Impulse Noise Removal with Modified BDND and Adaptive Switching Median Filter

In this paper, a impulse noise removal approach is proposed where a modified boundary discrimative noise detection (MBDND) and an adaptive switching median filter (ASMF) are applied. Note that the BDND has difficulty in cases with high noise densities. A boundary resetting scheme is introduced in the BDND. By this doing, the problem of miss detection in the high noise density is prevented. Note that the restored image obtained from the modified noise adaptive soft-switching median filter (MNASM) is of strong smoothing effect because of larger windows are employed. Thus the proposed ASMF uses a criterion to expand the window that results smaller windows are used in the filtering process. Consequently, the retored image from the proposed ASMF is expected to have better visual quality over that from the MNASM. Two examples are provided to justified the proposed noise removal approach MBDND/ASMF where comparisons are made with the BDND/MNASM. The results indicate that the MBDND/ASMF outperforms the BDND/MNASM both in the PSNR and visual quality of the restored image. Key-Words: noise removal, noise detection,switching median filter, BDND, impulse noise