Title :
Characterisation of welded area by learning and image processing
Author :
Mitéran, J. ; Gorria, P. ; Laggoune, H. ; Paindavoine, M.
Author_Institution :
Lab. GERE, Bourgogne Univ., Le Cresout, France
Abstract :
The study that we have done concerns the realisation of a hardware and software system which allows the geometric characterisation of welded area by image processing. We propose here a method for an automatic determination of the parameters of the welded area, which avoids macrophotography and human intervention: this increases the precision of the measure. The main problem (segmentation of textured images) is solved using a supervised method of classification, known as “stress polytops”, developed by Laboratory GERE. The ease of use of the software allows the acquisition, the segmentation and the complete analysis of an image in less than five minutes on a PC486 computer
Keywords :
automatic optical inspection; image segmentation; image texture; learning systems; welding; 5 min; image acquisition; image analysis; image processing; learning; stress polytops; supervised classification; textured image segmentation; welded area geometric characterisation; Area measurement; Hardware; Humans; Image analysis; Image processing; Image segmentation; Laboratories; Software systems; Stress; Welding;
Conference_Titel :
Industrial Electronics, Control and Instrumentation, 1994. IECON '94., 20th International Conference on
Conference_Location :
Bologna
Print_ISBN :
0-7803-1328-3
DOI :
10.1109/IECON.1994.397976
