基于时序信息的红外图像缺陷信息提取

主动红外热像检测技术中，红外图像的缺陷信息提取是其核心内容。传统的红外图像处理方法在一定程度上可以消除噪声、提高图像的对比度，但是仍存在一些问题，如:需要手动选择特征信息丰富的红外图像，红外图像增强和图像分割过程中会引入主观成分，仅仅分析单张红外图像可能存在信息丢失等问题。针对上述问题，本文根据主动红外热成像的数据特征提出了一种基于时序信息的红外图像缺陷信息提取方法。首先，通过室内实验制作含缺陷分层的混凝土试块；然后，利用主动红外热像检测技术进行三维红外图像数据的采集，提取每个像素点的时序信息；最后，采用基于时序信息的K-means方法进行缺陷特征提取。结果表明，基于时序信息的缺陷提取方法是可行的，其可以提取到隐藏的分层缺陷信息，提取效果优于基于空域信息的K-means方法。

Infrared Image Defect Information Extraction Based on Temporal Information

In active infrared thermography technology, the extraction of defect information from infrared images is crucial. Traditional image processing methods can eliminate noise and improve image contrast, but several challenges remain, such as selecting the infrared image manually, subjectivity in the process of infrared image enhancement and segmentation, and information loss in the process of a single infrared image. To overcome these challenges, this study proposes a method for extracting defect information from infrared images based on time sequence information. First, concrete blocks with delamination are fabricated by indoor experiments. Then, active infrared thermal image detection technology is used to collect the infrared image data and temporal information is extracted for each pixel. Finally, the K-means method is used for defect feature extraction based on temporal information. The results show that the defect extraction method based on temporal information can extract hidden defect information. Furthermore, its hierarchical defect information extraction effect is better than that of the K-means method based on the spatial domain.