Study On Weave Structure Analysis Method Based On Digital Image Technique

As the textile industry competition intensifies internationally, we must improve the quality and upgrade the levels of textiles so as to win in these fierce competitions. Textile fabric is an inherent quality of textile goods. And the analysis of textile fabrics and imitation of the sample design is an important measure for improving the quality of fabrics. At present in the textile industry, the fabric analysis and identification mainly depends on manual operation with experiences or the professional tools. This manual way of fabric structure analysis and parameter extraction by experts, although authoritative, but needs high operating requirements and is not easy to master either, let alone the tedious and monotonous working period. Therefore, it is urgent to develop new better technical lines and algorithms for analyzing and recognizing parameters automatically to replace the work nowadays.The development and application of digital image processing and artificial intelligence in many areas has greatly alleviated the workload and improved the efficiency of work to shorten the production cycle and enhanced the work and the quality of the product as a result. So fabric analysis also needs to use computer technology to help people to complete the work. Therefore, it is necessary to find much more advanced ways to recognize the yarn parameters and analyze the organizational structure of the fabric automatically instead of manually.This paper, on basis of study on digital image processing technology of automatically extracting and analyzing fabrics parameter, finds out its own working flow and technical lines to accomplish this job automatically and effectively.The first step is the acquisition of high-quality transmission of fabric picture with a CCD camera and transmission source of continuous variable zoom stereomicroscope. For the characteristics of the organization points in this transmission image is easy to separate. In the image preprocessing stage we choose the index histogram transform method to extend high gray value, and use Gaussian filter to remove fabric surface noise, such as hairiness, so as to facilitate the follow-up work.This paper presents a new method to identify the characteristics of textile interlacing point, that is the artificial identification method, which is nowadays considered as a better way to analyze the characteristics textile interlacing point with more reliability. Firstly, as to the whole points containing information of its neighbors', the characteristics is identified by comparing the characteristics between the point and those around it. Secondly, for the incomplete points, the already known pattern is applied to infer the characteristics of this kind of points. And finally, the latitude and longitude density of the fabric is indirectly calculated by taking the advantage of the relationship between the fabric image magnification size and pixel width.According to the different image segmentation methods, two different specific schemes are proposed in this paper. The first one is for the simple elementary weave. And the segmentation method that is to analyze the brightness projection on the basis of the fabric reflection image is applied. Then a series of analysis steps are designed based upon the after segmentation nature of the points. Also the experimental verifications have shown good identification results. The second is about the relatively complex fabric images and a form of watershed segmentation method is introduced. In the earlier segmentation stage of this part, the morphologically gradient reconstruction method is adopted to cut the local extreme value so that each point with only one extremum, which can also avoid the over-segmentation phenomenon in the latter period of segmentation. Then the following step is to segment the structure points by means of watershed image segmentation method. And the segmented point regions are correct and remain the original shape the size. In the end, the identification and the fabric density calculation methods mentioned above are used for further analysis. Due to the adoption of the relatively common segmentation method, this scheme can gain better identification outcomes for various fabrics. The algorithms and technique lines provided above have certain theoretical value and worthiness for reference, especially the pattern draft recognition method has some innovation in this field. Even for more complicated derivative weave fabric, it also has a good recognition result. The color and weave effect recognition method is also useful and practical. In a word, the method discussed in this article is valuable for popularizing in our practical working field.