| In the textile industry, the structure of fabric analysis is usually analyzed by workers so the analysis procedure is very time consuming and inefficiency. In order to improve work efficiency, Digital Image Processing is utilized to achieve automatic recognition of fabric structure parameters. It plays an important role to improve automatization and artificial intelligence in textile production. Nowadays recognition of fabric structure parameters commonly uses two-dimensional Fourier Transform.This thesis proposes a new automatic recognition algorithm of fabric structure parameters based on mathematics morphology, which includes weft and warp density, fabric cycle and structure. In the algorithm, mathematics morphology operation is applied and the actual arrangement features of yarn in different fabrics is considered; while different fabrics are deal with different Image processing algorithm. Firstly, after pre-processing of fabric image, histogram features are extracted which are used to recognize fabric structure types. Secondly, Combining Cluster Analysis method, an adaptive structure operator choosing algorithm is submitted which can perform automatic select of suitable structure operator for fabric image. Thirdly, morphology operation is used on the fabric images using these structure operators in order to obtain weft arrangement image and warp arrangement images. Post-processing and analysis are implemented on the weft arrangement images and warp arrangement images and finally the structure parameters of fabric are gotten. Because the characteristics of the fabric are considered and the results of image processing can be changed by directly adjust the shape of structure operator in mathematic morphology, the identification accuracy of structure parameters is improved and the defects of traditional method are avoided.The algorithm is tested on fabric images and the experiment results show that it is efficient and has higher recognition rate. |
PDF Full Text Request