Convex Frechet Distance in Specific Emitter Identification

Specific emitter identification (SEI) currently plays an important role in electronic warfare. This technology is used to distinguish different emitters of the same system through a certain distance function to measure the similarity between the fingerprint feature of unknown target and each standard feature in the template library, and finally make an identification decision. Among most existing distance functions, frechet distance has shown remarkable performances both in the recognition accuracy and computational complexity, compared to hausdorff distance and dynamic time warping. To enlarge the advantages of original frechet distance, a novel improved distance named convex frechet distance, is proposed in this paper. By fully considering the structural property of the fingerprint feature, convex frechet distance achieves maximum combination of the algorithm principle and target characteristics, and further improves the recognition accuracy and computational complexity.