摘要
在分析CS(Chirp Scaling)成像算法的基础上,发现弯曲因子Cs和等效调频率因子Ks对多普勒模糊数DAN(Doppler Ambiguity Number)的变化非常敏感,不正确的DAN,将使Cs因子和Ks因子失效,最终导致SAR图像的清晰度下降和位置发生变化.据此提出估计星载SAR DAN的新算法:将DAN的变化与描述图像清晰度的评价函数(熵或均方差)联系起来,通过观察所得图像的熵或均方差随DAN值的变化,找到最小熵或最大均方差时所对应的DAN,即为正确的值.用Radarsat的实际数据验证了算法的有效性.分析表明,基于图像均方差的DAN估计优于图像熵的DAN估计,前者在成像区域场景反差较大和较小时均适用,而后者在成像区域场景反差较小时表现不稳定.
It is found that in the Chirp Scaling algorithm, the curvature factor C, and the effective FM chirp rate Ks are very sensitive to the change of Doppler-Ambiguity-Number （DAN）. When DAN is inaccurate, Cs and Ks will be invalid and the definition of resulted SAR image will degrade, i.e. the definition-evaluation-function of SAR image （entropy or mean variance） varies with DAN. Based on the analysis, a new method was proposed for DAN estimation in spaceborne SAR imaging by finding the minimum of image entropy or the maximum of image mean variance. The raw data of Radarsat were used to validate the method. The results indicate that the algorithm based on mean variance is better than the algorithm based on entropy. The former is applicable no matter the contrast of image area is high or low, but the latter is only suitable for high contrast area.
出处
《测试技术学报》
2008年第3期250-255,共6页
Journal of Test and Measurement Technology
