基于自适应流形滤波的边缘感知编辑

Edge-aware Editing Based on Adaptive Manifolds Filtering

边缘感知编辑是图像和视频编辑中运用非常广泛的编辑方法。全局优化计算是处理此类问题的最常用手段,但是在处理高分辨率图像视频时存在诸多性能问题。为此,提出一种采用自适应流形滤波的方法,用于处理多种边缘感知编辑。借助自适应流形滤波具有局部边缘感知以及全局影响扩散的特性,实现多种边缘感知编辑的处理效果,同时避免传统的全局优化方法所产生的时间损耗和较高的计算复杂度,得到较好的视觉效果。实验结果表明,该方法具有较好的边缘保持性和全局性,并能够实时处理高分辨率的视频图像。

Edge-aware editing is the most widely used method in image and video editing. Global optimization is commonly used to solve these problems. There are still some problems in processing high resolution image and video. So this paper presents a scheme using adaptive manifolds filtering for edge-aware editing. Instead of using the traditional optimization which may bring the time-consume and the high computational complexity,it utilizes adaptive manifolds filter,which is an edge-preserve filter with the local edge-aware and global influence to propagation the edits. A good visually result can be received with better performance. Experimental results demonstrate the good edge-preserving,globalization influences of the approach.