Reconstructing non-ideal-bordered field by simple self-correlative algebraic reconstruction technique

Abstract:

A new algorithm for iterative reconstruction is suggested. It is named simple self-correlative algebraic reconstruction technique (SSART). With numerical simulation experiments, SSART was applied in reconstructing a non-ideal-bordered field in order to test its reconstructive effect. For contrast, three current representative algebraic reconstruction(ARTs) including basic ART, simultaneous ART (SART) and modified SART (MSART) were simulated simultaneously. The calculated results and reconstructive accuracy are discussed with three kinds of error indexes, namely mean-square error (MSE), absolute value error (AVE) and peak error (PE). As the results, the indexes of reconstructive accuracy are much improved by the proposed SSART. The MSE and PE have been decreased by 63.6% on the order of magnitude 10−4 and 88.9% on the order of magnitude 0–2, respectively, compared to that of ART. It is concluded that SSART is one of the most super iterative reconstructing techniques. This project was supported by China Postdoctoral Science Foundation (2004036121), the National Natural Science Foundation of China (10404022), and Education Bureau of Shandong province (J04A63)

Key words:;KeywordHeading";>CLC number TP391