A Target Behavior Law Mining Method Based on Similar Duplicate Record Detection

Abstract

In the complex marine environment, with the continuous enrichment and development of target detection methods, massive target track data is stored and accumulated in the database, which contains a large amount of information and knowledge. It is possible to analyze and predict the behavior and intention of the target by means of target behavior law mining. Most of the existing target behavior law mining methods are based on the idea of clustering. When working in the complex environment of the sea battlefield, the working steps are complicated and the effect is poor. Therefore, in combination with situation analysis requirements and based on the idea of similar duplicate record detection, this paper proposes a mining method of maritime target behavior law by defining multi-dimensional record matching similarity (MDRS). The experimental analysis is carried out on the simulated military scene. The results show that the proposed method can accurately and effectively mine the behavior of the target.