基于子带内小波系数相关性的遥感图像滤波

RS imagery filtering based on the relativity among wavelet coefficients in the sub-bands

滤波是遥感图像处理和应用过程中必不可少的一个环节。传统的小波域滤波方法对所有的细节小波系数逐一进行独立的收缩处理,而没有顾及小波系数之间的相关性,因而不具备较强的鲁棒性和适应性。文中提出一种基于小波系数聚集性分布的滤波策略。该方法认为小波系数的分布具有集中性,对当前的系数在邻域内计算其能量,而后将其纳入系数的收缩策略,并通过调节相应参数改进滤波策略。实验结果令人满意,证明该方法进行图像滤波的有效性。

Filtering is one of the essential courses for a remote sensing imagery in its processing and application.Traditional wavelet based methods for image filtering shrinks all the detail coefficients independently and ignores the relativity among wavelet coefficients,thus the adaptability and robust of these methods are poor.The paper proposed a new filtering strategy based on the concentrating distribution of wavelet coefficients.The proposed method believes there exists relativity among wavelet coefficients in the sub-band;it calculates the energy of the current detail wavelet coefficient in a neighborhood and then brings it into the shrinkage scheme and modifies the scheme by adjusting the parameters.The final experiments show the feasibility of the new strategy.