Machine Vision Based Inspection of Textile Fabrics

Several issues on automatic inspection of textile fabrics are discussed in this paper. To avoid the intense computation for real time inspection, we suggest a parallel pyramid hardware architecture consisting of several channels with CCD cameras in different resolutions working simultaneously. In inspection algorithms, we use a hierarchy of improved BP neural networks which demonstrate a tree structure in the progress of defect detection and classification. Attention is also directed to the system cost and viability under practical industry conditions. Preliminary experiment results show that our suggested hardware and software structuresare very promising.

• Type: Conference
• Year: 1994
• Where: MVA
• Authors: Chuanjun Wang, Chih-Ho Yu
• Keywords: BP Neural Networks, MVA 1994, MVA 2007, Parallel Pyramid Hardware, Real Time Inspection