Micromotion Feature Extraction and Distinguishing of Space Group Targets

Abstract

Traditional micro-Doppler (m-D) analysis theories largely focus on isolated targets, making these theories difficult to utilize in monitoring and recognizing space group targets. This letter proposes an algorithm for separating ballistic group targets based on the extraction of micromotion features. Modeling the ballistic targets as cone-cylinder models, the skeleton extraction method in morphology image processing is first utilized to suppress the sidelobes of range profiles. A sliding window whose length of frames can be adaptively changed according to the curve characteristics is established to separate the m-D curves. Then, different recording criteria are adopted considering different types of intersections. After separating the m-D curves, the group targets are distinguished by extracting the different tendency gradients of group targets. In addition to being capable of distinguishing group targets, the proposed algorithm is robust to influences from noise. Simulations are performed to validate the effectiveness of the proposed method.