Résumé
This paper deals with the problem of object detection from digital maps. We are interested in detecting objects in a map which are defined in the legend. We will explore different similarity measures to compare the legend objects to those detected in different areas of the map. Our object detection method is evaluated on maps representing wastewater networks. In particular, we are interested in the detection of objects that represent lifting stations and manholes. The ultimate goal, after detecting correctly the target objects, is to repair misfunctions or inconsistencies in the water supply or evacuation network. The experimental results show that our similarity measures give good accuracy results on the detection of the objects of the legends.