Study On Pattern Recognition And Automatical Notation Creation Of Weft Jacquard Knitted Fabric Using Digital Image Processing Techniques

Digital image processing is a technology focusing on the digital image signal, which rapidly progressed in recent years. Image segmentation which subordinates digital image processing takes significant impact on analyzing knitted fabric, especially on pattern segmentation and recognition of weft jacquard knitted fabric. The aim of this paper is to improve the image recognition of multicolor weft jacquard knitted fabric image from grey to 24 bits colors. It creatively applies mean-shift algorithm which is normally used on pattern recognition to automatic segmentation of weft jacquard knitted fabric image.The author firstly obtained the required fabric image via digital camera, and then utilized median filtering and mathematical morphology processing to pretreat the fabric image. After the pretreatment, the mottle and noise had been well controlled.On the course of image segmentation, this paper explained the principles and the methods of the image segmentation which were based on k-mean clustering algorithm, threshold algorithm and mean-shift algorithm respectively, then compared these three image segmentation algorithms by segmented images which base on total 6 kinds of fabrics varying from 2 colors to 4 colors. It was found out that the best way to segment colored weft jacquard knitted fabric was mean-shift algorithm.In conclusion, this paper created colored knitting notation by the image which processed by mean-shift algorithm of the weft jacquard knitted fabric.On the platform of Windows XP Profession, this experiment used Visual C++ to complete the pretreatment and code the software of these three segmenting algorithms, and realized the object to do segmentation of fabric image by the software that had been coded. At last, on the platform of Matlab, it created colored knitting notation automatically by the image which processed by the software based on mean-shift algorithm of the weft jacquard knitted fabric.In addition, in order to achieve the paper's object, it also compared these three ways through their algorithms, realizing processes and segment results. It solved problems and modified errors of the program during the comparison. At the end of this paper, the author pointed out three opinions to improve the automatic analysis system for weft jacquard knitted fabric.This paper dealt with the weft jacquard knitted fabric in the given image area, and created colored knitting notation automatically according to the fabric image, but its effect was not so perfect when the fabric structure was not well knitted and its color was not well scanned. this should be deeply studied later.