Robust estimation of the optical flow based on VQ--BF

摘要: To increase the robustness of optical flow computation in context mobile robotics, we introduce an image filtering process based on codebook computed by Vector Quantization (VQ) algorithms, which usually are used for compression and codification purposes. The Self Organizing Map is to compute adaptively vector quantizers color imaged sequences. each sequence then as a smoothing filter, VQ Bayesian Filter (VQ-BF), preprocessing images sequence. filtered basis via pixel region correlation algorithms. gives good estimation at edges, whereas robust dense flow.