Abb. INB-Logo

Tools + Demos


Zurück / back

Robust Localisation of Circular Objects

 Fabian Timm, Thomas Martinetz, and Erhardt Barth

ObjectDetection.jpeg

 

The detection of circular objects (annuli or circles) is a fundamental problem in pattern recognition. In the case of eye tracking, for example, the centre of the pupil has to be determined accurately and in real time.  Also, in industrial applications such as automatic inspection of manufactured components, often circular objects have to be detected.

 

We created images of 150 x 150 containing an annulus or a circle with a random centre (near to the image centre) and a fixed size. The grey values of the annulus/circle were uniformly distributed within the interval [40,70] and the grey values of the background were uniformly distributed within [50,80]. Thus, the images contained white noise and the image contrast was very low. We further added multiplicative noise (speckle) and applied a motion blur to the images with white noise. For each object and each type of noise we created 100 images, which gives in total a dataset of 600 images.

 

The dataset can be downloaded here.

 

If you use this dataset for any publication, please refer to this: F. Timm, T. Martinetz, E. Barth. Robust Localisation of Circular Objects.