复杂背景下运动点目标的检测算法

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。

Detection algorithm for moving point target with complex background

Detection of moving point target in infrared sequence images with complex background is the emphasis and difficulty of target detection. A new simple detection algorithm for moving point target is presented. The algorithm using different mean filtering, firstly, inhibits the background to improve the SNR of the images according to the differences among the point target, background interference and noise. The second step is to obtain new sequence images by using a new threshold method. The target is identified effectively by a modified discontinuous frame difference optical flow field algorithm. From the simulation experiment, the algorithm can successfully overcome the shortcoming of the classical optical flow field algorithm that couldn't detect the target whose displacement is less than one pixel between two continuous frames. Compared with other algorithms, the algorithm is not only computationally simple, but also has a high detection performance.