Detecting defects, and especially reliably measuring defect sizes, are critical objectives in automatic NDT defect detection applications. In this work, the Sentence software is proposed for the analysis of pulsed thermography and near IR images of composite materials. Furthermore, the Sentence software delivers an end-to-end, user friendly platform for engineers to perform complete manual inspections, as well as tools that allow senior engineers to develop inspection templates and profiles, reducing the requisite thermographic skill level of the operating engineer. Finally, the Sentence software can also offer complete independence of operator decisions by the fully automated "Beep on Defect" detection functionality. The end-to-end automatic inspection system includes sub-systems for defining a panel profile, generating an inspection plan, controlling a robot-arm and capturing thermographic images to detect defects. A statistical model has been built to analyze the entire image, evaluate grey-scale ranges, import sentencing criteria and automatically detect impact damage defects. A full width half maximum algorithm has been used to quantify the flaw sizes. The identified defects are imported into the sentencing engine which then sentences (automatically compares analysis results against acceptance criteria) the inspection by comparing the most significant defect or group of defects against the inspection standards.
Automatic thermographic image defect detection of composites
Thermosense: Thermal Infrared Applications XXXIII ; 2011 ; Orlando,Florida,United States
Proc. SPIE ; 8013
2011-05-06
Conference paper
Electronic Resource
English
Automatic thermographic image defect detection of composites [8013-28]
British Library Conference Proceedings | 2011
|Thermographic Detection of Defects in Refractory Composites
Online Contents | 1998
Fusion of geometric and thermographic data for automated defect detection
British Library Online Contents | 2012
|Fiber optic thermographic detection of flaws in composites
Tema Archive | 2009
|