Due to the recent development of the display technology, digital image processing has been used in many fields. In general, digital image processing data represent a visual error while entailing heat deterioration due to external causes in the process of acquiring, processing, and analyzing them. Therefore, in this paper, various studies have been conducted to remove noise added in the image.
There are many types of noises added in the image depending on the causes and forms. As for representative examples, there are salt & pepper noise and AWGN(adaptive white Gaussian noise). There are linear and non-linear filters designed to remove such noise. As for linear filter, there are AF(average filter) and WF(wiener filter). As for non-linear filter, there are MF(median filter), HMF(hybrid median filter), and MMF(modified median filter). As a representative linear filter, AF calculates the average of neighboring pixels in the local mask and processes them. Therefore, it features in outstandingly removing noises in AWGN and low frequency. MF, a non-linear filter, arranges the scale of neighboring noises in the near the mask in a descending order and processes them with a median. Therefore, it features in effectively removing salt and pepper noise. However, previous filters have been insufficient in removing noise in the environment with highly dense salt & pepper noises.
Therefore, this study has suggested a deformed switching filter that maintained the central pixel in the local mask not as a noise but as an original pixel through judging whether it was noise and arranged and processed the noise according to the density of noise in the local mask. In case of low density of noise in the local mask, dual linear interpolation has been applied and processed. In case of med-density, spatial weight filter has been applied. In case of high-density, medial filter has been utilized for processing the noise. In addition, if center pixels in the local mask were all noises, algorithm that calculated the average value of processed pixels has been suggested. In addition, PSNR(peak signal to noise ratio) has been used to evaluate the function of algorithm in removing noises and compare them with other existing methods.