The textile industry is a most important and rapidly growing
sector that makes a considerable impact to the economical state of Sri Lanka.
Quality is a major parameter in textile, thus good quality products increase
the profit of the industry as well as the customer satisfaction. If the defects
in the fabric are not detected properly significant financial loses can be
occurred. Hence, the fabric inspection process should carry out vigilantly. The
key issue, therefore, is how and under what circumstances fabric inspection
will lead to quality improvement. As a result, the automated fabric inspection
systems have been designed to improve the efficiency of the inspection process.
the fabric inspection is still undertaken manually by skilled staff with a
maximum accuracy of only 60%-70%. The modern fabric manufacturing industry
faces a lot of challenges due to high productivity as well as high quality
manufacturing environment. Because the production speeds are faster than ever
and because of the increase in roll sizes, manufacturers must be capable of
identifying defects, locating the sources of defects, and making the necessary
corrections in less time so as to lessen the amount of low quality fabric. This
in turn places a greater strain on the inspection departments of the
manufacturers. Due to the factors such as tiredness, boredom and carelessness,
the staff performance is often unreliable. Therefore, the best reliable
evaluation is through the application of an automated inspection system.
From the early
beginning, the target is to achieve optimum potential benefits such as high
quality, low cost, comfort, accuracy and speed in the manufacturing process. As
the technology is revolutionized the fabric inspection process has been
developed from manual to automated machinery to help achieve all those benefits
in the manufacturing process. The application of automated fabric inspection
would seem to offer a number of advantages, such as improved quality, reduced
labor costs, the elimination of human errors and increase the profit of the
industry. Therefore, the automated visual inspection is gaining progressive
importance in fabric manufacturing industry.
inspection system usually consists of a computer-based vision system. Because
they are computer-based, these systems do not undergo the drawbacks of human
visual inspection. The application of digital image processing is useful in
textile manufacturing and inspection. In recent years, it has proven to be the
most promising, rapid and reliable solution for the development of automated
fabric inspection systems. Considerable efforts have been taken to develop
and/or improve the task of automated fabric inspection systems. As all fabric
has the periodic regular structure, analyzing the parameters such as variance,
intensity of the fabric presents a possible way to predetermine the occurrence
of defects in the fabric.
method in this thesis represents an effective and accurate approach to
automatic defect detection. It is capable of identifying various types of
defects by monitoring the fabric structure. Presence of a defect over the
periodical structure of a fabric causes changes in the variance and intensity
parameters of the fabric. By thresholding the defected area to create a binary
image, it is possible to identify the exact location of the defect. Further,
for accurate detection of the defects the noise removal, morphological erosion
and dilation, connected component analysis are carried out.
The fabric defect
could be simply defined as a change in or on the fabric construction.
Fabric defects can occur due to
machine faults, color bleeding, yarn problems, scratch, poor finishing, dirt
spots, excessive stretching, and crack points. Because of the wide variety of
defects, it will be useful to apply the study on the most major fabric defects.
The chosen major fabric defects are: hole, missed-yarn, oil mark, knitting
fault and pin holes.
A fabric fault detection system designed with MATLAB is
used for this procedure. It is implemented on the above mentioned chosen fabric
defects as well as the defect-free samples to identify as defect or non-defect.
To verify the success of the defects it is implemented on one hundred and fifty