Bottle image de-noising using adaptive threshold based on nonsubsampled contourlet transform

Non-subsampled Contourlet transform(NSCT) with translational invariance was applied to image denoising, which could capture the intrinsic geometrical structure of bottle image. After the scale of NSCT is determined, using NSCT to transform the noisy image of glass bottle. A low frequency component and some high frequency components will be obtained. Through using the coefficients of high frequency components from different directions of the same scale, adaptive thresholds will be got. Combined with hard threshold function, these high frequency components will be treated and new high frequency components will be acquired. Using inverse non-subsampled Contourlet transform to deal with the low frequency component and new high frequency components, a de-noised image of glass bottle is obtained. The experimental results show that the method can get higher SNR value of de-noised bottle image and better visual effect compared with other methods.