The Image Segmentation of Weft Knitted Fabric Defects by PCNN-based Algorithm

This paper proposes a PCNN-based algorithm and applies it to automatic segmentation of weft knitted fabric defects image. The segmentation method in this article consists of two main stages: preprocessing stage which equalizes the shadow area cased by Illumination device and segmenting stage which is implemented by the proposed Pulse Coupled Neural Network (PCNN). In order to verify the segment result, we apply another modem segmentation algorithm: Cellular Neural Network (CNN) to compare the segment effect with PCNN. The interlock fabrics with the defects of hole, course mark and dropped stitch are used as experiment materials. The experiment proved the performance of PCNN method is better than CNN, which gives better segment effect and the processing speed is much faster.