The video signal is easily interfered by noise over the wireless system. The well-known "median filter" and its modified methods could completely remove the noise when the noise/image ratio is low. However, as this ratio becomes more and more high, the filtering performance degrades very fast and most of noise can not be removed. To resolve high-noise problem, we proposed a new adaptive algorithm to efficiently reduce noise without visible distortion. The experimental results show that the proposed methods can attain very high quality in 50% noise environment. Moreover, this adaptive approach also keeps lower complexity in order to meet real-time applications.