Force histograms and radial density for invariant image retrieval

Abstract

In Content-Based Image Retrieval (CBIR), indexing techniques based on global features that relate color and texture are commonly used to represent a description of images. However this approach is unable to capture local information of parts of the image that contain different characteristics. Therefore, some necessary local features of image may be might do not taking in account. This research introduces the new methods in which using the force histograms and radial density to capture some invariant local features. The results show that the force histograms and radial density use to solve the problem caused by variations in scaling, rotation, and translation. Furthermore this technique provides that efficient features to retrieve the sought image where difference pattern of objects may affect the retrieval accuracy.