一种针对超声检测图像的自适应阈值设置方法  

An Adaptive Threshold Setting Method for Image Processing in Ultrasonic Testing

作　　者：杨晔[1] 潘希德[1] 庄健[1] 

机构地区：[1]西安交通大学材料科学与工程学院,西安710049

出　　处：《西安交通大学学报》2015年第1期127-132,共6页Journal of Xi'an Jiaotong University

基　　金：国家自然科学基金资助项目(51375363);广东省战略性新兴产业核心技术攻关资助项目(2012A090100010);陕西省科技厅工业攻关资助项目(2013GY2-04)

摘　　要：在对图像阈值分割进行分析和已有算法的基础上,提出了一种针对超声波无损检测C扫描图像的自适应阈值设置方法,以实现对超声波C扫描图像的准确定量评估。该方法根据C扫描图像的灰度分布自动地计算相应的阈值,并进一步对特征区域进行分离和评估,从而克服固定阈值方法受超声波能量偏差影响大、难以准确评估的缺点。将该方法获得的结果与最大类间方差、最小交叉熵等方法的处理结果以及通过腐蚀试验获取的实际检测界面结果进行了对比,结果表明:该自适应阈值设置方法与其他分割方法相比,评估的结果更加准确,运算时间更短,并且结果受超声波能量偏差的影响最小。Based on the analysis of the image threshold segmenting method and the existing algorithms, an adaptive threshold setting method （ATSM） is presented to make accurate quantitative assessment on the ultrasonic C-scan images. In this method, the threshold is calculated automatically according to the gray level distribution of the C-scan image and then the feature region is segmented and evaluated by the threshold. Therefore, ATSM can overcome the shortcoming that constant threshold evaluation method （CTEM） is greatly influenced by the deviation of ultrasonic energy, which leads to inaccurate results. According to the results obtained by corrosion test, ATSM is more accurate and stable, and spends less operation time than other algorithms, such as the maximum between-cluster variance algorithm and the minimum cross-entropy algorithm.

关 键 词：超声检测图像 自适应阈值设置方法 最大类间方差 最小交叉熵 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]