Edge-Aware Editing Based On Adaptive Manifolds Fiitering

Edge-aware editing methods, including colorization, edit propagation, segmentation and matting, are widely used in Computational Photography and Image Processing applications. In traditional image editing environments, such as Photoshop, it usually requires the users to precisely select the area in which they want to edit pixel colors. However, the precise selection is difficult, especially for something like hair. Therefore, the researchers have proposed some methods for edge-aware editing via global optimization. These methods need to solve large-scale optimization problems to get better results, and consequently result in performance limitations.This paper sythetically analyzes both the property and development of edge-aware editing and edge preserving filtering, and presents a scheme which employs adaptive manifolds filtering for edge-aware editing.We utilize adaptive manifolds filter, which is an edge-preserve filter with the local edge-aware and global influence to propagation the edits, to realize many kinds of application, such as colorizaion, edit propagation, segementation and matting. We also compare the results to other approach and analyse the performance.By using adaptive manifold filtering for edge-aware editing, we avoid time-consume and high computational complexity of traditional optimization. Users could get better results with simple operation. The experimental results demonstrate the good edge-preserving, globalization influences and user-friendliness of our approach.And we could also process high-resolution image and video in real time.