Image Analysis for the Detection of Barré

Barré is a major problem for the textile industry.  Barré is detectable after fabric is dyed and the detection of barré can depend upon the color of the dyed fabric, lighting conditions, fabric pattern, and/or the color perception of the person viewing the fabric.  The standard method for measuring barré is to compare fabric samples to digitally produced images on the AATCC rating scale.  The use of a human observer to detect and rate barré is problematic due to differences between operators and the need for a trained and skilled human operator.  An image analysis system was developed to examine the potential for a computer-based image analysis system to examine fabrics for the presence of barré and the extent to which barré may exist within a fabric.  Knit fabrics present some unique challenges due to their inherent 3-dimensional nature and repetitive construction.  Techniques for capturing the image for analysis by the software had to be developed in conjunction with software which was able to process the image and allow for the measurement of barré.