Kan Wang

Dr. Kan Wang

Institute of Fluid Physics, China Academy of Engineering Physics
orcid

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
25 January 2022 in Applied Sciences
Verified
Journal Article
Open access
PDF file
KWKan Wang
Kan Wang
ZLZeren Li
Zeren Li
XWXu Wang
Xu Wang
The low accuracy of detection algorithms is one impediment in detecting ceramic tile’s surface defects online utilizing intelligent detection instead of human inspection. The purpose of this paper is to present a CNFA for resolving the obstacle. Firstly, a negative sample set is generated online by non-defective images of ceramic tiles, and a comparator based on a modified VGG16 extracts a reference image from it. Disguised rectangle boxes, including defective and non-defective, are acquired from the image to be inspected by a detector. A reference rectangle box most similar to the disguised rectangle box is extracted from the reference image. A discriminator is constituted with a modified MobileNetV3 network serving as the backbone and a metric learning loss function strengthening feature recognition, distinguishing the true and false of disguised and reference rectangle boxes. Results exhibit that the discriminator appears to have an accuracy of 98.02%, 13% more than other algorithms. Furthermore, the CNFA performs an average accuracy of 98.19%, and the consumption time of a single image extends by only 64.35 ms, which has little influence on production efficiency. It provides a theoretical and practical reference for surface defect detection of products with complex and changeable textures in industrial environments.

9 September 2021
Verified
Conference Paper
Closed access
PDF file
KWKan Wang
Kan Wang
ZLZeren Li
Zeren Li
The application of intelligent inspection to the online inspection of ceramic tile surface defects not only ensures the stability of product quality but also reduces the cost of labor for enterprises. Existing detection methods often achieve good results for the detection of defects in known background textures, however, its false detection rate increases dramatically, and its practicality is greatly reduced when there are unknown background textures or defects with similar characteristics to known background textures. This paper proposes a non-defective and defective samples synchronous comparison detection(N-DSCD) algorithm which combines traditional detection with deep convolutional neural network(DCNN). Firstly, a reference image library of non-defective ceramic tiles which contains all random backgrounds of the same batch of tiles is constructed based on DCNN extracting image features in real-time. After that, a reference image with the most similar features to the image to be detected is searched out from the image library relying on the feature comparison method. Next, we feed the image to be detected and the reference image to the defect detector at the same time, the difference in results is used as the basis of judging whether the ceramic tile has defects or not. Tested on a total of 12,000 images of ceramic tiles with 20 different textures found that: the computational time of a single ceramic tile increased by only 120 ms, and the average false detection rate of 20 unknown background texture tiles was reduced from 23.1% to 5.8%. It greatly improves the practicality of the ceramic tile surface defects online inspection system and has wider economic benefits.