FAST SMALL-KERNEL K-NEAREST NEIGHBOR NOISE-SMOOTHING ALGORITHM

A basic operation in image processing is the removal of noise from a n oise-corrupted input picture. The K-nearest neighbour (KNN) is a very effective noise-smoothing filter: it removes both additive and multipl icative noise without blurring the edges and lines in the input pictur e. The main drawback to this filter is its high computational complexi ty. In this paper, the author describes a fast algorithm specifically designed for the small-kernel KNN filter. The computational complexity of the new algorithm is sufficiently low to facilitate real-time nois e-smoothing.