Research On Automatic Identification Of Fabric Weave Patterns Based On Image Processing

Traditional weave recognition which mainly rely on manual detection analysis have a certain authority, but been time-consuming, inefficient, and apparently cannot meet the current demand of textile industrial automation and high efficiency. Digital image processing technology for the automatic identification of fabric weave patterns can not only simplify the operation process of the traditional textile industry, greatly improve work efficiency, and have great significance on automation and intelligence of textile enterprise.So, this paper mainly studies how to use image processing technology to achieve the automatic identification of fabric weave patterns. This article mainly include the following:the fabric image automatically tilt correction, automatic identification fabric warp and weft density, weave point feature extraction and selection, automatic identification and verification of the weave patterns.Firstly, In order to improve the precision of identification of weave patterns, this paper gives the corresponding tilt detection and correction algorithm for tilt image and tilt yarn. Then, using the gray projection method to achieve the positioning of the yarn of the fabric image, thereby completing the automatic detection of the density of the fabric yarns.In order to achieve the automatic identification of the crossed points, it should extract the features of the point image, and then make unsupervised clustering recognition on the crossed points according to the features. This article extract texture features of the crossed point image based on gray level co-occurrence matrix(GLCM). Meanwhile, in order to optimize features and reduce the amount of computation, principal component analysis must be implemented to extract the most significant child feature.Nuclear fuzzy clustering algorithm is using to classify the crossed point, and determine the clusters’attributes of the crossed point according to the directional features of the texture. Due to the possibility of the individual crossed points’ misjudgment, this paper uses the standards weave patterns’database to automatically check the results of the preliminary identification. So the identification method has some fault-tolerant mechanisms, making the final identification result is more accurate.The experimental results show that the algorithm can identify the basic fabric weave patterns correctly, and has a larger theoretical significance and application value.