Background subtraction using competing models in the block-DCT domain

Many image analysis applications rely on background subtraction as a pre-processing step. Hence it should be efficient and robust. We present a background subtraction algorithm that uses multiple competing hidden-Markov models (HMMs) over small neighbourhoods to maintain a locally valid background model in all situations. We use the DCT coefficients of JPEG encoded images directly to minimize computation and to use local information in a principled way. Region level processing is reduced to the minimum so that the extracted information that goes to higher level processing is unbiased.