为了降低红外焦平面告警系统中较高的虚警率, 提出了一种速度限制基础上基于曲率限制的目标搜索算法。以视场角5.3°×4.0°凝视型红外告警系统为例, 在MATLAB中对红外图像序列进行了目标搜索实验, 验证了该算法的可行性。按照蒙特卡罗法思路模拟了包含大量随机噪声的目标场景, 研究了该算法相比传统多帧检测算法对随机噪声的抗干扰性。对该算法的搜索帧数进行了扩展, 给出了不同搜索帧数的搜索效果和时效性表现。

A curvature threshold based target searching method is reported in order to reduce the high false alarm rate in staring infrared warning system. A staring infrared warning system with view field of 5.3°×4° is chosen to simulate the infrared search process in MATLAB, proving the feasibility of the proposed algorithm. Target scene with a large amount of random noise is simulated according to Monte-Carlo method to compare the search performance of the proposed algorithm with the classical 3D matched-filter method. The number of search frames is discussed to expand the proposed algorithm, and the search performance and temporal limit are also reported for different search frames.