Advanced intelligent sensing system using sensor fusion

A general three-layer hierarchical structure of an intelligent sensing system is discussed. In the bottom layer multiple sensors acquire information from the external world. In the top layer the main computer organizes available knowledge and sensed information and makes a decision or generates a final output. In the middle layer, signals from multiple sensors are integrated and fused with knowledge of the object. The essential features of the object are extracted there. An adaptive sound sensing system with visual cue is described as an example. The system acquires sound by multiple microphones, and then extracts nonstationary target sound out of a heavy noise environment using a visual cue signal. The prototype system and its algorithm have been tested experimentally