sensorInformation fusion methods can be divided into four categories: combination, synthesis, fusion, and correlation.

Combination is a processing method that obtains output by combining multiple sets of data from multiple sensors in a parallel or complementary manner. This is a fundamental approach that involves coordinating, synthesizing, and selecting output methodssensor. Usually it is mainly applied at the hardware level. A typical example is to use visual detection to determine the orientation of an object, then accurately measure the distance of the object using a laser rangefinder, and display the distance parameters simultaneously on the video screen.

2. Comprehensive

Synthesis is an effective method for obtaining clear information in information optimization processing. A typical example is in virtual reality technology, where two separate cameras are used to simultaneously capture two images of different sides of an object. By combining these two images, an accurate and three-dimensional image of the object can be restored.

3. Integration

When willsensorWhen data sets are correlated or sensor data is correlated with internal knowledge models of the system to generate a new expression of information, this processing is called fusion. The definition of fusion mentioned here is narrow, and a typical example is the fusion of robot vision and touch to obtain optimized information on the space and shape of objects and the environment.

4. Related

Obtaining certain results by processing sensor information requires not only individual information processing, but also processing through correlation in order to obtain informationsensorThe relationship between data sets is used to obtain accurate information and eliminate useless and erroneous information. The main purpose of relevant processing is to integrate and optimize information from processes such as recognition, prediction, learning, and memory. For example, in passive sonar target recognition technology, the computer processes the received sonar information to obtain the motion information and characteristic power harmonic information of the ship target, and then processes the above data set (even including the data provided by shipborne radar) based on prior knowledge to ultimately determine which type of surface ship, submarine, or torpedo the target belongs to.