Abstract:

In view of the fact that the design of jacquard warp knitting patterns is time-consuming, this paper proposes a rapid segmentation method to divide the multi-textural regions of jacquard warp-knitted fabric which could be used for automatic identification of fabric patterns to improve the efficiency of design. After pretreatment, the images scanned were decomposed by a two-layer two-dimensional wavelet transform and the standard deviations of five channels were extracted as the eigen values. Then, after giving the cluster centers, a multi-channel clustering was made combined with a K-means clustering algorithm. Finally the removal of noises caused by classification errors was needed , after which an accurate identification image was obtained. The experiments show that this method can achieve automatic texture segmentation of jacquard warp-knitted fabric with more than three textural regions. The identification results have high regional consistency, and the segmentation accuracy is up to 92%. The method can also adapt to a variety of mesh regions. Besides this, the approach is fast and can simplify craft personnel’s traditional process of pattern tracing classification when it is combined with CAD. Through this method, the efficiency of jacquard warp-knitted product designing can be improved a lot.