2016 13th Conference on Computer and Robot Vision (CRV) (2016)
Victoria, BC, Canada
June 1, 2016 to June 3, 2016
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/CRV.2016.56
We present a novel method for detecting waving humans at long ranges in outdoor environments, using a consumer video camera on a mobile ground-based robot. The proposed algorithm analyzes the average pixel intensity of motion-containing regions in an image stream, identifying those regions which show a strong periodic signal in the frequency range of human waving gestures. The system achieves robustness to sporadic false positives such as waving trees, flags and walking pedestrians by using a filter to reject nonstationary and transient detections. In real-world experiments we determine the effective detection range of our algorithm and show that a robot equipped with a low-resolution consumer camera is able to approach a single waving human from a starting position up to 35 meters away when the person is roughly 20 pixels high, even in the presence of human and non-human periodic distractors such as foliage.
Cameras, Robot vision systems, Time series analysis, Detectors, Time-frequency analysis
J. Bruce, V. Monajjemi, J. Wawerla and R. Vaughan, "Tiny People Finder: Long-Range Outdoor HRI by Periodicity Detection," 2016 13th Conference on Computer and Robot Vision (CRV), Victoria, BC, Canada, 2016, pp. 216-221.