基于指数平滑的动态规划检测前跟踪算法

针对传统动态规划检测前跟踪(Dynamic Programming Track-Before-Detect, DP-TBD)算法在低信噪比(Signal to Noise Ratio, SNR)环境下跟踪性能较差以及容易出现团聚效应的问题, 提出一种基于指数平滑法的DP-TBD算法.该算法的创新之处在于:利用指数平滑法预测当前帧的目标状态, 当对当前帧代价函数进行优化时利用预测的目标状态对前一帧搜索窗内的代价函数进行加权.仿真结果表明, 文中所提算法能够有效抑制团聚效应, 且算法的检测性能和跟踪性能都比传统算法有所提高, 并且信噪比越低, 性能提高越明显.因此文中算法相对于传统算法来说更适用于低信噪比环境.

Dynamic programming track-before-detect algorithm based on exponential smoothing method

Accounting for the issues of bad tracking performance and agglomeration phenomenon of conventional dynamic programming track-before-detect (DP-TBD) algorithm in low signal to noise ratio(SNR) situation, a DP-TBD algorithm based on exponential smoothing method is proposed in this paper. The innovation lies in an algorithm that the merit function in search window at previous frame is weighted with the predicted target state which is obtained by exponential smoothing method while the merit function at current frame is optimized. Simulation results indicate that the proposed algorithm can mitigate the agglomeration phenomenon efficiently and has better detection and tracking performance over the conventional algorithm. Furthermore, the lower the SNR is, the greater the improvement will be. Therefore, the proposed algorithm is more applicable in the low SNR environment than the conventional ones.