一种自适应毫米波被动成像的超分辨率算法  

An Adaptive Super-Resolution Algorithm for Passive Millimeter-Wave Imaging

作　　者：陈友仙[1] 聂建英[1] 

机构地区：[1]福州大学数学与计算机科学学院,福建福州350108

出　　处：《激光与光电子学进展》2012年第11期90-96,共7页Laser & Optoelectronics Progress

基　　金：江苏省资源环境信息工程重点实验室基金(JS201104);国防重点预研项目(51305050204)资助课题

摘　　要：针对目前毫米波被动成像的图像分辨率很低的问题,提出了一种新的有效的超分辨率算法。通过改进维纳滤波方法,对毫米波图像进行低频分量恢复,将得到的低频分量代替通带内频谱分量,充分利用了图像中的信息,使之具有自适应功能。将改进的维纳滤波的优点和正则最大实验概率(MAP)超分辨率算法外推高频分量的优点相结合,通过正则MAP算法迭代获得外推高频分量,对图像的傅里叶变换作频域校正,再求图像的逆变换,对其作进一步校正,逐次进行上述过程,直到达到提高图像的分辨率的目的。对提出新的自适应超分辨率算法进行了多次实验验证。结果表明,提出的新算法能够增强图像的分辨率,收敛速度快,峰值信噪比高,并且视觉效果优越于维纳滤波和正则MAP算法。实验结果证明了所提出方法的有效性。To solve the problem that the resolution of the millimeter-wave passive imaging is low, a new effective super-resolution algorithm is proposed. Wiener filtering algorithm is improved, and the low-frequency components are restored. The low-frequency components of instead of the pass-band with spectrum component is put to making full use of the information in the image and make it self-adaptive. Then the adaptive Wiener filtering and regular maximum a posteriori （MAP） super-resolution algorithm which can extrapolate high-frequency component are combined. The regular MAP iterative algorithm is used to obtain extrapolated high-frequency components, the frequency domain correction is conducted for Fourier transform of the image, and inverse transform of the image is carried out for further correction, the above process is performed successesively until the resolution of the image is improved. The new adaptive super-resolution algorithm is tested through the simulation experiments. The simulation results show that the new algorithm can enhance the resolution, improve the peak value signal-to-noise ratio and has fast convergence. The visual effect is superior to the other two algorithms; which indicate that the new algorithm is effective.

关 键 词：图像处理 超分辨率算法 自适应算法 维纳滤波 无源毫米波成像 

分 类 号：TN015[电子电信—物理电子学]