TV-Based image reconstruction of multiple objects in a fixed source-detector geometry

Abstract

In this paper, we study how to reconstruct multiple objects tomographically from their overlapped x-ray projections. We propose a new rotation-translation scanning mode for a fixed source-detector geometry, in which the objects are individually translated periodically while they are rotated during the data acquisition. The reconstruction scheme performs SART reconstruction and TV minimization alternatingly. The scanning parameters are evaluated in a series of numerical experiments. Promising results are obtained in the cases of two and four objects respectively. The major advantage of our approach is that the scanning time can be reduced by about 46% (for two objects) and 67% (for four objects) respectively, improving the imaging facility throughput significantly.