Author(s)

Jiale Qiao, Jindong Zhang, Yuze Wang, Chenhui Yu, Sai Gao

Corresponding Author

Jiale Qiao

ABSTRACT

This paper presents a multi-sensor information fusion system based on SVM (Support Vector Machine) and D-S evidence theory, and applies it to embedded devices. Each time the data collected by multi-sensor is received, the data will be brought into the decision tree of the binary classification SVM and the decision vector will be output. The decision result of the system considers that the reliability of different sensors is different, so the probability redistribution of decision vectors is carried out by using the reliability. Finally, D-S evidence theory is used to fuse each decision vectors after probability redistribution. The simulation results show that the algorithm takes into account the different reliability of the sensor and is simple in calculation and short in processing time. It is suitable for embedded devices with low computing speed.

KEYWORDS

Multi-sensor, embedded device, simulation results show that algorithm