Graph-based methods for simultaneous smoothing and sharpening of color images

Abstract

[EN] In this work we introduce an image characterization of pixels based on local graphs that allows to distinguish different local regions around a pixel. This separation also permits us to develop a method for determining the role of each pixel in a neighborhood of any other, either for smoothing or for sharpening. Two methods for simultaneously conducting both processes are provided. Our solution overcomes the drawbacks of the classic two steps sequential smoothing and sharpening process: enhancing details while reducing noise and not losing critical information. The parameters of the methods are adjusted in two different ways: through observers visual quality optimization and with an objective optimization criterion. The results show that our methods outperform other recent state-of-the-art ones.