变电站遥视系统中运动目标检测

The Detection of Moving Target in the System of Substation Remote-supervision

运动目标检测是当前无人值守变电站遥视系统的智能视频监控技术需要解决的问题。采用基于负熵最大判据的快速算法(FastICA)又称为定点法(Fixed-point),通过随机梯度法调节分离矩阵W来达到对信号源进行优化的目的。试验表明,在背景亮度发生变化时,FastICA法也可以很好地检测出运动目标,并将其与背景分离,还可确定运动物体的运动轨迹。试验证明了FastICA是一种鲁棒性较强的运动目标检测方法,具有较强的抗图像背景灰度变化的能力。

The detection of moving targets is one of the urgent problems on intelligent video surveillance technology in the remote-supervision system of unattended substations. By the way of FastICA based on negative entropy optimum criterion, also called Fixed-point, and random gradient method, we adjust separation matrix W and achieve the aim of optimizing signals. Experiments show that FastICA can detect moving targets,separate them from background and locate the moving orbits in spite of the change of background luminosity. Experiments also show that it is a robust method and has a stronger capacity of anti-greyvaring of a picture.