Improving global neighborhood structure map denoising approach for digital images

Abstract

This paper proposes a new noise reduction model for digital images. In the proposed model, the intensity similarity between the center pixel and its neighboring pixels within a certain window for constructing a Global Neighborhood Structure (GNS) using Dominant Neighborhood Structure (DNS) maps of central pixels has been measured. The intensity similarity was calculated by using the Canberra Distance measurement equation; where the conventional GNS map approach used the Euclidean distance principle. To evaluate the performance of the proposed model, several noise attacks were imposed on two public image datasets and experimental results demonstrated that the proposed model outperforms the conventional GNS map based denoising technique by exhibiting higher PSNR and SNR values.