摘要
从高速流场扰动的失真图像中恢复出满足制导系统所需精度的目标图像是需要研究的课题。Holmes的基于泊松模型的极大似然校正算法存在平凡解的问题,针对此问题,提出在原算法的极大似然概率估计函数中增加适当的修正项来约束迭代校正算法向正确的方向收敛。讨论了修正项中参数的优化问题,提出按照使校正后图像的熵最小的优化准则来选取合适的修正参数。最后使用了实际风洞模拟实验中获得的气动光学效应降晰图像来测试改进算法的校正效果,同时给出原算法的校正结果以便比较。测试结果表明,改进算法有效消除了原算法的平凡解问题,校正效果较原算法有较为明显的提高。
It＇s necesary to study that target＇s image meeting the precision needed by guidance system from distortion image of high-speed disturbance flow field. The Maximum-Likelihood Poisson-based image restoration algorithm proposed by Holmes which has a trivial solution, and a proper revising term of Maximum-Likelihood Poisson-based probability estimation function was added to avoid this trivial solution and made the iterative restoration algorithm converge toward to the right direction. How to optimize those parameters in the revising term was also discussed, and an optimizing criterion which minimized the entropy of the restored image was proposed to select suitable parameters. Finally the peformance of improved algorithm was tested by blur images degraded by aero-optics effect in real wind hole simulation experiment, and that of original algorithm was also given. The results show that improved algorithm succeeds in deleting trivial solution and gives evidently better restoration results than the original one.
出处
《红外与激光工程》
EI
CSCD
北大核心
2008年第3期547-550,共4页
Infrared and Laser Engineering
