A linear multisensor PHD filters via the measurement product space

The probability hypothesis density (PHD) is the first moment of RFS. Its integral over any region gives the expectation number of targets in that region. In the finite set statistics (FISST) framework, the PHD recursion, or PHD filter, approximate the multi-target Bayes recursion. This paper deals with the multisensor PHD filter under a linear correlation condition through multisensor product space and the measurement dimension extension (MDE) approach, which remains the similar appearance like the conventional PHD filters except the product space and some parameters in the filters. However, in the product space the dimension extended measurements may greatly increase the computational load. Therefore, we propose a fast algorithm for the linear multisensor PHD (LM-PHD) filters to increase the running speed and with cost of slightly sacrificing the tracking performance.