Linear surface interpolation for salt-and-pepper noise removal

In this paper, we present a novel algorithm for the recovery of images, corrupted by the salt-and-pepper noises. Specifically, we adopt a linear interpolation scheme to reconstruct the identified noisy pixels. Our extensive experimental results show that the resulting performance of our proposed method is higher than many of the traditional methods of noise removal. More explicitly, in the case of low density noises, the reconstructed images are very close to the original images, resulting in very high PSNR values for the recovery. The computational simplicity of our algorithm has made it a very good alternative for other high complexity de-noising methods.