A model-based method for the recognition of 2D occluded objects ispresented. Initially, to reduce the number of models to be consideredfor identification from the model database, two types of signatures areused. One is based on curvatures and the other on G curve. Next, exactmatching is used to find the correct model. The results are illustratedfor an occluded scene consisting of industrial objects. The relativeperformance of the proposed two types of signatures is also given
展开▼