Prediction of the quality of resistance welds by computer based color image analysis
This report discusses experiments which have been completed to correlate the quality of electric resistance pinch welds with an automated computer analysis of the weld surface. The pinch welds were performed on small diameter stainless steel tubes after they were annealed in air at several different temperatures to form an oxide layer on the weld surfaces. The images of the tube bore were collected with a borescope, stored in a computer and analyzed. The analysis consisted of computing a parameter which gave a representation of the color integrated over the inspected region. This color parameter was then used to rank the tubes in order of their relative oxidation level. Once this was performed the tubes were welded and low magnification metallography was performed on the welds. It was found that the color analysis gave a perfect correlation with the oxidation levels and that the weld quality was inversely proportional to the amount of oxidation. It was also shown that the color analysis was robust in the sense that the sorting was independent of the borescope illumination level over a large range for both oxidized and unoxidized stems. Thus the color parameter chosen was an excellent predictor of the weld quality.
- Research Organization:
- Westinghouse Savannah River Co., Aiken, SC (United States)
- Sponsoring Organization:
- USDOE, Washington, DC (United States)
- DOE Contract Number:
- AC09-89SR18035
- OSTI ID:
- 10106319
- Report Number(s):
- WSRC-MS-92-051; CONF-9208178-1; ON: DE93002402
- Resource Relation:
- Conference: International Metallographic Society annual meeting,Denver, CO (United States),2-5 Aug 1992; Other Information: PBD: [1992]
- Country of Publication:
- United States
- Language:
- English
Similar Records
Fill tube bore inspection with machine vision
Fill tube bore inspection with machine vision