Detection of a Region of Interest in the Images Based on Zipf Laws

Models of power laws such as Zipf's law and inverse Zipf can be applied to images. For this, we define patterns and coding that reduce the number of different patterns to characterize the complexity of the structural contents of the image. It allows to envisage applications of these models such as the detection of a region of interest in an image, different codings can be used to define the patterns, the image is segmented into sub-images and is performed by a classification of sub-images according to the properties of the laws associated powers. The result of this classification is used to detect a region of interest in the image. To improve the results given by Zipf's law or inverse Zipf individually, we used simultaneous the methods of Zipf and inverse Zipf and two different encodings, the general rank and 9 classes. Regions of interest obtained with these methods correspond to those who are recognized by a human observer.