炭素制品X射线图像的小波边缘提取

针对经典的边缘检测算法在炭素制品X射线有噪图像的边缘检测效果问题,提出了基于小波变换的多尺度局部模极值边缘检测算法.该方法充分利用了小波变换优越的时频局部化分析能力及图像边缘点和噪声点的小波变换局部模值和方向在不同尺度下呈现不同的特性的特点.首先对图像进行小波增强处理,改善了图像的质量,在此基础上,确定了检测X射线图像边缘的最优边缘检测滤波器(小波基)和尺度,给出了小波多尺度局部模极大值的算法,并将该算法与经典的边缘检测算子进行了比较.实验结果表明,该方法明显比传统检测算子的效果要好,为进一步的缺陷模式识别等高层处理奠定了良好的基础.图3,参15.

Wavelet edge extraction on X-ray images of carbon produce

Regarding the effectiveness problems of classical edge detection algorithm on X-ray noise image of carbon material,an multi-scale local maxima edge detection algorithm based on wavelet transforms was advanced.The method made the best of the advantage of time-frequency localization of wavelet transforms,the characteristics that local maxima,and direction of edge points and noise points take on in different scale.Firstly,wavelet enhancement was made on the image by the sake of improve the image quality,based on this,the optimal edge detection filter(wavelet base) and scale were confirmed,wavelet multi-scale local maxima algorithm was described,then,the algorithm and classical operator was compared.The result of experiment shows that the method excel classical operator,it establish well foundation for pattern recognition.3figs.,15refs.