Data-driven image stylization using graph-based filtering

In this work, we consider the problem of image abstraction and stylization. A graph-based framework is proposed to render real images into painterly-style ones and create a simple stylized format from color images. The goal is to abstract images by simplifying their visual content while preserving edges and emphasizing most of the perceptually important information. To this end, the low-contrast regions of an image are first smoothened using iterative graph filters in both the vertex and spectral domains. The abstracted luminance channel is quantized and sharpened using an iterative highpass graph filter in the spectral domain. The effectiveness of the proposed graph-based image stylization method is verified through simulations. It is shown that the proposed method can yield significantly better visual quality for stylized images as compared to other existing works.