摘要
针对遥感图像薄云覆盖下地物细节模糊问题,给出一种融合对偶树复小波变换和支持向量滤波的遥感图像薄云覆盖下地物信息恢复算法.利用对偶树复小波变换和支持向量滤波器将薄云覆盖遥感图像分解为高频方向子带和低频子带;分别对高频方向子带系数进行自适应增强,对低频子带系数加以抑制;对不同方法处理后的低频系数采用基于匹配度的选择和加权相结合方法进行融合,对高频系数采用基于轮廓波对比度的方法进行融合,获得地物细节清晰的融合图像.实验结果表明,算法在视觉效果和定量指标上优于对偶树复小波变换方法和支持向量滤波方法.
Aiming at the problem of the fuzzy details of the ground objects in the remote sensing image withthin cloud cover,a ground object information recovery algorithm for thin cloud covered remote sensing imag-es is proposed by fusion of dual tree complex wavelet transform and support vector filtering. Firstly,based onthe dual tree complex wavelet transform and the support vector filter respectively,thin cloud covered remotesensing images are decomposed into high frequency directional subbands and low frequency subbands. Thenthe high-frequency directional subband coefficients are enhanced adaptively and the low frequency subbandcoefficients are suppressed. Finally,the low frequency coefficients are fused by using the method of combin-ing selection and weighting based on the matched degree and the high frequency coefficients are fused basedon the contourlet contrast to get the fused images with clear ground object details. The experimental resultsshow that the proposed algorithm is superior to the methods of dual tree complex wavelet transform and sup-port vector filtering in visual effects and quantitative indexes.
出处
《淮北师范大学学报（自然科学版）》
CAS
2017年第3期53-59,共7页
Journal of Huaibei Normal University：Natural Sciences
基金
安徽省自然科学基金项目(1408085MF121)
安徽大学学生科研训练计划项目(KYXL2016064)
偏振光成像探测技术安徽省重点实验室开放课题(2016-KFKT-003)
