Method of interval fusion with preference aggregation in brightness thresholds selection for automatic weld surface defects recognition
| Parent link: | Measurement.— .— Amsterdam: Elsevier Science Publishing Company Inc. Vol. 236.— 2024.— Article number 114969, 18 p. |
|---|---|
| Autore principale: | |
| Ente Autore: | |
| Altri autori: | |
| Riassunto: | Title screen Checking the weld joint quality is carried out during visual inspection process and depends significantly on an operator’s experience. In the article, an approach is proposed to automatically determination of geometric attributes and classification of a defective region, where the segmentation of the analyzed optical image of a weld (i.e., its division into defective and defect-free regions) is carried out using a combination of two procedures: region growing and edge detecting. The brightness thresholds for these procedures are calculated by the robust method of interval fusion with preference aggregation (IF&PA) based on the analysis of the fragmented image histograms and its gradients. The results obtained by the two procedures are consolidated to obtain a refined defective region. Testing the proposed technology on 150 real optical images showed its ability to identify geometric features and detect certain weld defects with higher accuracy compared to conventional Otsu and k-means methods Текстовый файл AM_Agreement |
| Lingua: | inglese |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | https://doi.org/10.1016/j.measurement.2024.114969 |
| Natura: | Elettronico Capitolo di libro |
| KOHA link: | https://koha.lib.tpu.ru/cgi-bin/koha/opac-detail.pl?biblionumber=673341 |
| Riassunto: | Title screen Checking the weld joint quality is carried out during visual inspection process and depends significantly on an operator’s experience. In the article, an approach is proposed to automatically determination of geometric attributes and classification of a defective region, where the segmentation of the analyzed optical image of a weld (i.e., its division into defective and defect-free regions) is carried out using a combination of two procedures: region growing and edge detecting. The brightness thresholds for these procedures are calculated by the robust method of interval fusion with preference aggregation (IF&PA) based on the analysis of the fragmented image histograms and its gradients. The results obtained by the two procedures are consolidated to obtain a refined defective region. Testing the proposed technology on 150 real optical images showed its ability to identify geometric features and detect certain weld defects with higher accuracy compared to conventional Otsu and k-means methods Текстовый файл AM_Agreement |
|---|---|
| DOI: | 10.1016/j.measurement.2024.114969 |