小波分析在视频检测中的应用

Application of wavelet analysis on video detection

针对复杂背景下机器视觉系统中运动目标的检测,提出一种基于小波分析的运动目标检测方法。首先,利用双Haar小波软阈值法,去除视频图像的噪声。然后,建立了一种基于小波变换的运动目标检测算法,直接在二进小波变换域提取运动区域,从而检测出运动目标。实验结果表明,本文方法可以有效地从图像序列中检测出完整的运动目标,而且在平滑噪声的同时还可以保护图像边缘细节不受损失,效果要好于传统帧差法。

Aiming at the problem of detecting moving object from complex background with machine vision system,A new method for moving object detection was proposed.First,we remove noises of video image by the double Haar wavelet transform.Second,a binary discrete wavelet transforms based moving object detection algorithm was put forward,which directly detected moving object in the binary discrete wavelet transforms domain.Experimental results show that the algorithm can pick up the whole moving object in the natural image series of complicated background.It can restrain noises effectively,meanwhile,protect the image edge.The experimental results are better than tranditional frame-subtraction method.