Unattended ground sensors (UGS) are widely used to monitor human activities, such as pedestrian motion and detection of intruders in a secure region. Efficacy of UGS systems is often limited by high false alarm rates, possibly due to inadequacies of the underlying algorithms and limitations of onboard computation. This paper presents a symbolic method of feature extraction and sensor fusion, which is built upon the principles of wavelet transform and probabilistic finite state automata (PFSA). The relational dependencies among heterogeneous sensors are modeled by cross-PFSA, from which low-dimensional feature vectors are generated for pattern classification in real time. The proposed method has been validated on data sets of seismic and passive infrared (PIR) sensors for target detection and classification. The proposed method has the advantages of fast execution time and low memory requirements and is potentially well-suited for real-time implementation with onboard UGS systems.