The basic principle of multi-sensor data fusion is to make full use of the multi-sensor resources, through the reasonable control and use of these sensors and observation information, the redundant or complementary information of various sensors in space or time is combined according to some criteria to generate the consistency interpretation or description of the object under test. The purpose of this paper is to derive more useful information based on the observation information separated by each sensor, and derive more useful information through the optimization and combination of information. The ultimate goal is to improve the effectiveness of the whole sensor system and eliminate the limitation of single or small number of sensors by using the common or joint operation advantages of multi sensors.
The basic goal of data fusion is to deduce more information through data combination rather than any individual element in input information, which is the result of the best synergy. The multi-sensor data fusion system can maximize the information of the target and environment by using the multi-sensor resources effectively. There is essential difference between multi-sensor data fusion and classical signal processing method. The key lies in that the multi-sensor information processed by information fusion has more complex forms, and can be processed at different information levels, including data layer, feature layer and decision-making layer.
The structure of data fusion includes series, parallel and hybrid fusion.
1. when series fusion is conducted, the current sensor shall accept the output result of the previous level sensor. Each sensor has the function of receiving information processing information and information fusion. The processing of each sensor has a great relationship with the information form of the previous sensor output. The last sensor combines all the information of the front stage sensor, and the output result will be the conclusion of the series fusion system.
2. when parallel fusion is conducted, each sensor directly transmits the output information to the sensor fusion center, without any influence between the sensors. After the fusion center comprehensively processes the information in the appropriate way, the output results will be obtained.
3. the hybrid fusion method is a combination of serial fusion and parallel fusion, or the whole serial local parallel or the overall parallel local serial.
