毫米波辐射图像的运动模糊参数辨识-红外技术.pdfVIP

第32 卷 第5 期 红 外 技 术 Vol.32 No.5 2010 年5 月 Infrared Technology May 2010 〈图像处理与仿真〉 毫米波辐射图像的运动模糊参数辨识 胡泰洋，肖泽龙，许建中 （南京理工大学 电子工程与光电技术学院，江苏 南京 210094 ） 摘要：针对平面扫描毫米波辐射成像系统和被成像场景之间的运动模糊问题，提出了一种基于Radon 变换和微分自相关运算的运动模糊参数辨识方法。首先对原始运动模糊图像两次傅立叶同态变换谱的 二值化图像进行Radon 变换以估计运动模糊方向，然后在此方向上利用运动模糊图像的微分自相关运 算得到运动模糊长度，最后以运动模糊方向和运动模糊长度为参数结合维纳滤波算法对运动模糊图像 进行恢复，并引入基于图像灰度梯度向量模方和的运动模糊图像清晰度评价函数，实现对原始运动模 糊图像及恢复图像质量的评估，从而达到衡量本文算法辨识精度的目的。实验结果表明，该算法能够 准确辨识毫米波辐射图像的运动模糊参数，并具有良好的抗噪声干扰性能及稳定性。 关键词：毫米波辐射成像；平面扫描；运动模糊；参数辨识 中图分类号：TP751，TN911.73 文献标志码：A 文章编号：1001-8891(2010)05-0273-06 Motion-blur Parameters Identification of Millimeter-wave Radiometric Images HU Tai-yang ，XIAO Ze-long，XU Jian-zhong (School of Electronic Engineering and Optoelectronic Technology, Nanjing University of Science and Technology, Nanjing 210094, China) Abstract ：According to the problem of motion-blur imposed by the relative motion between planar scanning millimeter-wave radiometric imaging system and imaging scene, a novel motion-blur parameters identification method which was based on Radon transform and differential autocorrelation was proposed in this paper. Firstly, the motion-blur direction was estimated by means of the Radon transform to the binary image of the spectrum of original motion-blur images after twice Fourier homomorphic transform. Then the motion-blur length was calculated through the differential autocorrelation calculation to original motion-blur images in the estimated direction. Finally the original images were restored by Wiener filter algorithm which utilized the motion-blur direction and motion-blur length as parameters, and