论文部分内容阅读
基于运动点目标在邻帧差分图像中所具有的近邻反相特征,即运动点目标的两个位置相邻近、灰度值一正一负,提出一种在复杂背景下,基于红外序列图像的运动点目标检测算法。本算法利用该特征在邻帧差分图像中检测反相点对,进而构造反相点对矢量图,最后依据累积反相点对矢量图中多矢量首位相接的连续性检测出运动的点目标。文中给出并证明应用本算法能以概率1 检测到运动点目标的收敛性定理。对典型复杂背景下 10 幅 1000 帧图像的仿真结果表明,当信噪比大于或等于 1.5 时,可以有效检测出运动点目标。
Based on the near-neighbor inversion feature of the moving-point target in the adjacent-frame differential image, that is, the moving-point target is adjacent to two positions and the gray-value is positive and negative, a new algorithm based on infrared sequence image Motion point target detection algorithm. This algorithm uses this feature to detect the inversion point pairs in the adjacent frame difference image, and then constructs the inversion point pair vector diagram. Finally, the moving point target is detected according to the continuity of the cumulative inversion point to the multi-vector first phase connection in the vector image . In this paper, we prove and prove that this algorithm can detect the convergence theorem of moving-point targets with probability 1. The simulation results of 10 1000-frame images under the typical complex background show that the moving-point target can be effectively detected when the signal-to-noise ratio is greater than or equal to 1.5.