应用于压力容器红外无损检测的SIFT图像拼接方法  

SIFT image stitching method applied to infrared nondestructive testingof pressure vessels

作　　者：杨晓 黎琼炜 殷春[1] 陈凯[1] 邱根[1] 薛婷 黄雪刚[3] Yang Xiao;Li Qiongwei;Yin Chun;Chen Kai;Qiu Geng;Xue Ting;Huang Xueguang(School of Automation,Engineering,University of Electronic Science and Technology,Chengdu 611731,China;Air Force Research Institute,Beijing 100080,China;Hypervelocity Aerodynamics Institute,China Aerodynamics Research&Development Center,Mianyang 621000,China)

机构地区：[1]电子科技大学自动化工程学院,成都611731 [2]空军研究院航空兵研究所,北京100080 [3]中国空气动力研究与发展中心超高速空气动力研究所,绵阳621000

出　　处：《电子测量与仪器学报》2019年第12期31-40,共10页Journal of Electronic Measurement and Instrumentation

基　　金：自然科学基金(61873305,61671109,U1830207);四川省科技厅项目(2018JY0410,2019YJ0199)资助。

摘　　要：压力容器在航空航天以及工业中都有着重要的作用其安全性是十分重要的,由于压力容器使用环境复杂作用性强,会产生如疲劳裂纹腐蚀的损伤,并且其不可分割移动的特殊性需要对其使用基于红外热传导的无损检测方法,对压力容器升温后降温过程采集红外视频流对其进行信号处理获得红外重构图像。由于压力容器体积较大需要对其多次拍摄获得多幅红外重构图像,为了分析检测高压容器整体缺陷,使用了基于尺度不变特征变换(SIFT)的红外重构图像拼接方法对图像进行拼接融合,首先使用SIFT对图像进行特征点提取,然后对特征点进行匹配,使用随机一致性(RANSAC)去除误匹配点以及获得仿射变换矩阵参数,最后对图像进行亮度调整和融合后获得最终拼接图像,图片拼接结果反映了压力容器的完整缺陷,实验结果验证了方法的有效性。Pressure vessels play an important role in aerospace and industry and the safety is very important. Due to the complex and versatile environment of the pressure vessel, damage such as fatigue crack corrosion will occur, and the particularity of its indivisible movement requires the use of non-destructive testing method based on infrared heat conduction, and the signal processing is performed to obtain the infrared reconstructed image, the infrared reconstruction image is obtained by collecting the infrared video stream and processing the signal of the pressure vessel during the process of heating and cooling. Due to the large volume of the pressure vessel, multiple infrared reconstructions are needed for the multiple shots image. In order to analyze and detect the overall defect of high-pressure container, this paper uses SIFT-based infrared reconstruction image stitching method to splicing and merging images. Firstly, feature points are extracted by SIFT, then feature points are matched, and RANSAC is used to remove mismatched points. And obtain the affine transformation matrix parameters, and finally adjust the brightness and fusion of the image to obtain the final mosaic image. The image mosaic result reflects the complete defect of the pressure vessel. The experimental results verify the effectiveness of the method.

关 键 词：压力容器 图像拼接 红外重构图像 SIFT算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TN219[自动化与计算机技术—计算机科学与技术]