In this paper, we propose an adaptive decision-based median filtering algorithm, called adaptive pixel-correlation filter (APCF), with an adaptive threshold for removal of impulse noise. By exploiting the correlations between a pixel and its neighbors, an adaptive threshold can be designed based on the global statistical properties. Due to the fact that both horizontal and vertical correlations for a pixel more significant than that of other orientation, such a characteristic is used to design a filter with a novel adaptive working window and weighted function for detecting impulse noise and preserving noise-free pixels. Numerical simulations show that APCF is more robust and effective than many other well-known median filtering algorithms.
展开▼