A late adaptive graph-based edge-aware filtering with iterative weight updating process
Hamidreza Sadreazami, Amir A. Asif, Arash Mohammadi · 2017
In this work, we propose a parallelized graph-based framework for lowpass and highpass edge-aware filtering for detail manipulation (smoothening/boosting). Our proposed filter abstracts images through simplifying their visual content while preserving edges and emphasizing most of the perceptually important information. The proposed filtering framework is realized by using the graph similarity and Laplacian matrices to obtain smoothened image at each layer. The resulted smoothened images are iteratively treated as inputs to the next layer of the filtering framework and the weights are updated accordingly. The efficacy of the proposed image abstraction method is confirmed by conducting simulations. It is shown that the proposed method provides abstracted images having higher quality than those resulted from the other existing works.