基于多目标优化的图象重建算法

An Image Reconstruction Algorithm Based on Multiobjective Optimiza tion

针对当前流行的最大似然估计算法所存在的,由于投影数据的不完备性所产生的重建结果的非唯一性和最大似然估计算法本身对统计误差的积累放大效应所产生的,当迭代次数超过某个极限时,使重建图象质量开始严重变坏的问题,该文提出了新的基于多目标优化的图象重建算法,大大改善了图象重建的效果。

The reconstruction result from the current image reconstruction algorithm-maximum likelihood algorithm is nonunique because of the incomplete projection data.The quality of reconstructed image will grow worse seriously when iteration number is larger than certain limit because maximum likelihood algorithm has the accumulation enlargement effect to statistical error.Based on the faults of maximum likelihood algorithm,this paper developes a new reconstruction algorithm based on multiobjective optimization to improve the inherent faults of the current algorithms based on single criteria optimization.Experiments show this algorithm can improve the quality of reconstructed image largely.