Anzeige
Erschienen in:
2017 | OriginalPaper | Buchkapitel
An Efficient Learning-Based Bilateral Texture Filter for Structure Preserving
verfasst von : Zhe Zhang, Panpan Xu
Erschienen in: Next Generation Computer Animation Techniques
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Abstract
Images generally contain rich visual information that can be decomposed into edges and textures. Particularly, human beings are more sensitive to edge information. However, it is difficult to separate edges from textures since they are tough to be differentiated by computers. In this paper, we provide a novel learning-based bilateral filter to effectively remove textures from the image. Firstly, edge features are extracted as the guidance image through structured forests learning method. Then the guidance image with very rough edge features needs to be optimized. Finally, the joint bilateral filter is applied to produce the filtered result according to the input image and the optimized guidance image. Comparing with some previous approaches, our method is simpler and faster, as well as more effective in preserving edge structures and removing textures.