Grammar-based image segmentation with taking care about runtime and accuracy constraints

Theory of Language and Compilation (TLC) is a generative mechanism that can recognize all the words for a given language. Thus, the aim of this paper is to exploit this theory on image processing domain and particularly on image segmentation field. We achieve to define an image as a set of letters based on image vocabulary and use grammar formalism for describing an object. This lasted one is assimilated as word recognized by an automaton. We will show how the idea of the TLC is applied for real and synthetic image with possibility of taking care about runtime and accuracy constraints. This type of task helps enough at some point in real time image and video processing application.