Thermography pattern analysis and separation
Thermography pattern analysis and separation
复制标题
DOI：
10.1063/1.4884644
复制
发表时间：
2014-06
影响因子：
4
通讯作者：
B. Gao;L. Bai;W. L. Woo;G. Tian
中科院分区：
文献类型：
--
作者：
B. Gao;L. Bai;W. L. Woo;G. Tian
热成像空间瞬态模式的分析具有相当大的潜力，使自动识别和量化的缺陷，无损检测和评估。这封信提出了一个非负模式分离模型的涡流脉冲热成像自动提取重要的空间和时间模式，根据瞬态热序列没有任何预先训练或先验知识。特别地，该方法是尺度不变的，使得可以提取具有热对比度的动态范围的表面发射率、热点和冷区域的大的差异。最后，对一个表面有光亮黑色条纹的钢试件上的人工槽进行了测试，验证了所提出的方法的有效性。
Analysis of thermography spatial-transient patterns has considerable potential to enable automatic identification and quantification of defects in non-destructive testing and evaluation. This Letter proposes a non-negative pattern separation model for eddy current pulsed thermography to automatically extract important spatial and time patterns according to the transient thermal sequences without any pre-training or prior knowledge. In particular, the method is scale-invariant, such that large differences in surface emissivity, hot spots, and cool areas with dynamic range of thermal contrast can be extracted. Finally, an artificial slot in a steel sample with shining, black strip on the surface is tested to validate the proposed method.