3D DCT Based Filtering of Color and Multichannel Images
Nikolay N. Ponomarenko, Владимир Васильевич Лукин, Alexander A. Zelensky, Pertti T. Koivisto, Karen Egiazarian · Telecommunications and Radio Engineering · 2008
The performance of vector (3D) filtering of color and multichannel images in discrete cosine transform (DCT) domain is studied. It is shown that the vector filtering offers some advantages compared to the component-wise (2D) filtering. Moreover, it is shown through experiments that the DCT based processing can be more efficient than advanced nonlinear vector filters. As noise models, both additive noise and mixed additive and impulsive noise are used. Finally, the applicability of the DCT based filtering is studied when the noise is multiplicative; and a convenient way to reduce multiplicative noise to additive noise is proposed.