A Maximum Likelihood Registration Algorithm for Moving Dissimilar Sensors

Registration is the prerequisite for data fusion in sensor networks. A maximum likelihood(ML) registration algorithm is presented in this paper. Both complete and incomplete measurements provided by dissimilar sensors are taken into account. Sequential filtering technique is adopted to solve the problem of target state estimating with incomplete measurements. Then errors of target state estimate leaded by residual biases of sensors are analyzed. The likelihood function with respect to residual sensor bias in measuring coordinate system is derived. Finally, an iterated batch algorithm is presented to derive sensor biases. Simulations are performed to demonstrate the effectiveness of the presented algorithms.