Skip to content
DSML.KZ
Publications

Book chapter · CRC Press eBooks · June 27, 2024

A method for removing mixed noise in images

Narzillo Mamatov, Malika Jalelova, Abdurashid Samijonov, B.N. Samijonov

Narzillo Mamatov
Open via DOI

Abstract

Nowadays, digital images have become a widely used information tool in all areas of human activity. In particular, it is an important tool for data transmission and reception on social media platforms, product quality control and production in industry, and medical diagnosis. Digital images are made up of a collection of pixels, and in some cases, due to various factors, the color of the pixels in them changes abruptly or randomly, which creates noise in the images. For example, noise types found in digital images include Gaussian, Poisson, and salt-pepper noise. Many literatures have studied images affected by one or two types of noise. Images, however, might have a variety of noises blended together at once. Within this research, the above-mentioned three noise combinations are studied, and a mixed noise elimination approach is proposed. Most of the existing noise reduction approaches are dedicated to using only one method to eliminate one type of noise in the image. However, removing mixed noise with a single noise reduction filter may not provide the expected result. Therefore, it is necessary to select the most optimal existing filters for each type of noise and form hybrid filters. This, in turn, requires determining the sequence of applying appropriate filters to the noise sequence. For this, it is necessary to determine a sequence of filters suitable for a sequence of noises by performing a computational experiment. In this research, the image quality was evaluated based on the Blind Image Quality Index (BIQI) no-reference evaluation criterion, the image processing process was automated using this criterion, and an effective noise elimination rule was developed. In addition, in this work, the most optimal filter for effectively eliminating the noise sequence is determined from the six hybrid filters suitable for each noise sequence.

Image and Signal Denoising MethodsImage Enhancement TechniquesImage and Video Quality Assessmentmixed noise removalimage denoisingGaussian noise

More by Narzillo Mamatov