Abstract
Defects in fabrics usually appear as inhomogeneities of fabric texture patterns. Accurate and efficient detection of these inhomogeneities is crucial for fabric manufacturers. Many sophisticated visual inspection methods have been introduced based on texture analysis algorithms. Unfortunately, only few of these are suitable for real-time fabric inspections due to the high computational cost. An efficient and robust fabric defect detection method is proposed in this paper. It is based on the fact that the presence of defects breaks the texture's homogeneity and thereby changes the distribution of microedges in the edge map after the original texture image undergoes edge detection. In the proposed approach, a set of edge operators are introduced to reveal microedges in multiple directions. Local measures of microedge pixel densities are proposed to characterize the underlying fabric texture. A weighted distance derived from normal distribution is exploited to classify the inhomogeneous areas from normal fabrics. The proposed method is implemented on a smart camera using its on-board processing ability to achieve fast inspection. Experimental results on real fabrics demonstrate the effectiveness and robustness of the proposed method.
Original language | English |
---|---|
Pages | 360-365 |
Number of pages | 6 |
State | Published - 2005 |
Event | Proceedings of the 2005 IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2005 - Monterey, CA, United States Duration: Jul 24 2005 → Jul 28 2005 |
Conference
Conference | Proceedings of the 2005 IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2005 |
---|---|
Country/Territory | United States |
City | Monterey, CA |
Period | 7/24/05 → 7/28/05 |
ASJC Scopus subject areas
- General Engineering