基于有限线积分变换的序列图像运动目标检测

运动目标检测主要是从序列图像中将变化区域从背景中分割出来，它影响着运动目标能否正确地分类和跟踪。因此，运动目标检测是智能视频监控技术中的关键问题之一。有限线积分变换（finitelineintegraltransform，FLIT）作为近年来一种多尺度几何分析的新方法，能有效提取图像中的线性特征。检测方法是在FLIT的基础上再结合背景差法来实现的。具体而言，先对图像序列中的每帧网像（包括参考帧）作固定模板下的FLIT，再用当前帧的FLIT减去参考帧的FLIT，然后对不同方向提取的运动目标信息进行综合，最后采用数学形态学的相关处理来消除噪声。实验结果表明，使用该方法可以承受整体的或局部的、缓慢的或突变的光线变化，能有效地检测出运动目标。

Moving target detection based on finite line integral transformation in image sequence

The aim of moving object detection is to depart variety area from background through the image sequence, which affects whether the moving object is correctly classed and followed or not. Hence, moving object detection is one of the critical issues in intelligent visual surveillance. Finite line integral transform （FIAT）, as a new method based on the concepts of multiscale geometric analysis, can deal images with line singularities. Our work on moving object detection is based on FLIT, combining present frame. Firstly, FLIT with a fixed scale the difference between background and present of different orientations. Finally, we use Experimental results demonstrate changes in illuminations to detect with the difference between background and that the method can bear the the moving object effectively. whole or local, quick or slow