Wavelet-Based Multi-Sensor Optimal Information Fusion

The problem of multi-sensors information fusion is studied in time-frequency domain, and a new optimal criteria weighted by scalars for unify multi-sensor systems is presented. Wavelet transform is introduced in multi-scale signal filtering, the approximate component and details are both updated. The local sensor estimate is fused via an optimal algorithm weighted by scalars, then reconstruct at the finest scale. The method proposed: (1) "complete" multi-scale filtering for unify multi-sensor system, estimation performance is greatly enhanced; (2) distributed fusion weighted by scalars, only requires the computation of scalar weights, avoids the computation of matrix weights, the computational burden can obviously be reduced; (3) the simulation also shows it outperforms optimal fusion filter weighted by scalars and centralized multi-sensor fusion.