An association algorithm of ship-group targets based on topological and attributive characteristics

When the sensors' orientating precision and their identification performance are poor, it is difficult to get effective group members association by their position or attributive characteristics. This paper presents a novel algorithm for ship-group targets association of spaceborne electronic reconnaissance data and optical imaging data. The contributions of the paper are: (1) Firstly, a new shape descriptor, named Point Pair Topological Characteristics (PPTC), is proposed to describe the topological characteristics of a ship-group. (2) Secondly, based on the PPTC descriptor, we present a basic probability assignment function, which is used to evaluate the comparability between PPTC. (3) Thirdly, we combine both PPTC and attributive characteristics together to construct the initial association probability matrix. Thus, the independent topological and attributive information are integrated to a synthetic association measurement. Experimental results on both synthetic and real word data demonstrate the effectiveness and robustness of our method.