Access to SPIE eBooks is limited to subscribing institutions. Access is not available as part of an individual subscription. However, books can be purchased on SPIE.Org
Chapter 8:
Voting Logic Fusion
Voting logic fusion overcomes many of the drawbacks associated with using single sensors or sensors that recognize signals based on only one signature generation phenomenology to detect targets in a hostile environment. For example, voting logic fusion provides protection against false alarms in high clutter backgrounds and decreases susceptibility to countermeasures that may mask a signature of a valid target or cause a weapon system to fire at a false target. Voting logic may be an appropriate data fusion technique to apply when a multiple sensor system is used to detect, classify, and track objects. Figure 8.1 shows the strengths and weaknesses of combining sensor outputs in parallel, series, and in series∕parallel. Generally, the parallel configuration provides good detection of targets with suppressed signatures because only one sensor in the suite is required to detect the target. The series configuration provides good rejection of false targets when the sensors respond to signals generated by different phenomena. The weaknesses of these configurations become apparent by reversing their advantages. The parallel is subject to false target detection and susceptibility to decoys, since one sensor may respond to a strong signal from a nontarget. The series arrangement requires signatures to be generated by all the phenomena encompassed by the sensors. Thus, the series configuration functions poorly when one or more of the expected signature phenomena is absent or weak, such as when a target signature is suppressed. The series∕parallel configuration supports a voting logic fusion process that incorporates the advantages of the parallel and series configurations. These are rejection of signatures from decoys, clutter, and other nontargets and detection of targets that have one or more of their signature domains suppressed. We will show that voting fusion (one of the feature-based inference fusion techniques for object classification) allows the sensors to automatically detect and classify objects to the extent of their knowledge. This process does not require explicit switching of sensors based on the quality of their inputs to the fusion processor or the real-time characteristics of the operating environment. The sensor outputs are always connected to the fusion logic, which is designed to incorporate all anticipated combinations of sensor knowledge. Auxiliary operating modes may be added to the automatic voting process to further optimize sensor system performance under some special conditions that are identified in advance.
Online access to SPIE eBooks is limited to subscribing institutions.

Back to Top