OPTICAL TOMOGRAPY WITH MODIFIED RADIAL BASIS FUNCTION NETWORKS

摘要

We propose a new tomography reconstruction algorithm that uses modified radial basic function (RBF) networks. To reconstruct the density distribution, conventional tomography techniques need many equal-angle projection data. However in various cases, few data can be taken. To make up a lack of the information of the object in the reconstruction, we propose the modified RBF networks that are constructed by membership functions. Turning the membership function, the object can be reconstructed. The proposed algorithms have a simple structure and require a few projection data. The RBF networks-based tomography shows stable image reconstruction for limited projection data.