Computer tomographic reconstruction from partial view projections

Conventional algorithms for tomographic reconstruction require a discrete set of projections equally spaced over the full angular range of 180 degrees for a parallel beam or 180 degrees + fan angle for a fan beam. A crucial point is the availability of a complete set of data. However, in some cases this requirement cannot be fulfilled. This happens, for example, in case of large objects, such as a pipeline or a wall, or when part of the projections was lost. In order to perform a reconstruction of the image from the partial set of projections, the data must be preprocessed. Several algorithms have been reported in literature, but the majority of them require a-priori knowledge of the shape and composition of the sample or they present a high computational cost. Recently a new algorithm has been proposed. It allows to recover the lacking projections without any a-priori assumption and with a relatively low computational cost. It is based on a morphing technique, which affords in general terms the problem of curve matching and has been specialized to the case of tomographic reconstruction. It has been applied to medical (transmission) tomography. In the present work such algorithm is applied to micro-tomographic measurements of different types (transmission, Compton and fluorescence tomography), which have been performed on industrial samples.