Akademik Veri Yönetim Sistemi
Uluslararası Mühendislik Araştırma ve Geliştirme Dergisi, cilt.15, sa.2, ss.558-573, 2023 (Hakemli Dergi)
The development of technology has increased the use of imaging devices. In addition, the development of hardware in recent years has improved the quality of images. However, noise in images is inevitable due to many factors during or after the image acquisition process. Image filters are used to remove these noises and this process is an important research area. One of the most basic steps of image processing is denoising. Diffusion filters have been used for denoising for many years, as they protect the edges. The common problem of diffusion filters, which are still being developed today, is the user intervention requirements. The diffusion threshold value is currently selected by the user. In this study, a fully adaptive threshold value selection method is presented with the help of dissimilarity transform. The main contribution of the study is the proposal of dissimilarity transformation for anisotropic diffusion filters. Experimental results show that the threshold values determined by the proposed method effectively provide denoising without being affected by noise variance and preserve the edges.