Ground target track bias estimation using opportunistic road information

Track-to-track fusion is widely used by large-scale distributed surveillance systems to integrate multisensor tracking data. A critical step prior to track-to-track fusion is track-to-track assignment. However, when ground targets are tracked by airborne sensors, the target tracks contain not only tracking sensor errors but also navigation errors of the sensor platforms. If the errors are not compensated for properly, biased tracks create false assignments and lead to erroneously fused tracks. In this paper, track biases produced by individual trackers are characterized in terms of translation and rotation (and to a lesser extent by scaling) with respect to the common reference frame in which the track-to-track fusion will take place. Opportunistic information about the roads on which ground targets are moving is explored to estimate the track biases, akin to a system calibration, which can be used not only to remove biases from past and present tracks but also to provide corrections for future estimates. The proposed bias estimate method is based on binary image matching to estimate rotation and translation. A two-dimensional fast Fourier transform (2D FFT) is used to implement 2D search and correlation efficiently. Simulation results are presented to illustrate the proposed opportunistic road information based bias estimation (ORIBE) method and its performance as a function of target track accuracy and spatial resolution in forming track and road map images.