The block compressed sensing approaches based on matrix permutations reduce storage space costs, reduce transmission costs, and support the high quality reconstruction of images by reducing the blocking artifact. An enhancement in the literature includes the block compressed sensing method based on ripple matrix permutations, which further balances the textures of sub-blocks. Based on the block compressed sensing method by ripple matrix permutations, this paper proposes a novel method for high-frequency images that uses the energy distribution characteristics of high-low frequency images. The proposed method makes full use of this characteristic and performs matrix permutations only on the high-frequency image. This method initially performs a wavelet decomposition on the image. Subsequently, the transformed high-frequency image is subjected to a ripple matrix permutation to achieve texture balancing. Finally, the compressed sensing processing is performed on the high-frequency image. The simulation results show that the high-frequency part of the image wavelet domain is texture balanced while the low-frequency part remains unchanged. The image is reconstructed after the compressed sensing step, and the image quality is significantly improved.
Image Texture Balancing in the Wavelet Domain Based on a Ripple Matrix Permutation
Lect. Notes Electrical Eng.
International Workshop of Advanced Manufacturing and Automation ; 2020 ; Zhanjiang, China October 12, 2020 - October 13, 2020
23.01.2021
8 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Wavelet-Domain Texture Modeling for Image Compression
British Library Conference Proceedings | 1994
|Image Texture Classification Using Exponential Curve Fitting of Wavelet Domain Singular Values
British Library Conference Proceedings | 2006
|Boundary refinements for wavelet-domain multiscale texture segmentation
British Library Online Contents | 2005
|Texture Classification Based on the Wavelet Decomposed Image
British Library Conference Proceedings | 2006
|