RECOGNITION OF OCCLUDED-OBJECTS BY LOCAL FEATURE SEQUENCING
Jianming Song, Paul Gaillard · 1988
We present a new method for 2-D occluded-object recognition. In this method, objects are segmented then represented by local and (rotation-translation) invariant feature-vectors. A decision tree is used to match similar local features, which allows an efficient and complete selection of potential models. In order to validate each potential model, we explore the geometric relationships between feature-vectors using a feature sequencing process. An example of a 2-D application is given to show the effectiveness of the method.