Multiple target recognition based on blind source separation and missing feature theory

This paper considers the problem of classifying simultaneous multiple ground vehicles using their acoustic signatures recorded by unattended passive acoustic sensor array. The proposed approach relies on the blind source separation (BSS) algorithm based on time-frequency signal representations. Instead of estimating mixing parameters as the original algorithm do, we get the missing feature mask from the BSS step. Then an acoustic signature recognizer based on the missing feature theory recognizes each acoustic source. Recognition results are presented for several simultaneous vehicle acoustic signals. Compared with familiar ways, using both the missing feature theory and BSS algorithm results in high performance improvement