DocumentCode :
2369602
Title :
New image transforms using hybrid wavelets and directional filter banks: analysis and design
Author :
Eslami, Ramin ; Radha, Hayder
Author_Institution :
Dept. of Electr. & Comput. Eng., Michigan State Univ., East Lansing, MI, USA
Volume :
1
fYear :
2005
fDate :
11-14 Sept. 2005
Abstract :
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.
Keywords :
channel bank filters; image reconstruction; image resolution; wavelet transforms; PSNR; directional filter banks; hybrid wavelets; image reconstruction; image transforms; multiresolution geometrical image; nonlinear approximation property; Channel bank filters; Design engineering; Electronic mail; Filter bank; Frequency; Image analysis; Image coding; Image processing; Wavelet analysis; Wavelet transforms; directional filter banks; geometrical image transforms; wavelet transforms;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Image Processing, 2005. ICIP 2005. IEEE International Conference on
Print_ISBN :
0-7803-9134-9
Type :
conf
DOI :
10.1109/ICIP.2005.1529855
Filename :
1529855
Link To Document :
بازگشت