摘要
为提高前视红外图像复原的计算速度和复原效果,首先提出一种加速的Landweber迭代复原算法,使松弛迭代因子在迭代过程中,根据复原图像的梯度特征进行自适应选择,使收敛所需迭代次数降低为传统landweber算法的1/5~1/6,提高了算法的实时性。同时,设计出改进的Neumann边界条件,对观测图像进行延拓,将图像复原误差转移到图像主值序列之外,有效降低复原图像边界处的振铃效应。
Aiming at the problems of improving the performance and computational efficiency of the Forward Looking Infrared（FLIR） image restoration procedure,an improved landweber iterative accelerating technique is proposed first.Instead of being fixed,the relaxation factor is updated adaptively at each iteration considering the gradient character of the restoring image.Experimental results demonstrate that the required iterative times of the proposed method are reduced to 1/5~1/6 of the traditional landweber iterating method,which can improve the real-time processing.Meanwhile,techniques are developed for suppressing the ringing effects in the restored images.Improved Neumann boundary condition is introduced to pad the observed image,which shifts the restoration errors out of the image principal values.Results obtained by the proposed restoration technique leads to a larger reduction of the ringing effects than classical schemes.
出处
《武汉理工大学学报》
CAS
CSCD
北大核心
2010年第15期150-155,共6页
Journal of Wuhan University of Technology
基金
国家自然科学基金(60672057)
