Local information and global information fusion-based target classification identification method

Abstract

The invention discloses a local information and global information fusion-based target classification identification method. The method comprises the steps of extracting local information of a training sample set through a clustering thought to obtain K types of data, and performing calculation to obtain a clustering center corresponding to each type of the data; obtaining an initial classification result for each piece of the data in each type, performing correction on the initial classification result, and performing calculation to obtain a deviation between a corrected classification result and a true value; calculating a distance weight factor of the initial classification result of each piece of the data in each type and the true value; and for a target sample, calculating a distance between the target sample and each clustering center to adaptively select a corresponding effective correction matrix for correcting the initial classification result of the target sample, thereby obtaining a final classification result of a target. The clustering thought is applied to a process for improving the accuracy of data classification; and through the local information and global information fusion-based classification correction method, an output of a classifier is corrected to enable the identification output of the classifier to be closer to the true value, so that the classification precision of target identification can be effectively improved.

Description

【Technical field】 [0001] The invention belongs to the technical field of target recognition, and in particular relates to a target classification and recognition method in which local information and global information are fused. 【Background technique】 [0002] Multi-sensor target automatic recognition technology in confrontation environment plays a very important role in the military field. Due to the rapid development of methods such as non-cooperative target deception and interference, and the continuous use of multiple and heterogeneous sensors, the characteristics of uncertainty and high conflict in observation information have become increasingly prominent. In this context, how to improve the accuracy of target recognition has become an increasingly important issue. [0003] At present, the solutions to the target recognition problem of uncertain observation data mainly include multi-classifier fusion and construction of new classifiers. Multi-classifier fusion techn...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a target classification and recognition method that combines local i...

Images