Using Fractal Dimension as Texture Discriminator for Context Base Image Retrieval

The objective of the present work is to evaluate the potential of the use of the image fractal dimension as textural feature in a content-based image retrieval system. In order to compare and classify the regions of an image, we have used two distances between two partitions of the image: (1) a classic distance computed with some features (contrast, energy, entropy, homogeneity, correlation) extracted form cooccurrence matrix and (2) a distance between the fractal dimensions of the two regions. The experiments proved that the degree of confidence is increased when adding the fractal dimension to the other textural features.