Adaptive Salt-&-Pepper Noise Removal: A Function Level Evolution based Approach

Noise removal is an important procedure before ongoing further image processing. In this paper, a two-stage adaptive noise removal scheme is proposed. The first stage of the scheme is noise detection; the second stage is functional level evolution based noise cancellation. This cancellation only operates on the noise candidates detected in the first stage. Additionally, the proposed method acquires local information to suppress noise from the noisy pixelspsila neighbors rather than from original images. These features enable the scheme to preserve more image details with less computation effort, and make it able to be the basic frame for EHW based implementation of on-line adaptive impulse noise removal system. Simulation results show that the proposed approach provides at least equal performance to the adaptive median filter, and can cope with the corrupted images with up to 70% noise density.