A practical adaptive approach for dynamic background subtraction using an invariant colour model and object tracking

摘要: In this paper, three dynamic background subtraction algorithms for colour images are presented and compared. The performances of these defined as 'Selective Update using Temporal Averaging', Non-foreground Pixels the Input Image' Median' only different pixels. Then an invariant filter a suitable motion tracking technique, object-level classification is offered that recognises behaviours all foreground blobs. This novel approach, which selectively excludes blobs from frames, included in methods. It shown produces correct image each input frame. advantages third algorithm are: it operates unconstrained outdoor indoor scenes. Also able to handle difficult situations such removing ghosts including stationary objects efficiently. Meanwhile, algorithm's parameters computed automatically or fixed. efficiency new confirmed by results obtained on number sequences.