Automatic Image Measurement Technology in Endoscopy NDT

Abstract:

Endoscopy is an important non-destructive testing (NDT) method. It plays an important role in damage assessment and fault detection for aeroengine and structure [1,3]. However, the endoscopy usually relies on the operator’s experience to measure the defects manually in the application. Therefore, the current endoscopy method has some shortcomings that the accuracy and the efficiency of its measurement are low. To solve these problems, an automatic image measurement method based on the cubic spline interpolation is proposed in this paper. Firstly, the endoscopy image pre-processing such as gray processing, filtering, sharpening, segmentation and edge extraction are presented. Then, the pixel points in the processed edge image are handled by cubic spline interpolation and the spline function is got. After the starting point, vertex and endpoint of the defect are determined automatically, the width, depth and distance of defects are worked out. Finally, the method is applied in the fault detection for aeroengine’s blade. The experimental results show that this method can detect the width, depth and distance of the aeroengine’s blade defects exactly and quickly.

• Periodical: Key Engineering Materials (Volumes 413-414)
• Pages: 803-809
• DOI: https://doi.org/10.4028/www.scientific.net/KEM.413-414.803
• Online since: June 2009
• Authors: Guan Jun Liu, Yong Zhang, Jing Qiu, Ke Hong Lu
• Keywords: Aeroengine, Automatic Measurement, Endoscopy, Fault Detection, Image Processing

References

[1] Kim J.H.: Key Engineering Materials, Vol. 183(2000), p.1029.

[2] Gonzalez, Woods: Digital Image Processing (USA: Prentice Hall Press, 2008).

[3] Ashby J. M.: IEEE Aerospace Conference Proceedings, (2000), p.6211.

[4] Coimbra, M. T.: IEEE Transactions on Circuits and Systems for Video Technology, Vol. 23(2006), p.628.

[5] Suthaharan S.: Pattern Recognition Letters, Vol. 21(2000), p.549 Fig. 8. The cubic spline interpolation and its 1st derivative of two gaps blade edge.