Open AccessTire Defect Detection Using Adaptive Dictionary
Author(s) -
Yuanyuan Xiang,
Shengbao Li,
Chuanming Yin
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1982/1/012058
Subject(s) - artificial intelligence , computer science , residual , pattern recognition (psychology) , principal component analysis , image (mathematics) , wavelet , computer vision , texture (cosmology) , standard test image , orientation (vector space) , image processing , mathematics , algorithm , geometry
This paper proposes a new algorithm for detecting tire defects, which is based on principal component analysis and adaptive dictionary learning. Because the defects on the side of the tire are very small and difficult to detect, the dictionary is directly obtained from the test image itself instead of the reference image, thereby improving the soul of adapting to changes in lighting intensity and texture. When using the learned dictionary to represent the test image, patches involving abnormalities in the test image may have larger reconstruction errors than normal errors. Through threshold operation, the defect area can be segmented from the residual image. Compared with the wavelet-based method and the component decomposition method, the experimental results of test images with defects show that the algorithm can adapt to changing light intensity and texture, and also shows more accurate defect detection results.