Tolerance Based Adaptive Switching Mean Filter for Impulse Noise Removal

In digital Image Processing, removal of noise plays a vital role and it is a highly demanded area of research. Impulsive noise is most common in images and this is mainly occurs at the time of image acquisition and or transmission of images. The impulse noise removal technique came into picture because it acts as most powerful and perspective technique to remove the noise in the image and the necessity of image noise reduction continuously increases during the last decade. Impulsive noise can be divided into two categories, namely Salt & Pepper Noise (SPN) and Random Valued Impulsive Noise (RVIN). Corruption of image during image acquisition, image transmission and image storage, is said to be Impulsive noise. The ‘n’ number techniques proposed in order to remove this type of noise. It is keenly notified that techniques which follow the two stage process of noise detection and filtering of noisy pixels outperforms better than others. In this thesis such schemes of impulsive noise detection and filtering techniques are discussed.
The impulse noise is removed from corrupted image by means of adaptive switching mean filter. Presently To restore the image highly corrupted by impulse noise, the two-stage morphological noise detector is used presently to detect the accurate noise in the image and this noise detection technique does not provide the expected results. To overcome this problem a tolerance based noise detection technique is proposed in this thesis. This noise detection technique is combined with the adaptive trimmed mean filter with the ability of adaptively adjusting the filtering window size. Tolerance Based Adaptive switching mean (TBASM) filter out shows its performance comparably better than other switching filters by the effective combination of the novel noise detector with the distinctive mean filter. The final output of the simulation results deliberately reveals that the proposed filter can realize or identify the accurate noise detection and it can suppress impulse noise effectively while preserving details in the image very well, thus providing significantly better restoration performance than the traditional median filter and numerous well known switching-based filters.
Extensive simulation results and comparisons are done with competent schemes. It is notifying that, in general, that the proposed schemes are better in suppressing impulsive noise at different noise ratios than their counterparts.