Online globally consistent mosaicing using an efficient representation

摘要: This work addresses the problem of mosaic construction from image sequences. We propose an incremental, globally consistent method in which each new frame sequence is aligned to that has been constructed up time instant. efficient, stochastic representation, color pixel represented with a mixture Gaussians. The parameters Gaussian are learned on-the-fly. proposed particularly suited for long sequences, since it prevents accumulation small alignment errors and maintains constant memory requirements. Results synthetic real sequences presented.