A HYBRID DENOISING ALGORITHM FOR PATTERENED FABRIC IMAGES

Abstract

The textile industry is mostly concerned with the design, manufacture and supply of fabrics. It is one of the main sources of revenue generated industry. The price of fabrics is severely affected by the defects of fabrics that represent a major risk to the textile industry. A very small percentage of defects are detected by the physical inspection even with highly trained inspectors. Automatic defect detection system will increase the defect detection percentage. It is economically profitable. Fabric images taken for automatic defect detection process. Noise is commonly present in fabric images. Noise degrade the fabric’s value. Noise is very complicated to remove from the fabric images without the preceding facts of noise model. The purpose of noise models reviews is necessary for denoising methods. The sparse denoising methods suggest a class of denoising principle. The use of sparse and over complete representations to develop position of the image denoising system. K_SVD (K means Singular Value Decomposition) is a method that is used to represent the signal of each set, which can create dictionary. The atoms in the dictionary can be equalized with each signal using combination of the atoms in the dictionary. K_SVD can remove the noise present in the color images and gray scale images. It can remove Gaussian noise from an image. Wavelets are often used to denoise two dimensional signals in images. The Discrete Wavelet Transform (DWT) is a denoising method can be decomposed into its low-frequency and high-frequency parts of an image. The low-frequency parts represent the approximate energy of an image, while the three high-frequency parts represent the detailed information of an image, including the horizontal, vertical and diagonal parts. It can remove Gaussian noise, impulse noise, speckle noise. The proposed hybrid denoising algorithm ksvd_dwt is the combination of ksvd and dwt. It can remove sparse representation noise and wavelet signal noise from the images. The performance of this study can be measured in terms of Peak Signal Noise Ratio (PSNR) and Mean Square Error (MSE) values.