摘要
提出一种新的技术 ,它自适应地选取正则算子以取得较理想的恢复效果 通过理论分析和实验发现当恢复图像残差的频谱能量分布较均匀时恢复效果较好 这种分布均匀性可以用正则图像残差的各子频段能量偏离平均能量的程度最小来衡量 ,这个最小化问题以各种各样的正则算子组成的空间为搜索空间 由于一般的优化算法对此优化问题无能为力 ,演化算法用来求解此问题 ,从而自适应地选择正则化算子
A new technique is proposed to choose regularization operator adaptively in order to get good image restoration Theoretical analysis and experiment indicate that the restoration is good when the residue energy distribution of restored image is uniform The uniformity is measured by minimizing the dispersion of every subband of the residue from the average energy which is a function of regularization operator Evolutionary algorithm is employed to solve the minimization problem, so as to choose regularization operator adaptively, while other optimal algorithms are helpless to do that Experiment results show that the regularization operator selected by using the new technique is good for image restoration
出处
《计算机研究与发展》
EI
CSCD
北大核心
2003年第4期584-588,共5页
Journal of Computer Research and Development
基金
国家自然科学基金 (60 0 73 0 43
70 0 710 42
60 13 3 0 10
60 2 0 40 0 1)
武汉青年科技晨光计划基金 (2 0 0 2 5 0 0 10 0 2 )
