Detecting fabric defects by image analysis

Author

Yixiang Frank Zhang

Date of Award

12-1993

Degree Type

Thesis

Degree Name

Master of Science

Major

Computer Science

Major Professor

M. Thomason

Committee Members

Jens Gregor, Randall R. Bresee

Abstract

A method of detecting defects in solid-shade, unpatterned woven fabrics is described. Two types of defects are studied: knots and slabs. The proposed automatic inspection system consists of a unix workstation, video camera, frame grabber and illumination device. Fabric samples are illuminated with transmitted light. A digitized picture from the video camera is stored in the memory of the computer. The computer calculates features from the picture and classifies samples as defective or defect-free. The locations of defects are also indicated. Autocorrelation function is used to determine the size of the repeat unit in the picture. Two approaches to detecting defects, gray level statistics and morphological operations, are presented. Fabric products from a local textile mill are examples of objects for inspection.

Recommended Citation

Zhang, Yixiang Frank, "Detecting fabric defects by image analysis. " Master's Thesis, University of Tennessee, 1993.
https://trace.tennessee.edu/utk_gradthes/12070