PEMANFAATAN ALGORITMA MEAN FILTER DAN MEDIAN FILTER UNTUK PENINGKATAN KUALITAS CITRA

Abstract

Image filtering is one part of improving image quality, namely smoothing and eliminating noise in the image, both linearly and non-linearly. Mean filter is a type of linear filtering that works by replacing the intensity of a pixel value with the average of that pixel value with the values ​​of neighboring pixels.  Meanwhile, the median filter is a type of non-linear filtering that sorts the intensity values ​​of a group of pixels, then replaces the processed pixel values ​​with the median value. Image quality is measured by two quantities, namely MSE (Mean Square Error) and PSNR (Peak Signal to Noise Error). In this study, the largest increase in PSNR in images with Gaussian noise occurred in the mean filter method with an increase of 7.182 dB in the 3x3 and 5x5 masks. Meanwhile, for images with salt noise, pepper noise, and salt & pepper noise, there was an increase in PSNR of 13,704 dB, 5,888 dB, and 11,717 dB respectively on a 3x3 mask for images with salt noise and on a 5x5 mask for images with pepper noise and salt & pepper noise. With a greater increase in PSNR than the mean filter method, it can be concluded that the median filter method is the best method for reducing noise in the image.