Automatic Place Determination using Colour Histograms and Self-Organising Maps

摘要: In this paper we propose a model-free appearance-based method to automatically determine places as landmarks for topological navigation. Most of the current approaches selecting are based on template models or complex feature detectors. We use modified colour histograms, more precisely an entropy-constrained 3D~colour clustering image features which adapt distribution environment. An unsupervised neural network learning strategy is used by histograms. Results from experiments in indoor environment with robot equipped panoramic camera show that places, were clustered histogram space refer physically close positions world domain large degree and can be navigation purposes.