摘要
本文为得到正确的能被计算机自动识别的云纹干涉条纹,首先对图像作滤波、二值化、细化等处理。尽管滤波可以有效地消除图像中的噪声,但当图像中本身就存在很大的噪声,滤波也不能完全消除这些噪声的影响。因此在细化后的细化图像中局部可能会存在分叉以及两条或几条条纹在空域上连在一起。在本文中提出了对条纹跟踪进行控制的方法来达到条纹提取的目的。最后对跟踪到的曲线进行多项式拟合来消除噪声的影响。处理结果表明采取以上的对云纹干涉条纹图像进行处理的步骤所得到的图像可以很好的被计算机自动识别,并且图像中的数据更加有效、准确。
In order to get image stripe in the Moire Interferometry recognised directly by computer,the image stripe are first done by a set of image processings,such as filtering, binaryzation,thinning.Although filtering can denoise availably, filtering can not avoid completely the effect of the noise when there is a lot of noise in the picture.Therefore there may are bifurcate in the image and two stripes or more are conjoint in spatial domain.In order to get stripe,the method that takes control of trancing stripe is proposed in the text.At last,the polynomial is used to fitting a curve to data points from the tranced stripes to remove the effects of noise. The results indicate that after the image are done by a set of image processings,it can be recognized automatically by the computer,and the data from the image are more valid, More accurate.
出处
《中国西部科技》
2008年第8期14-16,共3页
Science and Technology of West China
关键词
云纹干涉图像
线性回归
多项式拟合
laser moire interferometry image： linearity regression
polynomial fitting
