Visualization of the electrical tomographic data distributions based on the self organization map neural tool

Electrical tomography (ET) techniques have greatly been developed for visualizing the distributions of materials in an industrial process, since they offer certain advantages over other tomography modalities, such as low cost, rapid response, no radiation and being non-intrusive. But so far the natural imaging mechanism remains unknown to some extent so that many existing ET algorithms suffer from the uncertainty or fuzziness problems. In this paper the electrical tomographic data are mapped into a vector space, and thus the cluster structures hidden in the electrical tomographic data are recovered based on self organization map tool. Two groups of experiments are performed to validate our proposed methods. These experiments show that the high-resolution ET images must have a clear cluster structure, while the low-resolution images corresponds to fuzzy and vague data distributions.