Color- and texture-based salient map hierarchy

In content-based image retrieval (CBIR), the extraction of salient maps delivers regions focusing on the predominant objects of the images. This kind of description is aimed to specify the queries and to organize the image database index. This paper proposes a process providing a hierarchy of salient maps. This hierarchy is a multiscale description from fine to coarse maps using an adapted similarity measure based on texture and color specificities.