One-step singular correlation and most relevant vector filtering for color images
Chao Liu · Journal of Computer Applications · 2012
Concerning the fuzzy and unclear details after color image denoising,the correlation properties of the adjacent pixels as well as the correlation among the color channels were analyzed.First,one-step singular correlation detection algorithm was used to detect the noises of each layer of the pre-treatment color image,and then the most relevant vector median was used to fill value of the noises.Finally,the color image filtering processing was realized.The experimental results show that this method not only accurately detects the salt-pepper noises,but also well restores and protects the original information such as the edge details.The color image filtering accuracy and performance criterion such as Peak-Signal-to Noise Ratio(PSNR) are further improved.