Removal of impulse noise in medical images with machine learning techniques

It is desired that an ideal filter preserves the details in the digital image while effectively removing impulse noise. The objective of standard signal processing is the elimination of impulse noise without harming details of the image. In part 1, we propose a new Median-Based Method: Switching Median Filter with a hybrid of adaptive median filter combination. Hereon, the proposed filter will be named as "Double Checked Fast Adaptive Median Filter" (DCFAMF). DCFAMF gives faster, simpler and better results in comparison with the traditional Median-Based Filters. This method can get rid of impulse noise. Moreover, it can keep the necessary details of the image (this is also true when the input image is very badly ruined by noise). No adjustments are needed for DCFAMF. Thus, it is much more suitable for automated systems. We do not need a pre-preparation for this technique. Besides, an adaptive artificial neural network model is developed in order to restore severely corrupted images. Networks trained at different noise intensities get activated according to intensity of the noise and estimate the most suitable neighboring pixel that can replace the noisy pixel. The proposed algorithm reduces impulse noise effectively while also protecting the details. Naïve Bayes classifier filter for the removal of random impulse noise in digital grayscale image is recommended. It has especially been used more frequently in recent times in the field of signal processing. Prior to restoring the noisy pixels of the image as is used here, the image is separated into more than one piece, and a learning set is formed using the noise free pixels in these pieces. These learning sets that are different for each piece are used in order to estimate the pixel that will replace the noisy one. The presented method is both simple and easy to apply.