摘要
针对低信噪比下红外图像中的弱小目标检测与跟踪问题,由于噪声的影响,信号不准确,图像不清晰。为解决上述问题,采用基于粒子滤波的检测前跟踪算法,直接利用传感器的原始数据,通过粒子滤波算法得到目标状态的后验概率分布,以目标出现概率作为检测目标的判断准则,检测出真实目标,并估计出目标在空间平面内的准确位置。并在抽取新生粒子时设定一个阈值,有效提高了抽取粒子的质量。对单个点的目标检测与跟踪系统进行了仿真。仿真结果表明,改进算法跟踪误差小,精度高,具有良好的检测与跟踪性能。
A track-before-detect algorithm based on particle filter was presented for weak target detection and tracking in infrared image in low signal to noise ratio. In terms of realization, the algorithm utilizes raw data of sen- sor, obtains posterior probability distribution of target through particle filter and takes target presence probability as detection criterion. Through detection criterion, the true target was detected and the aecurate position of target in the space plane was estimated. A threshold value was set when sampling newborn particles and improving the quality of particles. The simulation on a single target detection and tracking system shows that the algorithm has less tracking er- ror. higher orecision and better performance in detecting and tracking.
出处
《计算机仿真》
CSCD
北大核心
2012年第11期264-267,共4页
Computer Simulation
关键词
粒子滤波
检测前跟踪
后验概率分布
Particle filter
Track-before-detect
Posterior probability distribution
