Research On The Quality Analysis Of The Fabric Surface Based On Image Processing

In recent years, industrial production pay more and more attention to the quality of products, products requiring high reliability and accuracy. So only high-tech production environment is not enough, have a strong capability of product quality control and testing is also important. With the development of computer technology and artificial intelligent technology, automation, such as product surface test, measurement and identification and classification come into reality and widely used. The application of Fabric quality inspection falls into this category.Recurring to three toolboxes:MATLAB image processing toolbox, wavelet toolbox, neural networks toolbox. After image processing and BP neural network recognition the fabric defect detection. This research of fabric surface quality detection implement evaluation quality automation and quality control automation.In the first part, this paper introduce the significance of the fabric surface quality detection, the fabric surface quality detection methods, the Fabric and other related knowledge, and the method of fabric quality evaluation and quality control.Second part, against disadvantages of the traditional fabric defect detection, using image processing, fractal and wavelet technology in the fabric defect detection.Again, based on MATLAB, American four point standard as for rating standards, using BP neural network combined with genetic algorithm to evaluation the fabric surface quality by computer.Finally, summarized the detection of the defect, it is concluded that the fabric surface defect formation reasons and solutions, and establish corresponding relations between the defect and process that forming defect, the computer control implementation through the neural network, and quick control production process in the later.This study based on image processing, focuses on the fabric surface quality evaluation and the fabric surface quality control. In the experiment, white cloth as the research object, study the several common kinds fabric defects of woven, the correct recognition rate reached97%.