Detecting a Wrinkle in (Less) Time

Aaron J. Hand

When choosing an outfit to wear, one factor often prevails over others in the decision-making process: Which item doesn't need to be ironed? In this scenario, what could be better than a "wrinkle-free" shirt?

The textile industry routinely evaluates wrinkling as an important performance criterion, particularly for fabrics treated with wrinkle-resistant solutions. Workers are given a set of standards with which to grade the material -- from severely wrinkled to very smooth -- but this visual method tends to be time-consuming and unreliable. It is particularly difficult to inspect patterned or dark fabrics, and it is not easy for human inspectors to objectively assign a grade to the samples.

Researchers at the University of Texas have developed a laser triangulation method to automate the evaluation. Detailed in the October 1999 issue of Optical Engineering, the method places a laser line generator and a charge-coupled device (CCD) camera in a triangular setup with a fabric specimen to detect wrinkles and uses a neural network to grade the degree of wrinkling. Wrinkling classification -- based on the roughness, sharpness and density of a fabric's wrinkles -- is more reliable than visual inspection and is independent of fabric color or design.

The researchers first tried detecting surface profiles with a single laser line. Their latest prototype sends more wrinkling data per image to the computer with a multiple line generator from Lasiris Inc. of St. Laurent, Quebec, Canada, which splits the light from a red laser diode into 19 lines, covering an area of about 25 x 25 cm. In addition to being bright enough to detect, the laser's wavelength must match the high sensitivity range of the camera. In this case, the researchers use 670-nm lines to match the response curve of a CCD camera from JVC of Yokohama, Japan.

A particular advantage of triangulation is that detection is independent of fabric color or patterns, said Bugao Xu, an associate professor in the university's department of human ecology. Changes in color affect only the intensity of the laser lines. Because this method converts those lines into binary data, only the pixel coordinates -- not the intensities -- are used to calculate depths. The developers also have incorporated a rotating stage into their design to compensate for wrinkle linearity. The system averages the data from eight images taken at different angles to reduce the influence of wrinkle orientations.

Two companies have contacted the researchers, but they hope to gain additional industrial interest in commercializing the method.