Due to the inherent genericity of graph-based representations, and thanks to the improvement of computer capacities, structural representations have become more and more popular in the field of Pattern Recognition (PR). In a graph-based representation, vertices and their attributes describe objects (or part of them) while edges represent interrelationships between the objects. Representing objects by graphs turns the problem of object comparison into graph matching (GM) where correspondences between vertices and edges of two graphs have to be found [14].In the domain of GM, over the last decade, Graph Edit Distance (GED) has been given a specific attention due to its flexibility to match many types of graphs [2]. GED has been applied to a wide range of specific applications from molecule recognition to image classification [9]. Researchers have shed light on the approximate methods that can find suboptimal solutions hopefully close to the optimal ones but the gap between optimal and suboptimal solutions has not been deeply studied yet.Roughly speaking, two main families of GM have been found in the literature: exact and error-tolerant GM. In this thesis, we propose adding a new GM family, called anytime GM. In order to demonstrate the benefit of having such a family, a new optimized GED algorithm which is based on depth-first search is put forward. This algorithm, referred to as DF, speeds up the computations of GED thanks to its upper and lower bounds' pruning strategy and its preprocessing step. Moreover, DF does not exhaust memory as the number of pending tree search nodes is relatively small thanks to the depth-first search where the number of pending nodes, or so-called partial edit paths, is |V 1 |.|V 2 | in the worst case where |V 1 | and |V 2 | are the numbers of vertices in G 1 and G 2 , respectively. Accordingly, DF outperforms the best-first GED algorithm (A * ) [11] in terms of speed, precision and classification rates. DF is able to provide not only one solution but successive solutions for a better and better quality according to available resources. The anytime version of DF, denoted by ADF, is able to find an initial, possibly suboptimal, solution quickly, keep it in the memory and then continue searching for improved solutions until the convergence to a provably optimal solution. The simplicity of the approach makes it very easy to use; it is also widely applicable. It can be used not only when an optimal solution is desired, but also when we want to see the evolution of the quality of the suboptimal solutions found at each time t. Generally speaking, Anytime GM provides an attractive approach to challenging GM problems, especially when the time Correspondence to: zeina.abu-aisheh@univ-tours.frRecommended for acceptance by David Vázquez Bérmudez