Multidimensional Image Processing via Fast Feature-Preserving Filters

多维图像在自然科学和工业工程中已成为观察或测量数据的热门。在本研究中，我们发展了一个多维影像的快速特征保持滤波器近似演算法。所开发的算法的应用包括降噪，频率分解，区域分割，和图像的特征提取。过滤由5133体素组成的图像只需要3分钟，我们的算法与3.2 GHz的PC，其中精确的过滤需要大约30年，这样巨大的数据。

Multidimensional images had become popular in natural science and industrial engineering for observed or measured data. In this research, we have developed a fast approximation algorithm of fea-ture-preserving filters for multidimensional images. Application of the developed algorithm includes noise reduction, frequency decomposition, region segmentation, and feature extraction of the images. Filtering the image consisting of 5133 voxels took only 3 minutes by our algorithm with 3.2 GHz PC where the exact filtering requires approximately 30 years for such huge data.