【摘要】 由于APDCT和APIDCT分别在子带分解和图像内插中表现出优异的性能,因此提出了一种基于APDCT滤波和APIDCT内插的多级分解方法.用该方法代替原Contourlet第1步中的拉普拉斯变换,不仅在分级中应用了纯二维滤波器,而且分级重建时只需相加即可,大大简化了运算.实验表明,将提出的全相位Contourlet变换应用在普通图像去噪中,不仅信噪比有所提高,而且图像视觉效果明显优于原Contourlet.更多还原

【Abstract】 Because APDCT and APIDCT are excellent respectively in subband decomposition and image inter- polation,a novel mutliscale decomposition method,based on the APDCT filter and APIDCT interpolators to re- place Laplacian Pyramid at the fist step of Contoulet,has been developed,and a novel Contourlet transform bas been constructed.In this multiscale method,not only a true two-dimension filter is implemented,but the com- putation complexity is reduced by adopting only add operations during the reconstruction process.Experiments in denoising the typical image Lena showed that the performance of the proposed AP-Contourlet is obviously superi- or to the original Contourlet both in vision and in signal to noise ratio(SNR).更多还原