Multi-target Detection of Underwater Vehicle Based On Multi-sensor Data Fusion

This paper describes a method of forward-looking sonar, single camera and ultrasonic proximate sensor data fusion applied for underwater multi-target identification and positioning. A set of multi-target detection approach was proposed based upon the characteristics of sonar and the geometry of underwater objects. In long distance, multitarget would be detected by sonar; when targets were in the distance of 5m, they would be identified and roughly positioned by integrating sonar and camera image information; in working range 2m of ultrasonic proximate sensor, camera image and ultrasonic proximate sensor data fusion was used to achieve multi-target identification and accurate positioning. Experimental results obtained in an indoor pool demonstrated the availability of this method.