欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
为解决超分辨复原中遥感图像调制传递函数过零点引起的病态问题，提出了一种基于最大后验概率估计，利用遥感系统成像模型对图像进行频域校正的遥感图像超分辨率复原算法．该算法首先假设图像满足高斯马尔科夫先验模型，使用最大后验概率估计法实现图像频谱外推．然后根据成像模型对外推的频率分量进行频域校正，去除调制传递函数过零点附近的伪信息．最后使用优化最小化方法完成数值求解，降低了计算复杂度．实验结果证明，该算法考虑了图像成像模型本身特性，对合成图像和各种地物条件下的遥感图像都能取得快速有效的超分辨复原效果．
英文摘要：
To solve the ill-posed problem imported by zero-crossing of modulation transfer function （MTF） in the super-resolution restoration, a novel MRF-MAP （maximum a posteriori） based super- resolution restoration methods with acquisition system modeling based frequency domain correction is proposed. The Huber function is used in a Gaussian MAP solutions and completes spectral extrapolation. Markov random field（GMRF） to operate stable Then a frequency domain correction algorithm is adopted for the geometrical characteristics of the remote sensing system to revise the frequency spectrum above the cut-off frequency, A majorization-minimization approach is used to reduce the computational complexity. Experimental results indicate that by considering the modeling of image acquisition systems the proposed method can achieve high-quality super-resolution image for both composite image and remote sensing images with different physiognomy.
同期刊论文项目
同项目期刊论文
期刊信息
位置：