Welcome to the Qinsun Instruments Co., LTD! Set to the home page | Collect this site
The service hotline

Search


Related Articles

Product Photo

Contact Us

Qinsun Instruments Co., LTD!
Address:NO.258 Banting Road., Jiuting Town, Songjiang District, Shanghai
Tel:021-67801892
Phone:13671843966
E-mail:info@standard-groups.com
Web:http://www.qinsun-lab.com

Your location: Home > Related Articles > Texture structure analysis technology used in fabric defect detection method

Texture structure analysis technology used in fabric defect detection method

Author:QINSUN Released in:2023-02 Click:130

The effective detection of fabric defects is of great importance for textile companies to improve product quality and reduce product costs. At present, the detection of defects is still carried out by the human eye, but the detection cost for the human eye is high, the speed is slow, and the effect is poor, and it cannot detect more than 80% of the defects [1 ]. Therefore, the automatic detection method of dust defects has always been an important research hotspot, and many automatic detection methods have been proposed [1-14]. For example, literature [2-4] proposes the method of using a gray-scale co-occurrence matrix or gray-scale difference matrix; literature [5] proposes the method to use Fourier transform; literature [6-9] uses the method of wavelet transformation or tabloid packet transformation; literature [10, 11] suggested using the Garbor filter method. Because of the different types andforms of fabric defects, however, it is still an important research content to detect defects effectively.
Defect detection is actually a texture segmentation and identification process, because the texture structure of the defect in the fabric is different from normal fabrics, so they can be detected. From the perspective of image processing, defects can be divided into gray scale mutation defects and structural mutation defects [12]. The gray scale of the defects in the gray scale mutation type differs from that of normal tissues, such as oil stains, holes, etc.; the gray scale of the defects in the structural mutation type of defects is less different from the gray scale of normal fabrics, only the space between pixels There have been major changes in the relationship, such as the Merge Classic, Song Classic, etc. Some defects are both, such as warp shrinkage, defects of side supports and so on. Since the normal tissue picture is ais regular and ordered texture, this paper proposes a fabric defect detection method based on texture structure analysis, which not only can effectively segment gray scale mutation defects, but also has better segmentation of structural mutation defects. effectiveness and robustness of the method are verified by experiments.

1 Algorithm for material defect detection
The flowchart of the tissue detection algorithm is shown in Figure 1 and Figure 2 shows the processing effect of each step of the algorithm
1.1 Zero-mean image enhancement
/>Zero-mean image enhancement is an important step of the preprocessing algorithm in this algorithm. It can effectively eliminate the difference in light and shadow caused by the difference in light intensity during the image acquisition process, and facilitate the processing of the following steps. The construction method ofthe zero average image is: first divide the image into 8×8 sub-windows.
1.2 Constructing texture primitive templates
The fabric is formed by interweaving warp and weft yarns according to certain organizational rules, which has a certain periodicity and belongs to a regular and ordered texture image. In the normal fabric image, the gray value of the corresponding pixel in each texture primitive is relatively close, but in the defect area, whether it is a gray level mutation defect or a structural mutation defect, the gray value of the corresponding pixel in the texture primitive is the same as the normal fabric image. The presence of substances varies greatly. Based on this idea, here we construct a texture primitive template, first determine the size of the texture primitive, and use the following formula to calculate the image level, the autocorrelation function in the vertical direction.
1.3 Improving the defect area
The automatic detection of dust defects should continuously weaken the background, that is, the information about the normal texture area, and at the same time emphasize the information about the defect area, to finally detect the defect. Here we propose a method to improve the defective area using the texture primitive template. Since the proportion of the defective area is small, in the non-defective area, the distance between each texture primitive and the template primitive shown above is has been calculated. The difference is small . As for the defective area, due to the sudden change of the gray value or texture structure, it is quite different from the primitive template.
1.4 Calculation of Local Asperities
Because the appearance of defects destroys the global consistency of the normal texture image, here we use the method of calculating the local asperities of the image to locate the defects.
1.5Segmentation of defect image
For the segmentation of defect image, the automatic threshold segmentation method of the Otsu method is used here. Otsu\'s method is considered one of the best automatic threshold segmentation methods. The basic idea is to divide the image pixels into two categories based on a threshold and determine the optimal threshold by maximizing the variance between the two categories obtained after division [14].

2 Experimental analysis
To verify the correctness of this algorithm, we performed segmentation experiments on some defective images, and the segmentation results are shown in Figure 3. The experiments show that the algorithm in this paper has good detection capability for both gray level mutation defects and structural mutation defects. To verify the robustness of the algorithm, the impact of noise, image rotation and image scaling on the algorithm is shown below.analyzed, as shown in Figure 4, (b), (d) and (f) are the final segmented images using the algorithm presented in this paper, respectively. From the perspective of the segmentation effect, the algorithm has some robustness to image noise , image rotation and scaling.

3 Conclusion
This article proposes a new method to detect defects, based on the characteristics of the fabric texture itself. Because the texture of the fabric has good regularity and order, the primitive texture template can be extracted according to the characteristics of the regular texture. In addition, the appearance of the defect image will inevitably destroy the gray value distribution of the texture primitive, so that the difference between the texture primitive and the primitive template can be used to emphasize the defect information. Finally, the defect is located by calculating the disparity of the image and the defect image is segmentedcreated by using the Otsu method to automatically select the threshold. In the experiment, we segmented and detected a variety of common images of dust defects, and found that the algorithm has a good segmentation effect on the gray level mutation and structural mutation defects; through the noise, image rotation, and scaling experiments, it has better verified the robustness of the algorithm. In addition, because the algorithm does not include orthogonal transformation and complex graphics processing, the algorithm has high time efficiency and is more suitable for real-time automatic defect detection.

More about substance testing instruments: http://www.abrasiontesters.com/productlist/list-5-1.html

Prev:

Next: