Open AccessA Modified Mean Shift Method for Fault Region Extraction in Infrared Image
Author(s) -
Hui Ni,
Jin Xu,
Ruoyue Wang,
Jiangang Bi,
Feng Wang,
Xing Xu
Publication year - 2020
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 - 10.1088/1742-6596/1550/2/022010
Subject(s) - mean shift , cluster analysis , pattern recognition (psychology) , artificial intelligence , similarity (geometry) , computer science , fault (geology) , segmentation , pixel , image (mathematics) , image segmentation , k means clustering , geology , seismology
Aiming to detect the fault of electrical equipment with infrared thermography, this paper presents a modified mean shift clustering method for finding the region of fault. At the beginning, the whole image is separated by the highest threshold, as to find the coarse fault region. Then the mean shift clustering method is modified by introducing the weight factor associated with the neighboring pixels, in order to cluster the region with similarity. Meanwhile, the original way of mean shift clustering method to iterate in the image is abandoned, and we propose a threshold segmentation mechanism to solve. It thereby promotes the speed of clustering, and extracts the region of fault effectively. Finally, the experiments on real world infrared images show that our method has higher performance than some existing methods, including original mean shift clustering method.