A Fast Hybrid Noise Filtering Algorithm Based on Median-Mean

Xun Zhang,Hongzhou Liao,Xue Du,Bo Xu

DOI: https://doi.org/10.1109/icma.2018.8484392

2018-01-01

Abstract:Median filtering and mean filtering are two of the most common and practical methods in digital image processing, which are used to deal with impulse noise and gaussian noise respectively. However, the optical images often contain two kinds of noise at the same time. When the noise pollution is serious, the interaction between the noises can lead to a single filtering method that can not effectively eliminate the noise, leading to the failure of filtering. In order to suppress these two kinds of noises simultaneously, a fast algorithm of median and mean joint filtering is proposed. First, set consistent decomposition method is proposed and used in each sliding window for fast median filtering to improve the real time, and then weighted mean filter is combined. In order to solve impulse noise effect on the final output when using mean filtering, the adaptive window is proposed to effectively identify noise points, reducing the impulse noise effects on the output. The algorithm effectively matches the median filtering and mean filtering, and improves the real-time performance of the filter, which is conducive to the real-time target tracking task. The experimental result shows that the proposed algorithm can suppress the serious mixed noise effectively and takes less filtering time, so it has certain application value.