摘要
由于绝大多数人造物体图象具有规则的几何特性 ,其空间相关性较强 ,因此本文提出了一种基于小波变换极大模序列相关特性的二维物体特征点提取及非线性滤波的识别方法 ,并将此方法用于汽车正面图象的识别任务 。
As the most images of man-made objects have regular geometric at- tributes and strong space correlation,we propose a new method to abstract the attribute points of2 -D objectbased on WT sequence maximum correlation attributes and non-linear filter.We use this method in the task of automobile frontimage recognition and geta sat- isfactory result.
关键词
图象处理
小波变换
非线性滤波
汽车
车型识别
Image processing
Non-linear filter
Pattern recognition
/Wavelet transformation