投影小波域MAP估计无源毫米波图像超分辨算法  

Projected Wavelet-Domain MAP Estimation Super-resolution Algorithm for Passive Millimeter Wave Imaging

作　　者：金鑫[1] 熊金涛[1] 李良超[1] 杨建宇[1] 

机构地区：[1]电子科技大学电子工程学院,成都610054

出　　处：《电子与信息学报》2010年第4期889-893,共5页Journal of Electronics & Information Technology

基　　金：国家自然科学基金(60776823);电子科技大学校青年基金(JX0823)资助课题

摘　　要：在无源毫米波成像中,由于系统天线孔径大小的受限而使得成像的分辨率低。为了提高图像的分辨率,该文提出了一种投影小波域最大后验(MAP)估计毫米波图像超分辨算法(PWMAP)。该算法利用基于小波域广义高斯分布的MAP估计来恢复通带内的频谱;然后利用投影的非线性运算实现频谱外推。该算法不仅比以往的算法能提供更准确的先验建模,而且能在每步迭代时自适应地更新正则参数。实验结果验证了该算法的有效性。In passive millimeter wave imaging,the problem of poor resolution of acquired image stems mainly from system antenna size limitations. In order to achieve resolution improvements,a Projected Wavelet-domain Maximum A Posteriori （PWMAP） estimation super-resolution algorithm is proposed in this paper. This algorithm restores the spectrum in the pass-band based on wavelet domain using the generalized Gaussian distribution and the MAP estimate; then extrapolate the spectrum by using the non-linear projection operation. This algorithm can not only provide a more accurate priori model than previous algorithms,but also updates the parameter adaptively at each iteration. Experimental results show the effectiveness and superiority of the algorithm.

关 键 词：无源毫米波成像 超分辨 小波域 自适应 非线性运算 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]