Recognition of Some Texture Faults in Textiles Via Computer Vision
The purpose of this work is to detect, localize, and name the texture faults in the plainly woven cotton material by the use of computer vision. The main faults that will be recognized are snarls, missing wefts or slacks, and two or three extra wefts. Reference pictures are taken from the texture faults, digitized in 16 gray levels, and stored in the computer. Each reference picture is represented as a vector, and also orthogonaIized by using Gram-Schmidt method in the vector space.
A cotton material which has one or more of the texture faults is taken afterwards, simulated and stored in the computer. The center pixel of the faulty area is determined by blurring the picture then by finding the extreme dark or extreme light pixels. A piece of picture suitable with the dimension of the reference pictures (of texture faults) is taken around the center pixel of the faulty area. A vector is formed from this chosen piece of picture, and then the distances with the reference picture vectors are calculated. The reference picture that yields the smallest minimum distance determines the type of texture fault.
KeywordsCotton Material Fault Center Texture Pattern Reference Picture Average Gray Level
Unable to display preview. Download preview PDF.
- 1.Miller, J.G. and Townsend, M.W.: Testing and Control in the Wool Industry, Vol.3, Wool Industries Research Association, Leeds, England, 1955, 222–231.Google Scholar
- 2.Ballardand, B.H and Brown, C.M.: Computer Vision, Englewood Cliffs, NJ, Prentice Hall, 1982, 166–173.Google Scholar
- 3.Adar, N. and Barkana, Atila: Data Processing by Using Associative Memory Model, Anadolu University, M.S.Thesis, 1988.Google Scholar
- 4.Celik, O., Adar, N., Barjkana, Atalay: Recognition of some Texture Faults via Computer Vision, Anadolu University, Diploma Thesis, 1988.Google Scholar