An uncertainty analysis-based algorithm track fusion

The common track fusion algorithms in the current multi-sensor system have some defects such as serious imbalance between accuracy and computational burden, inefficacy to uncertainties, especially high fusion errors for inflection points of tracks and so on. In response to these defects, a track fusion algorithm based on a two-stage paradigm of uncertainty analysis and track state estimate fusion is presented in the paper, and its implementation process in the multi-sensor and multi-target environment is discussed. The algorithm is used to delete poor tracks by analyzing the information demand for the fusion and to quantify the uncertainties of every track by adopting the standard entropy. Then the data points with high errors in the tracks to take part in the fusion are corrected by orthogonal polynomial regression. In contrast to existing methods, the algorithm takes full consideration of and deals with the uncertainties effectively, plays down the high fusion errors for inflection points of tracks, and approaches a high accuracy with less computational burden, thus gaining a tradeoff between accuracy and computational burden. Simulation results show effectiveness and superiority of the algorithm.