序列图像的自适应背景提取及车型分类

Adaptive background update of sequence image and vehicle classification

提出了一种基于计算机图像视觉模式识别理论针对车型识别的自适应背景提取得的新方法。基于彩色图像的特征,利用三通道的差分图象获得对象掩模。并根据两种不同的情况分别采用了不同的背景更新方法;最后利用软传感器来触发车型提取工作。与同类方法相比,该方法速度快,准确率高,能很好地满足实时性的要求。

This paper presents a novel method of extracting the self-adaptation background applying to the vehicle shape recognization.Based on the colorful images characters,the object mask is obtained using RGB 3-channels difference images.Two different background update methods are chosen according to 2 situations and then the vehicle shape extraction is triggered through soft sensor.Compare with the other approach,it shows a higher efficiency and accuracy,which enables the excellent, application in the real-time system.