摘要
边缘作为一个重要特征是图像目标检测中基础而又困难的一个问题。用常规方法进行边缘检测时 ,噪声会影响到边缘特征提取的准确性。为了减小噪声对图像边缘特征的影响 ,改善边缘特征的定位精度 ,本文提出了一种新的图像边缘特征提取算法。该方法利用小波变换天生的多尺度特性 ,对小波变换各尺度下的细节图像用互能量交叉的方法进行噪声抑制和边缘识别。最后作了实例验证。
WT5BZ]The boundary of the object is an important feature for an image. Because of the influence of the noise the edge feature extraction is a basic and difficult problem. In order to restrain the noise of background, improve the edge accuracy of detection, this paper presents an edge feature extraction method for image by the multiscale characteristic of wavelet transform. Using the method of mutual energy cross for edge features in variety of scale and accuracy, this method can restrain noise and recognize edge. Simulations show this method is valid. [WT5HZ]
出处
《光学精密工程》
EI
CAS
CSCD
2000年第4期325-327,共3页
Optics and Precision Engineering
关键词
图像边缘特征
小波变换
多尺度分析
特征提取
image
edge feature
wavelet transform
multiscale analysis
mutual energy cross
