基于局部极值插值的保边缘图像平滑算法

EDGE-PRESERVING IMAGE SMOOTH ALGORITHM BASED ON LOCAL EXTREMA INTERPOLATION

提出一种新颖的基于局部极值的强边界与高对比度纹理的定义。使用这个定义提出一种保留边界的平滑算法,通过将算法迭代运用在基层上面进行多尺度图像分解。同时算法可以抽取图像高对比度纹理,但同时可以保留图像的强边界。该方法可以高效地使用在多尺度图像分解的各种应用中,如高动态范围图像的色调映射、图像细节增强、图像去噪。

This paper raises a new definition of strong edge and highcontrast texture which is based on local extrema. By using this defini tion we propose a smoothing algorithm which can preserve edges, by iterativety applying the algorithm to basic level, the multiscale image de composition is achieved. This algorithm can manage to extract highcontrast texture and preserve strong edges at the same time. The method can be effectively used in a lot of application of multiscale image decomposition such as HDRI tone mapping, image details enhancement, image denosing, etc.