Output details
11 - Computer Science and Informatics
University of Westminster
3D gesture recognition with growing neural gas
<23> Originality: This paper extends the growing neural network algorithm to address the correspondence problem and track the motion of 3D points obtained from cloud sequences. The algorithm is applied in the real-time recognition of 3D gestures.
Significance: The work is inspired by the need in medical applications for responsive touch free human-computer interaction for the consultation of visual data. The paper brings together collaborative research from the University of Westminster, University of Alicante, Spain and Griffiths University, Australia.
Rigour: Mathematical modelling of the algorithm followed by extensive experimentation of 3D gestures that include self-occlusion and close finger proximity.