基于非局部邻域的图像重新着色

Image recoloring based on nonlocal neighborhood

提出了一种新的基于非局部邻域的图像重新着色方法。首先提出了非局部彩色线性模型,然后构建了高维特征空间查找非局部邻域,最后将提出的非局部彩色线性模型最优化问题归结于求解稀疏矩阵。该算法继承了全局和局部邻域色彩传播方法的优点,既能实现全局色彩传播,即使当需要重新着色的像素离涂色线条距离较远时,也能实现局部或者直接的选择控制。与现有的采用全局或者局部邻域色彩传播的方法相比,该算法仅需要输入少量的用户涂色线条即可产生高质量的重新着色效果。

A novel nonlocal image recoloring approach is proposed. Firstly, a nonlocal color linear model optimization assumption is designed. Next, the nonlocal principle by computing K nearest neighbors in the high-dimensional feature space is implemented. Finally, the nonlocal color linear model optimization can be attributed to solve a sparse linear system. Our nonlocal color linear model optimization inherits the advantages of global and local color propagation methods, which can propagate color cues in global manner, also can propagate color to relatively far from the provided color line, while our approach can provide the user with good local control. Compared with the state-of-the-art methods, our approach can produce higher-quality results with only a small amount of user interaction than those only consider local propagation or global propagation approaches.