Abstract | ||
---|---|---|
This work proposes a framework for automated detection of blistering defects on metal surface. The framework takes an image as input, converts it to Histogram of Oriented Gradient based representation to capture contour information. Next, it performs search for the nearest neighbour in a database of existing example images. Label of the nearest neighbour is taken to be label (or result of analysis) of the query image. While being simple, the method has an advantage of being effective and efficient. It is demonstrated through experiments that contour is very helpful in detection of blistering defects on metal surface. Besides, not requiring significant resources for pre-processing, it is closer to real time processing and hence makes it possible for deployment to inspection robots. The method has also demonstrated state of the art performance on a challenging dataset for metal surface created by experienced engineers in industry. |
Year | DOI | Venue |
---|---|---|
2017 | 10.1109/COASE.2017.8256078 | 2017 13th IEEE Conference on Automation Science and Engineering (CASE) |
Keywords | Field | DocType |
automated vision based detection,automated detection,blistering defects,contour information,nearest neighbour,query image,inspection robots,gradient based representation | Histogram,Computer vision,Nearest neighbour,Software deployment,Computer science,Vision based,Prediction algorithms,Artificial intelligence,Robot | Conference |
ISSN | ISBN | Citations |
2161-8070 | 978-1-5090-6782-4 | 0 |
PageRank | References | Authors |
0.34 | 0 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Wei-Chian Tan | 1 | 0 | 1.35 |
Phoi Chin Goh | 2 | 0 | 0.34 |
Albert J. Causo | 3 | 38 | 3.60 |
I-Ming Chen | 4 | 567 | 87.28 |
Hoon Kiang Tan | 5 | 0 | 0.68 |