DSP实现低信噪比条件下点目标检测与跟踪

针对嵌入式实时图像处理系统中 ,复杂背景低信噪比条件下红外运动点目标检测及跟踪所采用的处理算法 ,包括基于空域低通滤波的多图像平均法和自适应阈值分割法以及基于目标运动时空连续性目标运动轨迹拟合及目标运动轨迹参数预测算法 ,进行基于二维数组组合方式存储结构的算法复杂度估计 ,并将上述算法在TI公司的TMS32 0C6 2 0 1EVM板上进行实时性能评测 ,说明上述算法在基于具有软件流水性能的TMS32 0C6x系列DSP嵌入式实时图像处理系统中实现的可行性。

Point target detection and tracking with DSP under the condition of low SNR

In view of infrared dim point targets detection and single target tracking in embedded real time image processing system,the complexity estimation of image processing algorithms is provided,including the algorithm based on average of multiframe and adaptive threshold adopted to restrain background Gaussian noise and the algorithm of estimating parameters of target movement based on the recursion of least square.And evaluation reveals that it is feasible to implement these algorithms in embedded real time image processing system with DSPs of TMS320C6x series as processor.