Freeman chain code method is used to calculate the perimeter of the abnormal area. By calculating the shape factors such as circularity (4π*area / (perimeter ^2)) and rectangularity (area / (minimum circumscribed rectangle area)), the geometric feature vector of the abnormal area is obtained. Combined with textile production process parameters such as weaving speed and yarn tension, an abnormality degree evaluation model based on support vector machine is constructed. The radial basis kernel function is used, the penalty parameter C is set to 1.0, and the gamma parameter is set to 0.1. The optimal parameters are selected through cross-validation. The geometric feature vector and process parameters of the abnormal area are input, and the types of detected defects, such as broken warp, broken weft, stains and severity (minor, medium, severe) are quantitatively analyzed to obtain the abnormal evaluation results. For a 1024×1024 pixel textile image, the Canny edge detection algorithm is first applied, and the low threshold is set to 50, the high threshold is set to 150, and the Gaussian filter standard deviation is 1.4 to extract the edge of the abnormal area. In the obtained edge binary image, three main abnormal areas are detected. Subsequently, the eight-connected region marking algorithm is used, and the connected region segmentation is performed by two-pass scanning. The first scan marks five potential abnormal areas, and the equivalent marks are merged after the second reverse scan, and finally three independent abnormal areas are determined. The geometric features of these three areas are calculated separately, among which the first area has an area of ​​2500 pixels, a perimeter of 200 pixels, a circularity of 0.78, and a rectangularity of 0.85; the second area has an area of ​​1800 pixels, a perimeter of 180 pixels, a circularity of 0.69, and a rectangularity of 0.79; the third area has an area of ​​3200 pixels, a perimeter of 240 pixels, a circularity of 0.70, and a rectangularity of 0.82. Combined with the production process parameters, weaving speed of 200 m / min, and yarn tension of 2.5 cN, these features are input into the support vector machine model. The model uses a radial basis kernel function, with a penalty parameter C of 1.0 and a gamma of 0.1, and is optimized by 5-fold cross validation. The model output shows that the first area is a broken warp with a moderate severity and a score of 0.75; the second area is a stain with a slight severity and a score of 0.35; the third area is a broken weft with a severe severity and a score of 0.92.

[0032] The size and shape of the structural elements are selected according to the complexity of the textile texture. Based on the morphological operation parameters, the rough outline of the abnormal area of ​​the longitude and latitude lines is extracted, and the connected region labeling algorithm is used to introduce local texture features for fine segmentation of the abnormal area.

[0033] The grayscale co-occurrence matrix features of the textile image are obtained, and the texture complexity index is calculated according to the grayscale co-occurrence matrix features, wherein the texture complexity index includes four features: contrast, correlation, energy and entropy. If the texture complexity index is greater than a preset threshold, the corresponding size and shape of the structural element are selected to obtain the initial value of the morphological operation parameter. The morphological operation parameters are adjusted by the gradient descent method, and a rough outline map of the abnormal area is obtained by opening and closing operations. For the rough outline map, a connected region labeling algorithm is applied, and the connected region labeling algorithm introduces the local binary pattern feature on the basis of the scanning method to calculate the characteristic distance between the central pixel and the neighboring pixel. If the characteristic distance is less than the preset distance threshold, the region is merged to obtain a preliminary segmented abnormal region image. The preliminary segmented abnormal region image is finely segmented by using a region growing algorithm, and the preliminary segmented abnormal region is used as a seed point to determine whether the grayscale difference and the texture feature similarity meet the preset growth criteria. If the growth criteria are met, the abnormal region boundary is extended until the rate of change of the number of pixels at the region boundary is less than the preset stop threshold, and a fine segmentation result image of the longitude and latitude abnormal region is obtained.

[0034] Specifically, the texture complexity index is calculated based on the gray-level co-occurrence matrix characteristics of the textile image, and the complexity index is calculated using four characteristics: contrast, correlation, energy, and entropy. By setting the complexity threshold, the corresponding size and shape of the structural element are selected to obtain the initial value of the morphological operation parameter suitable for the current textile texture characteristics. The gradient descent method is used to adjust the morphological operation parameters, and the learning rate is set to 0.01, the maximum number of iterations is 100, and the opening and closing operations are repeated until the contour extraction results of the abnormal area are stable, and a rough contour map of the abnormal area of ​​the longitude and latitude lines is obtained. The connected region labeling algorithm is applied to the rough contour map. On the basis of the traditional two-pass scanning method, the local binary pattern features of the 3x3 neighborhood are introduced, the characteristic distance between the central pixel and the neighboring pixel is calculated, and the distance threshold is set to 0.2. The region is merged to obtain a preliminary segmented abnormal area image. Based on the preliminary segmentation results, the region growing algorithm is used for fine segmentation. The preliminary segmented area is used as the seed point. The growth criterion is set to the grayscale difference less than 10 and the texture feature similarity greater than 0.8. The stopping condition is that the change rate of the number of pixels at the regional boundary is less than 1%. The abnormal region boundary is gradually expanded, and finally the fine segmentation result image of the longitude and latitude abnormal region is obtained. For a 1024×1024 pixel textile image, its grayscale co-occurrence matrix features are first calculated to obtain the contrast

[0035] 0.8, correlation 0.6, energy 0.3 and entropy 4.5. Based on these features, the texture complexity index is calculated to be 0.75. Since the complexity index is greater than 0.7, a 5×5 diamond structure element is selected as the initial parameter of the morphological operation. The gradient descent method is used to optimize the morphological operation parameters, and the initial learning rate is set to 0.01. After 73 iterations, the parameters converged and the optimized structure element size was 7×7. This optimized structure element is applied to open and close operations to extract the rough outline of the abnormal area. Then, the improved connected region labeling algorithm is applied to the contour map. The local binary pattern features are calculated in a 3×3 neighborhood to obtain a 59-dimensional feature vector. The feature distance threshold is set to 0.2, and adjacent pixels are merged. This process marks a total of 5 potential abnormal areas. Then, with these 5 areas as seed points, the region growing algorithm is applied for fine segmentation. The grayscale difference threshold is set to 10 and the texture feature similarity threshold is set to 0.8. After an average of 12 iterations, 3 of the 5 regions reached the stopping condition, which was that the change rate of the number of boundary pixels was less than 1%, and were finally determined to be abnormal regions. The areas of these 3 abnormal regions were 1200, 800 and 1500 square pixels, respectively, accounting for 0.28% of the entire image area.

[0036] The geometric features and texture features of the abnormal area are extracted. The geometric features include area, perimeter, and circularity. The principal component analysis method is used to reduce the dimension of the features and obtain the feature vector of the abnormal area to construct an abnormality degree assessment model based on support vector machine.

[0037] Acquire an abnormal region image, calculate the area by pixel counting method, and calculate the perimeter; calculate the circularity according to the area and the perimeter, and obtain the geometric feature vector of the abnormal region. Calculate the grayscale co-occurrence matrix for the abnormal region image, extract four statistics of energy, contrast, correlation and entropy; calculate the local binary pattern feature to obtain the texture feature vector. Merge the geometric feature vector and the texture feature vector; use the principal component analysis method to reduce the dimension of the merged feature to obtain the abnormal region feature vector after dimension reduction. Use the reduced dimension feature vector as input to construct an abnormality degree assessment model based on support vector machine; use the RBF kernel function to set the kernel parameter γ and the penalty parameter C. Determine the optimal parameters by the grid search method; classify the abnormal region and quantitatively evaluate the abnormality degree according to the optimal parameters to obtain the assessment result of the abnormal region.

[0038] Specifically, for the extracted abnormal area, the pixel counting method is used to calculate the area, the perimeter is calculated by the 8-connected Moore boundary tracking algorithm, and the circularity is calculated using the area and perimeter to obtain the geometric feature vector of the abnormal area. The grayscale co-occurrence matrix is ​​calculated for the abnormal area image, the distance is set to 1, and the directions are selected as 0°, 45°, 90°, and 135°. The four statistics of energy, contrast, correlation, and entropy are extracted, and the local binary pattern features are calculated at the same time. A circular 8 neighborhood with a radius of 2 is used to obtain a 59-dimensional histogram feature, and the two features are spliced ​​into a texture feature vector. The geometric feature vector and the texture feature vector are merged, and the principal component analysis method is used to reduce the dimension of the merged features. The cumulative contribution rate threshold is set to 95% to select the number of principal components to obtain the feature vector of the abnormal area after dimensionality reduction. Using the reduced eigenvector as input, an abnormality assessment model based on support vector machine was constructed. The RBF kernel function was used, the kernel parameter γ was set to 0.1, and the penalty parameter C was set to 1. The optimal parameters were determined by grid search and cross-validation, and the abnormal regions were classified and the abnormality was quantitatively assessed. The decision function value of SVM was used as the indicator of abnormality to obtain the assessment results of the abnormal regions. For a 1024×1024 pixel textile image, three abnormal regions were detected. The area was calculated by pixel counting method, and the area of ​​the first region was 1200 pixels, the second was 750 pixels, and the third was 980 pixels. The perimeter was calculated using the 8-connected Moore boundary tracking algorithm, and the results were 140, 95, and 120 pixels respectively. The circularity was calculated, and the results were 0.76, 0.83, and 0.85. The gray level co-occurrence matrix was calculated for each abnormal region, the distance was set to 1, and the directions were selected as 0°, 45°, 90°, and 135°. The four statistics of energy, contrast, correlation, and entropy were extracted. At the same time, the local binary pattern features are calculated, and a circular 8-neighborhood with a radius of 2 is used to obtain a 59-dimensional histogram feature. The geometric features and texture features are merged to obtain a 66-dimensional feature vector. Principal component analysis is applied, and the cumulative contribution rate threshold is set to 95%. Finally, 12 principal components are selected to reduce the 66-dimensional features to 12 dimensions. The support vector machine model is constructed, and the RBF kernel function is used. The optimal parameters are determined by grid search to be γ=0.08 and C=1.2. Using 5-fold cross validation, the model accuracy reaches 92%. The three abnormal areas are evaluated, and the SVM decision function values ​​are 1.5, 0.8 and 2.3, corresponding to slight, mild and moderate abnormalities, respectively. It takes 1.8 seconds and meets the real-time requirements.

[0039] S105. Based on the abnormality degree assessment model, a blending ratio defect classifier is constructed, and the warp and weft abnormality features, fiber distribution features and texture features are used as input to train the classifier to identify different types of blending ratio defects.

[0040] The warp and latitude abnormal features are extracted from the textile image, and the warp and latitude abnormal features include the abnormal area, perimeter and circularity; the warp and latitude are evenly divided into several regions, and the abnormal area ratio in each region is calculated to obtain the warp and latitude abnormal feature vector. The textile image is segmented into fiber regions by clustering method, and the number of clusters is set to the expected number of fiber types; the fiber distribution feature vector characterizing the blending ratio is obtained by calculating the area ratio and spatial distribution characteristics of different fiber types. The texture features of the textile image are extracted by gray-level co-occurrence matrix, and the energy, contrast, correlation and entropy statistics are calculated; the texture feature vector is obtained by combining the local binary pattern histogram features. The warp and latitude abnormal feature vector, the fiber distribution feature vector and the texture feature vector are fused by feature cascade method to construct a blending ratio defect classifier based on random forest; the classifier parameters are optimized by multi-fold cross-validation method to obtain a blending ratio defect classification model.

[0041] Specifically, according to the output results of the abnormality evaluation model, the abnormal features of the warp and latitude lines are extracted, including the area, perimeter and circularity of the abnormal area. Ten regions are evenly divided in the warp and weft directions, and the abnormal area ratio in each region is calculated to obtain the abnormal feature vector of the warp and latitude lines. The K-means clustering algorithm is used to segment the fiber region of the textile image. The number of clusters is set to the expected number of fiber types. The fiber distribution feature vector characterizing the blending ratio is obtained by calculating the area ratio and spatial distribution characteristics of different fiber types. The texture features of the textile image are extracted using the gray-level co-occurrence matrix and the local binary pattern algorithm. The distance of the gray-level co-occurrence matrix is ​​set to 1, and the directions are selected to be 0°, 45°, 90°, and 135°. Statistics such as energy, contrast, correlation and entropy are calculated. The texture feature vector is obtained by combining the histogram features of the local binary pattern. The warp and latitude abnormal features, fiber distribution features and texture feature vectors were fused by feature cascade to construct a blending ratio defect classifier based on random forest. The number of trees was set to 100, the maximum depth was 10, and the minimum number of leaf node samples was 5. The classifier parameters were optimized by 5-fold cross validation method to obtain the final blending ratio defect classification model, which was used to identify defect types such as uneven fiber ratio and uneven fiber distribution. For a 2048×2048 pixel blended fabric image, the abnormality assessment model outputs three abnormal regions with areas of 1200, 800 and 1500 square pixels, perimeters of 140, 110 and 180 pixels, and circularities of 0.76, 0.83 and 0.73. The image was divided into 10 regions in the warp and latitude directions, and a 20-dimensional abnormal area ratio feature vector was calculated. The K-means algorithm was used to segment the image into fibers. The number of clusters was set to 3, corresponding to cotton, polyester and viscose fibers. The area ratios of the three fibers were 45%, 35% and 20%, respectively. The spatial distribution characteristics were calculated to obtain a 15-dimensional feature vector. The gray-level co-occurrence matrix was calculated, the distance was set to 1, the directions were 0°, 45°, 90° and 135°, and four statistics of energy, contrast, correlation and entropy were extracted to obtain a 16-dimensional feature. The 59-dimensional local binary pattern histogram features were combined to obtain a 75-dimensional texture feature vector. The three types of features were cascaded and fused to obtain a 110-dimensional mixed feature vector. A random forest classifier was constructed, with 100 trees, a maximum depth of 10 and a minimum number of leaf node samples of 5. 1,000 annotated images were used for training, and 5-fold cross validation was used to optimize the parameters. The final model achieved an identification accuracy of 94.5% on the test set, successfully distinguishing four situations: uneven fiber proportions (37%), uneven fiber distribution (25%), mixed fibers (18%), and normal samples (20%).

[0042] S106. Based on the blended proportion defect classifier and the defect positioning requirement under a vibration environment, different types of defect proportion areas are tracked in real time to compensate for the image displacement caused by vibration, thereby ensuring that the defect detection result corresponds to the actual textile position.

[0043] According to the output results of the blending ratio defect classifier, the defect area boundary is extracted, and the center position coordinates of different types of defect areas are obtained by calculating the centroid coordinates of the boundary pixels. The displacement field between adjacent frames is calculated, and the global motion vector is obtained by statistically analyzing the displacement field. The Kalman filter is used to predict and update the center positions of different types of defect areas, wherein the defect center position and speed are set as state variables, and the detected defect center position is set as an observation variable. The global motion vector is obtained, and the prediction result of the Kalman filter is corrected in combination with the global motion vector to obtain the real-time position of various defect areas after vibration compensation. The affine transformation matrix is ​​calculated based on the real-time position of various defect areas after vibration compensation and the original detection position, and the original image is affine transformed. The image is resampled by a bilinear interpolation algorithm to obtain an image without vibration effects.

[0044] Specifically, according to the output results of the blending ratio defect classifier, the Canny edge detection algorithm is used to extract the defect area boundary, set the low threshold of 50 and the high threshold of 150, and calculate the centroid coordinates of the boundary pixels to obtain the center position coordinates of different types of defect areas. The Lucas-Kanade optical flow algorithm is used to calculate the displacement field between adjacent frames, and the window size is set to 15x15 pixels. The global motion vector is extracted by statistical analysis of the displacement field to obtain the overall displacement of the image caused by vibration. The Kalman filter is used to predict and update the center position of various defect areas, and the defect center position and speed are set as state variables. The detected defect center position is set as an observation variable. The prediction result is corrected by combining the overall displacement obtained by optical flow estimation to obtain the real-time position of various defect areas after vibration compensation. The affine transformation matrix is ​​calculated based on the position predicted by the Kalman filter and the original detection position, and the original image is affine transformed. The image is resampled by the bilinear interpolation algorithm to obtain a stable image that eliminates the influence of vibration, so as to achieve accurate correspondence between the detection results of different types of defects and the actual position of textiles. For a 2048×2048 pixel blended fabric image, the blended proportion defect classifier detects three types of defects: uneven fiber proportion, uneven fiber distribution, and fiber mixing. The Canny algorithm is used for edge detection, with a low threshold of 50 and a high threshold of 150 to obtain the boundaries of the three defect areas. The centroid coordinates are calculated, which are (512, 768), (1024, 1536), and (1536, 512), respectively. The Lucas-Kanade algorithm is used with a window size of 15×15 pixels to calculate the optical flow field and obtain the global displacement vector (5, -3) pixels. The Kalman filter is applied separately to each defect type, with the state vector being [x, y, vx, vy] and the observation vector being [x, y]. The initial state covariance matrix P is set to the unit matrix × 100, the process noise covariance Q is set to the unit matrix × 0.1, and the measurement noise covariance R is set to the unit matrix × 10. After the filter prediction step, correction is performed in combination with the optical flow displacement to obtain the new positions of the three defect areas: (517, 765), (1029, 1533) and (1541, 509). The affine transformation matrix is ​​calculated based on the original position and the new position, and the warpAffine function of the OpenCV library is used for image transformation. The interpolation method selects the INTER_LINEAR bilinear interpolation method for interpolation. Finally, a stable image is obtained. The positions of the three defects in the image correspond exactly to the actual textile positions, with a position error of less than 2 pixels and a time consumption of about 50ms, which meets the real-time requirements.

[0045] The above embodiment is only one of the preferred implementation modes of the present invention and should not be used to limit the protection scope of the present invention. Any changes or modifications that are made to the main design concept and spirit of the present invention and have no substantive significance, and the technical problems they solve are still consistent with the present invention, should be included in the protection scope of the present invention.

Claims

1. A method for detecting textile blending ratio defects, characterized in that: The method comprises: Continuously collect images of the blended textile under vibration, adjust the exposure time and frame rate of the camera device according to the vibration frequency and amplitude characteristics, and obtain a sequence of textile surface images; Preprocessing the obtained textile surface image sequence, segmenting the longitude and latitude areas according to the surface texture characteristics of the mixed textile; Perform frequency domain analysis on the segmented textile warp and weft images, extract the arrangement direction and spacing information of the textile warp and weft by calculating the spectrum energy distribution characteristics, and perform matching and comparison with the preset normal textile warp and weft arrangement template to determine whether the warp and weft are abnormal; The detected abnormal areas of the textile warp and weft are subjected to edge extraction and connected area marking segmentation, the area and perimeter geometric features of the abnormal areas are calculated, and an abnormality evaluation model is established in combination with the textile production process parameters to quantitatively analyze the defect type and severity. Based on the abnormality evaluation model, a blending ratio defect classifier was constructed. The warp and weft abnormality features, fiber distribution features and texture features were used as input to train the classifier to identify different types of blending ratio defects. According to the blended proportion defect classifier and the defect positioning requirements under vibration environment, different types of defect proportion areas are detected and tracked in real time to compensate for the image displacement caused by vibration and ensure that the defect detection results correspond to the actual textile position.

2. The method according to claim 1, wherein: The method of continuously collecting images of the mixed textile in a vibrating state, adjusting the exposure time and frame rate of the camera device according to the vibration frequency and amplitude characteristics, and obtaining a textile surface image sequence includes: A camera device is used to obtain a continuous image sequence of the surface of the blended textile under a vibrating state, wherein the exposure time and acquisition frame rate of the camera device are adjusted according to the vibration frequency and the vibration amplitude; For the continuous image sequence, preprocessing the image by adaptive histogram equalization to obtain an enhanced textile surface image sequence; An optical flow estimation algorithm is used to calculate the displacement field between adjacent frames, and the image is corrected according to the displacement field to obtain a stable textile surface image sequence.

3. The method according to claim 1, wherein: The method of preprocessing the obtained textile surface image sequence and segmenting the longitude and latitude regions according to the surface texture features of the mixed textile comprises: Acquire a textile surface image sequence, and perform filtering processing on the image sequence to obtain a denoised textile surface image; performing contrast enhancement processing on the denoised textile surface image to obtain a contrast enhanced textile surface image; For the textile surface image after contrast enhancement, a Gabor filter group is used to extract texture features to obtain a filter response; Performing principal component analysis on the filter response, selecting a preset number of principal components as texture feature representation, and obtaining a texture feature map after dimensionality reduction; If the texture feature map after dimensionality reduction exists, binarization is performed on the texture feature map; A morphological opening operation is performed on the binary feature map to obtain a longitude and latitude area segmentation result.

4. The method according to claim 1, wherein: The segmented textile warp and weft line images are subjected to frequency domain analysis, the arrangement direction and spacing information of the textile warp and weft lines are extracted by calculating the spectrum energy distribution characteristics, and matching and comparing are performed according to a preset normal textile warp and weft line arrangement template to determine whether the warp and weft lines are abnormal, including: Performing a two-dimensional fast Fourier transform on the segmented textile warp and weft images to obtain a frequency domain representation of the textile warp and weft structural features; Calculate the magnitude spectrum of the frequency spectrum diagram according to the frequency domain representation, and represent the magnitude spectrum using polar coordinates; For the amplitude spectrum represented by the polar coordinates, multiple directions are taken within a preset angle range at preset angle intervals, the energy distribution in each direction is calculated, and a characteristic vector representing the arrangement direction and spacing information of the warp and weft lines of the textile is obtained; The cosine similarity is used to calculate the similarity between the feature vector and the samples in the pre-established normal textile warp and weft arrangement template library, and if the similarity is less than a preset threshold, the warp and weft arrangement is judged to be abnormal; For the longitude and latitude area judged to be abnormal, the Euclidean distance between the feature vector and the template sample is calculated, the abnormal type is judged according to the Euclidean distance, and the specific type information of the abnormal longitude and latitude is determined; It also includes: dynamically adjusting the anomaly detection threshold according to the type of textile and production process parameters to determine whether the current longitude and latitude arrangement is abnormal, using a local affine transformation model to perform geometric correction on the detected local deformation area, and eliminating the spectrum analysis deviation caused by wrinkle factors.

5. The method according to claim 4, wherein: The abnormality detection threshold is dynamically adjusted according to the textile type and production process parameters to determine whether the current warp and weft arrangement is abnormal, and a local affine transformation model is used to perform geometric correction on the detected local deformation area to eliminate the spectrum analysis deviation caused by the wrinkle factor, including: Obtaining a reference threshold value corresponding to the textile type and production process parameter in a preset parameter library; Perform real-time statistical analysis on the current batch of samples, calculate the mean and standard deviation of the latitude and longitude density and spacing of the samples, and generate a dynamic anomaly detection threshold for the current batch based on the mean, standard deviation and benchmark threshold; The textile image is divided into regions, and the eigenvector representing the local texture structure is obtained by calculating the statistics of the gray-level co-occurrence matrix and the eigenvector of the local binary pattern of each sub-region; Using the feature vector to construct a local affine transformation model, mapping the feature vector to the parameters of the affine transformation, estimating the affine transformation matrix parameters, and performing geometric correction on the detected local deformation area; The corrected image is subjected to Fourier transform, the spectral energy distribution characteristics of the image are calculated, the characteristics are normalized and compared element by element with the dynamic anomaly detection threshold, and if the number of elements exceeding the threshold exceeds a preset percentage, the current arrangement of longitude and latitude is determined to be abnormal.

6. The method according to claim 1, wherein: The detected abnormal areas of the textile warp and weft are subjected to edge extraction and connected area marking segmentation, the area and perimeter geometric features of the abnormal areas are calculated, an abnormality evaluation model is established in combination with textile production process parameters, and the defect type and severity are quantitatively analyzed, including: According to the abnormal area image of the textile warp and weft lines, edge extraction is performed to obtain the edge binary image of the abnormal area; Performing connected region segmentation on the edge binary image to obtain a marked abnormal region image; For the marked abnormal region image, calculating the abnormal region area and the abnormal region perimeter; According to the area and perimeter of the abnormal region, the circularity and rectangularity shape factors are calculated to determine the geometric characteristic vector of the abnormal region; Construct an abnormality assessment model, determine the type and severity of the detected defects through geometric feature vectors and weaving speed and yarn tension process parameters, and obtain abnormality assessment results; It also includes: selecting the corresponding size and shape of the structural element according to the complexity of the textile texture, extracting the rough outline of the abnormal area of ​​longitude and latitude based on morphological operation parameters, using the connected region marking algorithm, and introducing local texture features to perform fine segmentation of the abnormal area; extracting the geometric features and texture features of the abnormal area, the geometric features include area, perimeter, and circularity; using the principal component analysis method to reduce the dimension of the features, and obtaining the feature vector of the abnormal area to construct an abnormality degree assessment model based on a support vector machine.

7. The method according to claim 6, wherein: The method selects the size and shape of the corresponding structural element according to the complexity of the textile texture, extracts the rough outline of the abnormal area of ​​the longitude and latitude lines based on the morphological operation parameters, uses the connected area marking algorithm, and introduces local texture features to perform fine segmentation of the abnormal area, including: Obtaining gray level co-occurrence matrix features of the textile image, and calculating a texture complexity index according to the gray level co-occurrence matrix features, wherein the texture complexity index includes four features: contrast, correlation, energy, and entropy; If the texture complexity index is greater than a preset threshold, the corresponding structure element size and shape are selected to obtain the initial value of the morphological operation parameter; The morphological operation parameters are adjusted by using a gradient descent method, and a rough outline of the abnormal area is obtained by performing opening and closing operations; For the rough contour map, a connected region labeling algorithm is applied, wherein the connected region labeling algorithm introduces local binary pattern features based on the scanning method to calculate the feature distance between the central pixel and the neighboring pixels; If the characteristic distance is less than a preset distance threshold, the regions are merged to obtain a preliminary segmented abnormal region image; A region growing algorithm is used to finely segment the abnormal region image segmented initially, and the abnormal region segmented initially is used as a seed point to determine whether the grayscale difference and the texture feature similarity meet a preset growth criterion; If the growth criterion is met, the abnormal region boundary is expanded until the rate of change of the number of pixels at the region boundary is less than a preset stop threshold, thereby obtaining a fine segmentation result image of the abnormal longitude and latitude region.

8. The method according to claim 6, wherein: The method extracts geometric features and texture features of the abnormal area, wherein the geometric features include area, perimeter, and circularity; and uses a principal component analysis method to reduce the dimension of the features to obtain a feature vector of the abnormal area to construct an abnormality degree assessment model based on a support vector machine, including: Obtain an image of the abnormal area, calculate the area using the pixel counting method, and calculate the perimeter; Calculate the circularity according to the area and the perimeter to obtain a geometric feature vector of the abnormal area; The gray-level co-occurrence matrix is ​​calculated for the abnormal area image, and four statistics, namely energy, contrast, correlation and entropy, are extracted; Calculate local binary pattern features to obtain texture feature vectors; Merging the geometric feature vector and the texture feature vector; The principal component analysis method is used to reduce the dimension of the combined features to obtain a feature vector of the abnormal region after the dimension reduction; Using the reduced-dimensional feature vector as input, constructing an abnormality assessment model based on a support vector machine; Use the RBF kernel function, set the kernel parameter γ and penalty parameter C; The optimal parameters are determined by grid search method; The abnormal regions are classified and the degree of abnormality is quantitatively evaluated according to the optimal parameters to obtain evaluation results of the abnormal regions.

9. The method according to claim 1, wherein: The blending ratio defect classifier is constructed based on the abnormality degree assessment model, and the warp and weft abnormality features, fiber distribution features and texture features are used as input to train the classifier to identify different types of blending ratio defects, including: Extracting abnormal features of longitude and latitude lines according to the textile image, wherein the abnormal features of longitude and latitude lines include the area, perimeter and circularity of the abnormal region; The longitude and latitude directions are evenly divided into several regions, the abnormal area ratio in each region is calculated, and the abnormal characteristic vectors of longitude and latitude lines are obtained; The fiber region of the textile image is segmented by clustering method, and the number of clusters is set to the expected number of fiber types; By calculating the area ratio and spatial distribution characteristics of different fiber types, the fiber distribution characteristic vector representing the blending ratio is obtained. The texture features of textile images were extracted using gray-level co-occurrence matrix, and energy, contrast, correlation and entropy statistics were calculated. Combining the local binary pattern histogram features, the texture feature vector is obtained; The warp and latitude abnormality feature vector, fiber distribution feature vector and texture feature vector are fused by feature cascade method to construct a blending ratio defect classifier based on random forest. The classifier parameters were optimized using a multi-fold cross validation method to obtain a blending ratio defect classification model.

10. The method according to claim 1, wherein: The blending ratio defect classifier is combined with the defect positioning requirements in a vibration environment to track different types of defect ratio areas detected in real time, compensate for the image displacement caused by vibration, and ensure that the defect detection result corresponds to the actual textile position, including: According to the output results of the blending ratio defect classifier, the defect area boundary is extracted, and the center position coordinates of different types of defect areas are obtained by calculating the centroid coordinates of the boundary pixels; Calculating a displacement field between adjacent frames, and obtaining a global motion vector by statistically analyzing the displacement field; The Kalman filter is used to predict and update the center position of different types of defect areas, where the center position and speed of the defect are set as state variables, and the detected center position of the defect is set as the observation variable; Obtaining a global motion vector, and then correcting the prediction result of the Kalman filter in combination with the global motion vector to obtain the real-time positions of various defective areas after vibration compensation; An affine transformation matrix is ​​calculated according to the real-time positions of various defective areas after vibration compensation and the original detection positions, an affine transformation is performed on the original image, and the image is resampled by a bilinear interpolation algorithm to obtain an image with vibration eliminated.

Images