板坯边角裂纹的小波检测方法

Detection the Transverse Corner Cracks in Steel Plates Using Wavelet

提出一种通过小波变换检测板坯边角裂纹的方法。该方法针对边角裂纹图像的特点,利用小波对其进行多尺度分析,在检测出边缘线的同时将图像分解为4个方向的细节分量;再选择恰当的阈值对垂直分量图进行分割,得到最终的检测结果。实验表明:与其他两种常用检测方法相比,该方法能更有效地提取出板坯边角裂纹。

An algorithm for the detection of transverse corner cracks in steel plate surface is proposed by wavelet transform. According to the characteristics of the images in transverse corner cracks, the wavelet transform is employed to perform the multi-scale analysis to detect the edge images, which is disintegrated into four directions at the same time. A proper threshold value is choosen to segment the image of vertical component to get final results. Experiments show that compare with other two common methods, transverse corner cracks in steel plates can be extracted by the proposed method more effectively.