TY - JOUR
T1 - The High-speed Fabric Defect Detection Algorithm Based on the Image Layered Model
AU - Pengfei Li, Yang Jiao, Junfeng Jing & Jiangnan Li
JO - Journal of Fiber Bioengineering and Informatics
VL - 2
SP - 161
EP - 173
PY - 2013
DA - 2013/06
SN - 6
DO - http://doi.org/10.3993/jfbi06201305
UR - https://global-sci.org/intro/article_detail/jfbi/4831.html
KW - Fabric
KW - Defect Detection
KW - Image Layered Model
KW - Otsu Method
AB - The high-speed fabric defect detection algorithm based on fabric image layered model is proposed to
achieve the goal of accurate defect detection in the fabric production process. The image layered model
assumes that fabric image is a superposition of the periodic texture background image, noise image,
and defect image. Thus fabrics can be divided and conquered. Firstly, image preprocessing and mean
sampling algorithms were used to suppress the background texture and interference image layer, and
variances sampling was used for enhancing defect image layer. Secondly, the Otsu method was applied
for determining the segmentation threshold to segment the defect image automatically, then clear and
accurate defect image was obtained via image post-treatment algorithm. Finally, defect positions were
marked by a labeling algorithm to prepare for subsequent offine processing. Experiments on common
defect images from a standard defect library were described, and experimental results show that the
proposed algorithm based on image layered model is reliable, accurate, real-time and well used in the
industrial field.