OPTIMALISASI KUALITAS CITRA DIGITAL DENGAN METODE KETETANGGAAN PIKSEL

Ketetanggaan Piksel

Abstract

This study aims to improve the image or image by eliminating noise in the color image that is converted to gray images with a number of pixel intensities of 0 to 255 using the pixel neighboring method. Pixel neighbor types used in the test include boundary filters, uniform filters and median filters. While noise added to the original image is salt noise and pepper and gaussian with a value of p = 0.01. From the results of this study, it can be seen how much interference can be corrected by using a type of neighboring pixel filter, with the amount of improvement including reducing noise, softening noise and eliminating noise. From the results of the tests carried out with the help of the octave 3.4.3 program, it can be seen that the boundary filter aims to prevent pixels whose intensity outside the intensity of neighboring pixels with white spots on the image can be removed. However, if you pay close attention, the operation also blurs the image, while in the equalization filter the processing effect with the alignment filter. Compared to the limit filter, the processing of the uniform filter does not eliminate white spots on the image of the car, but only slightly disguises the image, while in median filter The results show that noise can be removed, but the details on the image are retained. However, this is of course obtained with the added computational burden of "sorting".