超声波C扫描图像的缺陷模式识别

FLAW PATTERN IDENTIFICATION FROM ULTRASONIC C-SCAN IMAGE

本文在计算机图像处理理论和超声波理论的基础上,提出保边缘滤波;给出缺陷子图像分割和缺陷与背景的分离算法;再从处理后图像中提取能表征缺陷性质的几何特征和声学特征,以实现计算机对缺陷图像的理解和自动识别.

The application of computer imaging techniques on Non-Destructive Test has greatly help to understand the inspect result, and has greatly improved the reliability of NDT. In order to eliminate the deformation and blurring of ultrasonic C-Scan image, this paper presents a image filter method, which with less or no flaw-edge information lost. Aiming to the automatic scanimage recognition, a segmentation algorithm of flaw sub-image is given then. At last, the geometrical features and acoustical features of material and flaws from image are extracted and the method of flaw-patterm recognition is implemented.