FPGA based digital image restoration using multiresolution transform based filtering

Multiresolution processing of an image has become increasingly important in the present application areas like image restoration, filtering, enhancement, compression, etc. Although the Fourier transform has been the mainstay of transform-based image processing, a more recent transformation, called the wavelet transform, now makes it easier to compress, transmit, and analyze many images. Another multiscale approach which is an alternative to wavelet transform, is the multiresolution median transform. It is a non-linear transform and offers advantages for robust smoothing. This paper presents the VLSI implementation of multiresolution transform based filtering impulse noise from images. This transform has many advantages in comparison with wavelet transform, such as fast computation, error-free reconstruction, etc. synthesis is performed with Xilinx spartan-II FPGA which yields users high performance, unlimited reprogrammability, very low cost and provides system clocks upto 200 MHz.