Fabric deconvolution wiener filter and feature extraction regionprops for locating defects

Abstract

The core of this paper is to locate the defects in fabric by using image processing system. Automatic visual inspection methods are genuinely necessary in Textile industry, particularly when quality control of item enters into the industry. In the manual inspection method just less measure of defects are being identified while Automatic inspection method will increment to most extreme number. Here the rule detection used to distinguish the defects in fabric through deconvolution wiener filter algorithm. The deconvolution can be done with early known PSF (Point Spread Function) value. This will remove the unnecessary noise in images and producing a noiseless enhanced image. The given image is binarized and thresholded to get the desired output. After the filtering process is over the morphological transformations are done to extract the defected portion in the fabric. Then the features are extracted through the method regionprops and GLCM (Gray Level Co-Occurrence Matrix). Finally by extracting the features the classification of textile defects are done. The Experimental result shows that accuracy rate high compared to existing methods.