Computerized Tomography

A considerable amount of literature is presently available in the area of tomographic algorithms. These algorithms work with a set of projections of the field being investigated and reconstruct it to a certain degree of approximation [28]. They can be broadly classified as: (a) Transform methods, (b) Series expansion methods, and (c) optimization techniques. The first leads to the explicit calculation of the reconstructed field via the Radon Transform. Practical implementation of this method involves the use of convolution integrals and is called the convolution backprojection (CBP) algorithm. The second and third are iterative in nature and have been developed with a view towards handling a limited number of projections. In interferometry applied to measurement of temperature fields in fluids, the series expansion method is best suited. Censor [29] has reviewed the series expansion methods in terms of their rate of convergence and accuracy. Gull and Newton [30] have discussed the use of maximum entropy principle in tomographic reconstruction. A method of encoding prior information has been discussed. Verhoeven [31] has reported the performance of state-of-the-art implementation of a MART (Multiplicative Algebric Reconstruction Technique) algorithm to multidimensional interferometric data. Subbarao et al. [32] have compared ART (Algerbric Reconstruction Technique), MART, and  entropy-based optimization algorithms for three-dimensional reconstruction of temperature fields. A derailed error analysis has been presented: The principal findings of their study is that MART gives the best all around performance even with as few projections as two.