摘要
利用遥感信息技术准确掌握沿海滩涂湿地的现状,对于江苏沿海大开发这一国家战略的顺利实施具有重要意义。由于多极化合成孔径雷达图像(SAR)具有不同的极化信息,将极化SAR图像进行融合可以更有效地获取沿海滩涂湿地信息。针对多极化SAR图像的融合问题,提出了一种基于非亚采样Contourlet变换(NSCT)和脉冲耦合神经网络(PCNN)的图像融合方法,该方法采用简化的PCNN模型分别对图像的低频子带和高频子带系数进行智能决策。以江苏盐城地区的ALOS PALSAR双极化图像为例,对所提方法的有效性进行验证,从主观定性和客观定量两方面综合比较了新方法与几种传统的基于多尺度分解方法的融合效果。结果表明,新方法能够最大程度地保留原始极化SAR图像的信息,融合效果好于其他方法,更有利于沿海滩涂湿地信息的提取。
Grasping the present situation of coastal tidal flats by remote sensing information technology is important for Jiangsu coastal development policy. It is more effective to obtain the tidal flat information by fusing multi - polarization Synthetic Aperture Radar（SAR） images which provide different polarization information of the targets. In consideration with the problems of the fu- sion of multi - polarizaiton SAR, a new fusion algorithm based on Nonsubsampled Contourlet Transform （NSCT） and PCNN is proposed to fuse the multi - polarization SAR images. The simplified PCNN model is used to make intelligent decisions for the co- efficients of low and high frequency in sub - band respectively. Finally, the method is examined by using ALOS dual - polariza- tion SAR images of tidal flats in Yancheng City of Jiangsu Province and compared with some regular fusion algorithms based on multi -scale decomposition. The results indicate that the proposed method can reserve the original polarization information at the largest degree and its fusion effects are better, which can be more helpful for extracting the information of the coastal tidal flats in Jiangsu Province.
出处
《人民长江》
北大核心
2013年第5期52-56,60,共6页
Yangtze River
基金
国家自然科学基金项目(41274017)
江苏省科技支撑计划(BE2010316)
日本宇航局AlOS数据研究项目(PI534)
