首页 | 本学科首页   官方微博 | 高级检索  
     检索      

基于平稳小波变换的有噪弱目标检测方法的研究
引用本文:赵丹培,王延杰,李桂菊.基于平稳小波变换的有噪弱目标检测方法的研究[J].光学技术,2005,31(3):380-383.
作者姓名:赵丹培  王延杰  李桂菊
作者单位:长春光学精密机械与物理研究所,图像室,长春,130031
摘    要:利用平稳小波变换的多尺度边缘检测算法分别对模糊图像、低对比度图像和加入噪声的图像进行了边缘检测,验证了多尺度二次B样条小波的检测效果,也比较了三种小波局部模极大值方法在抗噪性、计算量及检测效果等方面的性能,并且针对对比度低,受噪声污染严重的目标图像,提出一种能够根据不同背景计算出自适应阈值的新方法,使其在抗噪的同时又能较好地提取出微弱目标边缘。实验证明,利用多尺度二次B样条小波边缘检测算法能有效地排除噪声干扰,准确地提取出微弱边缘,可以实现3%对比度下的有噪图像的目标探测问题。

关 键 词:平稳小波变换  弱目标  局部模极大值  B样条小波  自适应阈值
文章编号:1002-1582(2005)03-0380-04
修稿时间:2004年6月27日

Study of edge detection method for noise-involved weak object based on stationary wavelet transform
ZHAO Dan-pei,WANG Yan-jie,LI Gui-ju.Study of edge detection method for noise-involved weak object based on stationary wavelet transform[J].Optical Technique,2005,31(3):380-383.
Authors:ZHAO Dan-pei  WANG Yan-jie  LI Gui-ju
Abstract:A multi_scale method to detect image edges based on local modulus maxima of stationary wavelet transform was proposed for blurred image, low contrast image and Gauss noise_added image. The multi_scale edge detection effect of quadratic B_spline wavelet is validated. Throught the simulation, the three methods of local modulus maxima on wavelet were compared for capability of detection effect, anti_noise capacity and computation quantity. The contradiction between the edge detection precision and noise restraining ability was well resolved by applying a new method of adaptive threshold according to various backgrounds. This method can not only restrain noises but also extract the edge of weak object in allusion to low contrast image seriously polluted by noise. Experiments proved that this method with quadratic B_spline wavelet can effectively eliminate noise and accurately etract weak edge, resolved the problem of object detection for noise_invovlved image whose contrast is as low as 3%.
Keywords:stationary wavelet transform  weak object  local modulus maxima  B_spline wavelet  adaptive threshold
本文献已被 CNKI 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号