摘要
小波分析是图像与信号处理研究领域的一个重要工具,现有的小波分析方法大多集中于对线性小波的研究。智能交通中车型识别的基础是精确获取图像中汽车边缘和汽车特性,而图像中汽车边缘特性多表现为非线性性,采用非线性方法进行研究更为适合。文章在描述了非线性小波变换框架后,提出了一种用数学形态学实现非线性小波分解的方法来获取汽车边缘信息。为了正确识别车型,再进一步用修正后的数学形态学检测算子提取汽车边缘,得到了较为清晰的连续汽车边缘,为以后车型进一步分类提供了条件。经过实验检验该方法具有一定的可行性和稳定性。
Wavelet analysis is an important tool in the research area of image and single process.Most existing wavelet analysis approaches focus on linear wavelets.In Intelligent Transportation System(ITC),it is necessary to detect vehicle edge and get vehicle features.In images,vehicle edge contains nonlinear characteristic,so nonlinear wavelet and Morthological may be more suitable to describe and analysis those features.This paper presents a method how to use nonlinear wavelet and morphological to detect vehicle edge.To get vehicle edge information,it combines morphological nonlinear wavelet with morthological edge detection.First,based on the nonlinear wavelet multiresolution signal decomposition framework,it proposes a kind of morphological wavelet multiresolution signal decomposition and use morthological nonlinear Daubechies wavelet to detect vehicle edge.By it,it gets outline and location of vehicle.To sucessfully recognize and separate the unwanted components,to remove the varying facts and speck noise,the second step is using a morthological edge detection algorithm to detect edge again.At last,it gets a continue edge of vehicle.The result of this paper shows this scheme is satisfactory and efficiency.
出处
《计算机工程与应用》
CSCD
北大核心
2003年第12期63-65,95,共4页
Computer Engineering and Applications
基金
国家自然科学基金资助(编号:600734046)
关键词
形态学
非线性小波多分辨率分解
边缘检测
Morphology,Nonlinear wavelet multiresolution signal decomposition,Edge detection
