论文部分内容阅读
在复杂背景红外序列图像中,运动点目标的检测一直是研究的重点和难点。介绍了一种新的复杂背景下运动点目标的检测算法。首先根据点目标、背景干扰和噪声在红外图像中的差异,运用窗口大小不同的均值滤波器进行背景抑制以提高图像的信噪比,然后用一种门限法得到新的分割序列图像,最后采用改进后的隔帧差分光流场算法可有效地检测出点目标。仿真实验表明该算法优于传统光流场算法,能够检测帧间位移小于一个像元的运动目标,具有较好的检测性能,且实时性强。
In complex background infrared sequence images, the detection of moving-point targets has been the focus and difficulty of research. A new detection algorithm of moving-point target under complex background is introduced. Firstly, based on the difference of point target, background noise and noise in infrared images, background suppression is performed by using mean filter with different window sizes to improve signal-to-noise ratio of image, and then a new segmentation sequence image is obtained by a threshold method. Finally, The improved differential optical flow field algorithm can effectively detect the point target. Simulation results show that the proposed algorithm outperforms the traditional optical flow field algorithm and can detect motion targets with less than one pixel displacement between frames. It has better detection performance and real-time performance.