Speaker / Predavač : Dr Joakim Lindblad
Title / Naslov: How to improve perimeter estimation by using grey level information
Abstract / Abstrakt:
A novel method that provides precise and accurate estimates of the length of the boundary of objects (perimeter), by taking into account grey-levels on the boundary of digitized objects, will be presented. Assuming that intensity is proportional to area coverage of a pixel, we show that the presented method provides error free measurements of the length of straight boundary segments in the case of non-quantized area coverage values. For a more realistic situation, where pixel intensities are quantized, we present optimal estimates that minimize the maximal estimation error. The method utilizes only a small neighbourhood and is easy to implement. The so far obtained experimental results confirm a very good performance of the newly suggested estimator, both for straight line segments, and for more general objects.