Colour image denoising by eigenvector analysis of neighbourhood colour samples
[EN] Colour image smoothing is a challenging task because it is necessary to appropriately distinguish between noise and original structures, and to smooth noise conveniently. In addition, this processing must take into account the correlation among the image colour channels. In this paper, we intro...
| Autores: | , , |
|---|---|
| Tipo de recurso: | artículo |
| Fecha de publicación: | 2020 |
| País: | España |
| Institución: | Universitat Politècnica de València (UPV) |
| Repositorio: | RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia |
| Idioma: | inglés |
| OAI Identifier: | oai:riunet.upv.es:10251/166837 |
| Acceso en línea: | https://riunet.upv.es/handle/10251/166837 |
| Access Level: | acceso abierto |
| Palabra clave: | Colour image filter Colour image smoothing Eigenvectors Gaussian noise Principal components Vector filter MATEMATICA APLICADA |
| Sumario: | [EN] Colour image smoothing is a challenging task because it is necessary to appropriately distinguish between noise and original structures, and to smooth noise conveniently. In addition, this processing must take into account the correlation among the image colour channels. In this paper, we introduce a novel colour image denoising method where each image pixel is processed according to an eigenvector analysis of a data matrix built from the pixel neighbourhood colour values. The aim of this eigenvector analysis is threefold: (i) to manage the local correlation among the colour image channels, (ii) to distinguish between flat and edge/textured regions and (iii) to determine the amount of needed smoothing. Comparisons with classical and recent methods show that the proposed approach is competitive and able to provide significative improvements. |
|---|