Study And Implementation Of Algorithm Based On Mean Filtering For Mixed Noises

The contents of this paper are the part of digital image processing technology in image enhancement technology. The study in this paper was aimed at smoothing filtering images influenced by the mixed noise, which include Gaussian noise and salt & pepper noise. When the digital image is interfered with by both of these noise, the application of classic filtering algorithm is often not reach good results. The filter for mixed noise filtering algorithm is particularly necessary.The paper starts with the basic concepts of digital images. First of all, it discuss Gaussian noise and salt-pepper noise characteristics under the digital image. Then, in-depth study of the classical filtering algorithms, including the mean and median filtering algorithms, is made, which explains their basic ideas and illustrates the filtering algorithm specific course of action. Under the foundation above, the advantage and disadvantage of the two kinds of filtering algorithms were importantly demonstrated. For Gaussian noise, average filtering algorithm has a better effect, while the median filter seems ineffective in filtered salt & pepper noise.The median filter can maintain image detail very well, while the mean filtering algorithm is not that excellent .Based on the merits of the both of filtering algorithm, a new noise filtering adaptive algorithm is proposed for mixed noises. Gray values sequence are obtained from the neighborhood pixel through three different filtering windows. After abandoning the maximum and the minimum values, salt and pepper noise are filtered out; the sequence,which possesses the minimum variance,will be selected as the optimal sequence, whose, in the application of neighborhood averaging-method, average value is the final output value. Finally, under test simulation, it proves that the filtering algorithm of this article better than the traditional algorithms for mixed noises filtering.