Automatic Defect Segmentation for Plain Woven Fabric Images

摘要

An automatic defect segmentation method is investigated for plain woven fabric images. The original RGB images are converted into grayscale images by weighted averaging. In order to distinguish the defects from background, the converted images are enhanced using gray level adjustment and further filtered by ideal lowpass filter. Based on comparison, Roberts operator is chosen for edge detection and segmentation of the filtered images. Region of interest is finally segmented from the background which is verified by the experiments.