Cluster in images
- Giuliodori, M. Andrea
- Lillo Rodríguez, Rosa Elvira
- Peña Sánchez de Rivera, Daniel
Editorial: Universidad de Murcia. Departamento de Estadística e Investigación Operativa
ISBN: 978-84-691-8159-1
Año de publicación: 2009
Congreso: Congreso Nacional de Estadística e Investigación Operativa (31. 2009. Murcia)
Tipo: Aportación congreso
Resumen
A color image is mathematically composed of a set of three matrices that represent the color intensity of the pixels that constitute the whole image. By considering that every matrix is composed of variables and observations, we can use the classic multivariate data analysis as a tool for the statistical interpretation of a picture. In particular, one of the goals in exploratory image analysis is to classify images without eye observation. In this work, we analyze and interpret the statistical structure of an image in order to nd distinctive features to build clusters. We show a general method to classify and group the pictures considering three classication variables: the local variability, the eective variance and the spatial correlation.