Research On Automatic Recognition Technology Of Woven Fabric

With the rapid development of artificial intelligence and robotics technology,the level of automation in textile industry is also improving.Traditional fabric performance testing methods is overwhelmingly dependent on professionals.Despite the high accuracy and authority of this analysis method,which is dominated by professional experience.However,the manual detection method is time-consuming,labor-consuming and extremely inefficient.We are now advocating efficient economy.This method which have lagged behind undoubtedly hinders the advancement of automation and intelligence in textile industry.In order to overcome the shortcomings of traditional fabric testing methods,this paper proposes a new algorithm to identify fabric pattern based on digital image processing and pattern recognition technology.The main contents include:fabric image acquisition and preprocessing;fabric image structure automatic extraction;warp and weft yarn density detection;fabric pattern identification.First of all,Scanner is used to scan fabric samples to collect high-resolution image.Then the fabric image is preprocessed.In order to extract the fabric structure automatically,the color and texture features in the local range are extracted from super-pixels of fabric image.Principal Component Analysis(PCA)is used to reduce the dimension of features,and then the features which are reduced dimension are clustered by K-means to segment the fabric image.This paper realizes the automatic extraction of fabric structures based on fabric segmentation graph.In order to segment weft and detect weft density,firstly,the gray-scale integral projection curve of fabric image is calculated,and the weft segmentation and density detection are realized by multi-scale filtering.Finally,to recognize the fabric pattern,the concept of weft phase difference is proposed,then the fabric pattern is recognized and the warp density is measured based on weft phase difference.Experiments show that the algorithms studied in this paper have good performance of fabric pattern recognition and warp and weft yarn density detection.