TY - JOUR
TI - Two stages non-local means algorithm based on wavelet transform
AU - Ma Hongjin
AU - Zhang Chenxin
AU - Zhang Weiwei
JN - Thermal Science
PY - 2025
VL - 29
IS - 3
SP - 2015
EP - 2021
PT - Article
AB - This paper presents an innovative wavelet transform-based two-stage non-local  means algorithm that offers a new approach to image de-noising. It first  decomposes the noisy image into four sub-images with different frequencies,  which provides a more detailed analysis. Then, it fuses three high  frequency sub-images into one high frequency sub-image, which enhances the  image quality. The low-frequency sub-image and the fused high frequency  sub-image are de-noised using the non-local means algorithm, which is a  highly effective and robust de-noising technique. Furthermore, the de-noised  low-frequency sub-image and the high frequency sub-image are merged to form  the initial de-noised image. Finally, the initial de-noised image is  de-noised again using the non-local means algorithm to produce the final  de-noised image. The proposed algorithm is able to effectively remove noise  while preserving edges and details.