Adaptive Filter Design for High Percentage Noise Removal
R. Udayakumar · 2014
3 Abstract: The paper is aimed at the design of adaptive two-pass rank order filter to remove impulse noise in highly corrupted images. The rank order filter used is a median filter. Two-pass filtering is the application of this filter twice, which will produce better results when the signal to noise ratio is extremely low. An adaptive process will be included which will detect the irregularities in the spatial distribution of the estimated impulse noise. In this process, the median filtering is adaptively varied to meet the variation in the distribution of the impulse noise. Replaced pixels are kept unchanged during the second pass of filtering. The second filter will eliminate more noise and improve the resolution of the image. As a final result, the reconstructed image has a higher degree of fidelity and smaller amount of noise. Results from simulations are used to demonstrate the performance of this type of adaptation for various degrees of additive noise.