摘要
为了提高超声TOFD(Time Of Flight Diffraction)图像的配准精度,研究编写了基于离散傅里叶变换(DFT)的亚像素配准算法,使得原图与模板图的互相关度在0.99附近。为解决实际焊缝缺陷检测过程中处理数据量较大问题,文章通过图像配准来提取有效缺陷区域,减少后期图像处理的数据量和提高缺陷识别效率。现基于FFT的相关性,提出了一种超声TOFD图像配准算法,成功将灰度光学图像处理方法应用于超声波图像,提高配准精度到亚像素层次,得到的配准图像清晰,缺陷特征明显,其操作过程简单,配准位置准确。
To improve the registration accuracy of ultrasonic TOFD （Time of Flight Diffraction） image, this paper presented the sub-pixel registration algorithm based on discrete Fourier transform （ DFT）, making the cross-correlation in the vicinity of 0.99 between original and template image. To solve the large amount of data processing problems during actual weld inspection, it extracted effective defect area through image registration, so that reducing the amount of post image processing data and improving efficiency of defect recognition. Based on the correlation of FFT, it proposed a registration algorithm of ultrasonic TOFD image, the success of the optical gray-scale image processing method is applied to the ultrasound image, improved registration accuracy to sub-pixel level, registration images were clear and defect feature was obvious, and the process was simple, the registration location was accurate.
出处
《信息技术》
2015年第5期27-31,共5页
Information Technology
基金
科技部重大项目(2013YQ23057504)
