Automatic detection of stork nests on VHV towers

by João Fernandes¹, Maria Paula Queluz¹, Tomás Brandão², Francisco Azevedo³, João Gomes Mota^3
1IT/IST/Technical University Lisbon, ²IT/ISCTE-IUL, Lisbon, ³Albatroz Engineering, Lisbon

presented at ConfTELE 2013, Castelo Branco, Portugal, May 8-10, 2013.

Abstract

This paper proposes a method for the automatic detection of stork’s nests on the top of very-high voltage (VHV) towers, using video data from a regular inspection of aerial power lines. The method starts by detecting the towers, followed by a search for nests on each detected tower. At first, both detections are applied independently to the frames extracted from the videos. Afterwards, the temporal correlation between frames is exploited in an effort to improve the results. Tower detection relies on the search for straight lines that fit the tower model template. The search for nests is based on their typical color range and shape. Using five videos with a total duration of nearly one hour, the method was able to detect 83% of the existing VHV towers and
to correctly identify 79% of the towers in risk (i.e., towers with stork nests).

Keywords

VHV tower detection - nest detection - edge detection - straight line detection - motion estimation