The Study Of Segmentation And Connection Method On Jacquard Pattern

In recent years,the image processing techniques are gradually finding their applications in textile field. The textile quality examination, once completely relying on Manual Operation Inspection, is now transferring to Intelligence Robot Inspection (IRI). However, IRI is encountering some difficulties because of the complexity of the Warp Knitted Products' (WKP) structure and the uncertainties in the production process. By now, there has not been a universal arithmetic applied in WKP images segmentation.WKP images segmentation is a texture image segmentation, a type of image processing technique. Because of the irregularity and no pattern of curves and graphics in the KWP images, it is difficult to apply the segment technique. In the paper, we used the Mathematical Morphology Method which is based on the differences between the WKP textures and their background to segment the WKP images and had a good result. This is a new application of image processing techniques in the textile field.The following steps were included in this paper:1. We used the micro distance mode of digital camera or the CCD microscopically imaging system to obtain the WKP images. We then determine the range of magnification after many tests.2. Using the Matlab, we pre-processed the images by adopting the techniques of make gray, Histogram Equalization, Median Filter, and Top-hat Transformation to reduce the noises of the images.3. After binarizating the images, we drew the texel area histogram of multiply-connected domains. Based on the histogram, we obtained the Texture Multiply-Connected Threshold, which is used as an operand of Mathematical Morphology method.4. Finally, we segmented the WKP images and got the boundary of WKP areas by using the Mathematical Morphology Method. We made universal programs to achieve automatic segmentation of WKP. 5. In the algorithm, a new algorithm for tip-tilt correction was made. The experimental results show that the algorithm is effective on the regular Jacquard texture image.6. To obtain a large area for the fabric image, Improved template-matching algorithm applied to the image mosaic obtained good results .In this paper we only studied the segmentation for two types of WKP images. Some limitations are identified. Future study should explore how to quickly and accurately segment the Multiple Textures Images.