Share Email Print
cover

Proceedings Paper

Surface defects detection algorithm for a small sample of sealing rings for aerospace based on deep learning
Author(s): Hui Liu; Bo-xia He; Yong He; Hao Jiao
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

When using visual image technology to detect surface defects of sealing rings for aerospace, traditional image processing algorithms are constrained by surface defects types, scale differences, illumination variations, and non-uniformities of surface image. As a result, the application of each algorithm is relatively limited. To solve the problems, a general deep learning algorithm for surface defects detection of sealing rings is proposed. Texture features of defects, gradient features of edges and contour features are selected as main features of surface defects of sealing rings differentiated from the background, and a new cascade modular backbone network is design for extraction of these features. The average aspect ratio of label box for defects among dataset of defect images of sealing rings is calculated and aspect ratio distribution is counted to obtain the proportional parameter for generating prior box. Based on the calculated proportional parameters of prior box and scale features of defects, multi-scale defects detection network is designed. Meanwhile, defects classification and position regression compound loss function are defined. In the end, weighting parameters of the network are updated on the basis of end-to-end training. Experimental result shows that the accuracy of proposed algorithm for defects detection reach 94.03% under the condition that the original defects sample images is 107 and confidence is 0.6. Compared with RefineDet network, the depth of backbone network of feature extraction increases by 78% while the scale of parameters reduces by 84.82% and the accuracy of defects detection reduces by 3.01%.

Paper Details

Date Published: 31 January 2020
PDF: 9 pages
Proc. SPIE 11427, Second Target Recognition and Artificial Intelligence Summit Forum, 114274D (31 January 2020); doi: 10.1117/12.2553207
Show Author Affiliations
Hui Liu, Nanjing Univ. of Science and Technology (China)
Bo-xia He, Nanjing Univ. of Science and Technology (China)
Yong He, Nanjing Univ. of Science and Technology (China)
Hao Jiao, Nanjing Univ. of Science and Technology (China)


Published in SPIE Proceedings Vol. 11427:
Second Target Recognition and Artificial Intelligence Summit Forum
Tianran Wang; Tianyou Chai; Huitao Fan; Qifeng Yu, Editor(s)

© SPIE. Terms of Use
Back to Top
PREMIUM CONTENT
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?
close_icon_gray