Research Article
Fast Feature Extraction Method for Faults Detection System
@INPROCEEDINGS{10.1007/978-3-319-73317-3_35, author={Hongmin Wang and Xiaohui Zhu and Xiangyong Niu and Ping Xue}, title={Fast Feature Extraction Method for Faults Detection System}, proceedings={Advanced Hybrid Information Processing. First International Conference, ADHIP 2017, Harbin, China, July 17--18, 2017, Proceedings}, proceedings_a={ADHIP}, year={2018}, month={2}, keywords={Image processing Boundary tracking Freeman chain code Balanced binary search tree}, doi={10.1007/978-3-319-73317-3_35} }- Hongmin Wang
Xiaohui Zhu
Xiangyong Niu
Ping Xue
Year: 2018
Fast Feature Extraction Method for Faults Detection System
ADHIP
Springer
DOI: 10.1007/978-3-319-73317-3_35
Abstract
The feature extraction based on machine learning is significant in the detection system. The boundary information, the circumference and the area are the essential features in the identification and the classification of flaws. In order to get those information, this paper proposed a novel algorithm to get the boundary information using the boundary tracking, and to make each flaw independent by establishing a balanced binary search tree for data storage. By scanning the image and the image boundaries based on binarization transformation, there is no need to fill the region, nor need to use the chain code to count the number of regions and the boundary information. According to the established balanced binary search tree, we can calculate the number of the pixel of the area of each fault, the edge information of the boundary, and the circumference. The algorithm has the advantages of fast speed, less computation, better noise suppression and accurate results.
