Redundant structure detection in attributed adjacency graphs for character detection in comics books

Graphs are popular data structures used to model pair wise relations between elements from a given collection. In image processing, adjacency graphs are often used to represent the relations between segmented regions. Such graphs can be compared but graph matching strategies are essential to find similar pat- terns. In this paper, we propose to detect the recurrent characters of a comics book. In this method each panel is represented with an attributed adjacency graph. Then, an inexact graph matching strategy is applied to find redundant structures among this set of graphs. The main idea is that the same character will be repre- sented by similar subgraphs in the different panels where it appears. The two-step matching process consists in a node matching step and an edge validation step. Experiments show that our approach is able to detect redundant structures in the graph and consequently the recurrent characters in a comics book. The originality of our approach is that no model is required, the algorithm detects all by itself all redundant structures.