Abstract

Sensor data fusion is the process of combining data collected from multi sensors of homogeneous or heterogeneous modalities to perform inferences that may not be possible using a single sensor. This process encompasses several stages to arrive at a sound reliable decision making end result. These stages include: senor-signal preprocessing, sub-object refinement, object refinement, situation refinement, threat refinement and process refinement. Every stage draws from different domains to achieve its requirements and goals. Popular methods for sensor data fusion include: ad-hock and heuristic-based, classical hypothesis-based, Bayesian inference, fuzzy inference, neural networks, etc. in this work, we introduce a new data fusion model that contributes to the area of multi-senor/source data fusion. The new fusion model relies on image processing theory to map stimuli from sensors onto an energy map and uses non-linear convolution to combine the energy responses on the map onto a single fused response map. This response map is then fed into a process of transformations to extract an inference that estimates the output state response as a normalized amplitude level. This new data fusion model is helpful to identify sever events in the monitored environment. An efficiency comparison with similar fuzzy-logic fusion model revealed that our proposed model is superior in time complexity as validated theoretically and experimentally.