Today object recognition and pose estimation is an urgent problem in automation. In industrial environments the shape of objects is known a priori. In most cases they are rigid and often they have plane surfaces, bolts or at least drill-holes for the assemblage. But the recognition has to be fast and efficient. Also the objects may be rather similar so it can be impossible to make a distinction based on a single view.
Recognition and pose estimation requires the matching of structures in an image with corresponding structures in the model. As the number of comparisons necessary for this process is very large, it is urgent to optimize the efficiency of this operation. So the goal of this work was to optimize the number of parameters that have to be compared, already in the modeling phase.
Several approaches use geometric primitives like vertices, edges and surfaces for object recognition    . Additional informations about objects are topological characteristics  which can also be used. To increase the efficiency of the feature extraction process, a restriction on small areas of the range image and local structures of the model is made .
In this work we introduce a method of modeling three dimensional objects suitable for fast recognition of objects in industrial environments. The model consists of plane and cylindrical surfaces which are very common. The restriction on geometric structures reduces the number of parameters relevant for the matching. But the modular structure of the model allows the addition of every surface description that may be useful for future applications.