Composition and Detection Rate of a Symmetry Axis Localization Algorithm for Digital Images

In this paper we describe a novel approach for the extraction of object features from a digital image captured in an industrial environment. The developed algorithm localizes the projected position of the symmetry axis of cylindrical objects. Conventional approaches to this task are often based on e.g. edge detection or image matching to determine object features. These methods tend to be time consuming and are not suited for online applications. The proposed algorithm utilizes sufficient statistics to reduce the amount of data which needs to be evaluated. In addition it determines the symmetry axis not from local but from global image features. This strategy significantly reduces the computation load, while purveying the desired information. Finally, we provide an application example and discuss some properties of the presented algorithm.