摘要
在对电力钢管塔施工焊趾表面进行裂纹智能检测时,需要考虑天气、机械等因素对图像造成的模糊干扰,特征呈现多尺度属性,检测准确率不高,需要对特征增强。提出基于融合多尺度特征的电力钢管塔焊趾表面裂纹图像增强方法。首先,利用小波变换作为多尺度分析工具,提取电力钢管塔焊趾表面裂纹图像的多尺度特征;其次,利用贝叶斯对图像实行平滑处理;最后,通过伪暗通道计算图像实行去雾处理,检测图像显著性区域,根据其向量分差对显著性区域实行灰度映射,最终完成图像增强。试验结果表明:这种增强算法下的电力钢管塔施工焊趾表面裂纹图像SNR和MSE分别平均为7.4和1.2,对比度测量值较其他方法至少提高了3.81和1.71,处理后的图像质量高。
In the intelligent detection of cracks on the welding toe surface of power steel pipe tower construction,the fuzzy interference caused by weather,machinery and other factors on the image needs to be considered.The features showed multi-scale properties,and the detection accuracy was not high.Therefore,the features need to be enhanced.An image enhancement method based on fusion of multi-scale features for surface cracks at weld toe of power steel pipe tower was proposed.Firstly,wavelet transform was used as a multi-scale analysis tool to extract the multi-scale features of the surface crack image of the weld toe of the electric steel pipe tower.Secondly,Bayesian was used to smooth the image.Finally,the image was defogged through pseudo dark channel calculation to detect the significant area of the image,and the significant area was gray-scale mapped according to its vector difference.Finally,the image enhancement was completed.The experimental results showed that the SNR and MSE of the surface crack image of the construction toe of the power steel pipe tower under this enhancement algorithm were 7.4 and 1.2 respectively,and the contrast measurement value was at least 3.81 and 1.71 higher than that of other methods.The processed image quality was high.
作者
薛钦
卢峰
徐俊
董寒宇
柏菊红
XUE Qin;LU Feng;XU Jun;DONG Hanyu;BAI Juhong(Huzhou Electric Power Supply Company,State Grid Zhejiang Electric Power Co.,Ltd.,Huzhou 313000,Zhejiang pro.China)
出处
《焊接技术》
2022年第12期69-74,I0008,共7页
Welding Technology
关键词
电力钢管塔
图像增强算法
多尺度特征
贝叶斯
小波变换
噪声干扰
electric steel pipe tower
image enhancement algorithm
multiscale features
Bayes
wavelet transform
noise interference
