Recognition of Woven Fabric based on Image Processing and Gabor Filters

In this paper, we present a novel automatic method for recognition of woven fabric. The improved image processing method is suited for texture segmentation and the frequency domain analysis method is fit for detection of texture orientation in fabric texture. Their excellent performances are put together to realize the recognition of woven fabric and get the corresponding notation image. This method combines the improved image processing method with the frequency domain analysis method. First of all, It completes preliminary classification of yarn crossing points through image pre-processing and K-means clustering. Secondly, the preliminary classification comes to be accurate yarn segmentation by applying gradient accumulation of gray levels approach. Finally, using Gabor filters based analysis on texture orientation (vertical or horizontal) features, the states of yarn crossing points are automatically determined. The experimental results show that the algorithm proposed in this study can locate and recognize the cross points in common fundamental fabric weaves and effectively reduce the lighting effect from fabric image capture.