Removal of Random Valued Impulse Noise using Enhanced Dual Threshold Based Median Filter with Edge Preservation

In case of random valued impulse noise (RVIN) removal, the main issue is to identify the presence of impulse noise in the filtering window because the differences between pixel under observation and its neighbors within the filtering window may be very small. For this case two schemes are suggested. In the first scheme, the relative difference between the pixel and other pixels within the filtering window is increased by using a nonlinear function to facilitate impulse detection. In the second scheme, an adaptive threshold, which is based on the local statistics within the filtering window, is used for detection of impulse noise. In both the schemes, the detection is followed by a variant of weighted median filter and the schemes are iterative in nature. A suitable stopping criterion is also suggested which is based on the characteristics of the image to be filtered. For the removal of impulse noise, a new approach that is Enhanced Dual Threshold based Median Filter (EDTMF) for impulse noise removal for both low and high level impulse noise density levels. In this work, proposed filter is used for digital image random valued impulse noise reduction. In the impulse noise removal there are two main stages, firstly, the detection of the impulse noises on the basis of maximum and minimum value of pixels in a small window. In the second stage, removal of noise on the basics median calculation. In the filtering stage, the noise-free pixels remain unchanged in black and white, windows and noisy pixels are restored using a median filter.