New image transforms using hybrid wavelets and directional filter banks: analysis and design
New image transforms using hybrid wavelets and directional filter banks: analysis and design
复制标题
DOI：
10.1109/icip.2005.1529855
复制
发表时间：
2005-11
期刊：
影响因子：
--
通讯作者：
R. Eslami;H. Radha
中科院分区：
文献类型：
--
作者：
R. Eslami;H. Radha
我们提出了一个新的家庭完美的重建，非冗余，和多分辨率几何图像变换结合使用的小波变换与修改版本的方向滤波器组（DFB）。在所提出的DFB版本中，我们使用水平或垂直方向分解。利用小波变换具有高效的非线性逼近特性，我们通过将修改的规则DFB应用于几个最细小波级别的子带来添加方向性的重要特征。通过这种方式，我们可以消除使用DFB时通常引入的大部分伪影。建议的混合小波和DFB（HWD）变换家族提供了小波和轮廓波变换的视觉和PSNR的改善。
We propose a new family of perfect reconstruction, non-redundant, and multiresolution geometrical image transforms using the wavelet transform in conjunction with modified versions of directional filter banks (DFB). In the proposed versions of DFB, we use either horizontal or vertical directional decomposition. Taking advantage of the wavelet transform that has efficient nonlinear approximation property, we add the important feature of directionality by applying the modified and regular DFB to the subbands of a few finest wavelet levels. This way we can eliminate a major portion of the artifacts usually introduced when DFB are used. The proposed hybrid wavelets and DFB (HWD) transform family provides visual and PSNR improvements over the wavelet and contourlet transforms.