A New Look: How Machine Vision Is Changing Textile Inspection
With the value of global textile exports estimated at $238 billion in 2016 by the World Trade Organization, the textile industry is one of the largest and most competitive in the world. In 2017, Asia Pacific was the largest region in the textile manufacturing market, accounting for 68% of the market share, North America accounted for 9% and Eastern Europe for 3%.
The competitiveness of the textile industry makes quality control of prime importance. In an industry where buyers judge manufacturers on their ability to consistently deliver high quality textiles at low costs, manufacturers must institute quality control processes right from the start of production until the very end to ensure that the final products are of high quality. Manufacturers must ensure that the best raw materials are sourced, the fabric construction process is as flawless as possible, and that the finished textile has minimal defects. These considerations are key for manufacturers seeking to maintain their share of this competitive industry.
A textile is a material made from either natural fibers— like cotton, wool, silk and hemp— or synthetic fibers— like nylon, polyester, spandex and rayon. Inspecting textiles during the production process ensures that the finished goods delivered by manufacturers are free from structural and surface defects, as this can help keep costs down by 45-65%. However, given that the production processes for textiles vary depending on the intended end use— which may include apparel, insulation, automotive interiors, and home decoration— the risk for defects to be introduced onto a given textile product are significant. For example, visible or invisible defects may be introduced onto a textile through the selection of poor quality raw materials or during the production process. For yarns, defects may occur during the knitting, dyeing or finishing phases. These defects are undesirable because they detract from the aesthetic appeal and performance of the textile, increase wastage, and lead to customer dissatisfaction.
Machine vision systems typically involve industrial cameras. With the help of image processing software, these systems are able to assess the quality of work and provide feedback to guide production decisions. Machine vision is particularly useful for the inspection of materials moving at high speeds of up to 120m/minute across a production line. A system’s ability to work around the clock, and maintain a high level of consistency and quality helps manufacturers improve their bottom line.
Line scan cameras are increasingly used as part of these systems to detect defects during textile production. Line scan cameras use a single line of pixels to construct a continuous two-dimensional image as a web of material moves past. Line scan cameras are particularly desirable for their ability to work with continuous webs of material and consistently detect pattern changes, changes in color and texture, and other defects on textiles moving at predictable rates across a production line.
A main advantage of line scan cameras is their ability to deliver smear-free images at high speeds, with greater processing efficiency and lower cost per pixel, compared to traditional area cameras. Teledyne DALSA offers a series of high resolution, high sensitivity line scan cameras capable of capturing high quality images to assist in defect detection.
Textile manufacturers typically capture images using one or more line scan cameras in a single row, with a narrow, single light line. Newer multi-line line scan cameras can be combined with different LED light sources to detect a range of defects across the full width and length of a textile moving along a production line. It is desirable to have the illumination over the field of view be fairly uniform and of high-intensity. While most modern line scan cameras can correct for some non-uniformity in the light, the high-intensity is necessary because of the shorter integration times for line scan imaging. Shorter integration times allow objects to move quickly in front of a camera without being affected by motion blur. The data generated by the line scan camera may be used to create two-dimensional images or automatically create a map that shows exactly where the defects are located on the surface of the textile. A quality control inspector then reviews the defect map to ensure its validity. Some of the typical defects quality control inspectors look for include water damage, misprints, foreign fibers, oil spots, among others. Images processing software then analyzes the images or defect map to construct a virtual cutting plan for an inspected textile. This process allows the manufacturer to virtually construct a cutting plan that will produce the greatest yield with minimal defects, before cutting the physical textile. After an ideal cutting plan is generated, the manufacturer can then implement the plan and prepare the textile for shipment.
Textile Waste Is So Last Season
It is estimated that around 10-20% of textile waste is from the production process. This wasted material is referred to as pre-consumer or post-industrial waste. Pre-consumer waste includes anything left over from textile production like the leftovers from a cutting plan, the ends of rolls, and rejected fabrics. To minimize wastage, one of the most widely used grading systems in the textile industry is the 4-point system. With this system, fabrics are assigned points based on the length or width of the defect.
The total points are calculated per a 100 square yards of fabric, and based on a company’s criteria, the fabric is rolled and graded. Generally, if a 100 square yard of fabric has 40 points or less, it is considered to be of good quality.
For textiles that fall below the threshold, more often than not, they are sent to landfills or incinerated. However, some companies repurpose this waste into materials for industrial use, including car seat and mattress stuffing, furniture, paper, among others. Better management of waste material is ideal for manufacturers because it increases profitability, reduces the purchase of raw materials, and contributes to preserving our environment.
Increasingly, clothing designers are starting to regard industrial textile waste as a useful resource that can be used to produce new clothes. Designers collect discarded pieces from industrial manufacturers and combine them to form new clothes. However, a major difficulty with repurposing pre-consumer waste is having to deal with very diverse pieces that may differ in color, size, weight and texture. This makes scaling production increasingly difficult, as pieces have to be carefully selected before use. But for small clothing designers who want to create unique pieces with low cost materials, using pre-consumer waste is an ideal choice.
Machine vision systems allow manufacturers to produce high quality textiles while minimizing cost and maximizing profits. By detecting defects before textiles reach buyers, machine vision systems are enabling manufacturers to supply virtually defect-free goods, minimize waste, and promote a sustainable environment.
