Abstract
In industry, there is an increasing demand for intelligent and efficient approaches in order to automate the production process. Modern photogrammetry can make a substantial contribution in this field by using methods of machine vision. Often, objects must be automatically recognized in images. The requirements that are induced by the industrial background are very high: To facilitate a fast production process, it is necessary to recognize objects in real-time. Furthermore, objects that consist of several moving parts occur in many applications. These compound objects complicate the recognition considerably. This paper proposes a novel approach for recognizing compound objects in real-time while only using off-the-shelf hardware. The approach is based on a hierarchical model of the compound object. This leads to substantial improvements of the achieved performance. A further feature of the approach is its high degree of automation in order to achieve an easy use, and hence to facilitate a broad applicability.
Translated title of the contribution | Recognition of laws of objects in pictures under real time requirements |
---|---|
Original language | German |
Pages (from-to) | 184-194 |
Number of pages | 11 |
Journal | ZFV - Zeitschrift fur Geodasie, Geoinformation und Landmanagement |
Volume | 129 |
Issue number | 3 |
State | Published - 2004 |