This paper proposed a swift adaptive algorithm to restore images from impulse noise

by expanding the real signal pixels so as to smooth the noisy ones. First

the algorithm uses a critical region to distinguish signal pixels from noise and set a marked matrix. Then calculate the density of signal

which will be used to control the extent of growth adaptively. The final restored image will be obtained nearly after a simple correction on the expanded image matrix. The proposed algorithm exceeds the current ones markedly in both denoising effect and computational complexity.