Textile-inspection system measures colors accurately

Rotary screen printing is a complex, high-speed method that applies up to 20 separate colors to a continuous textile web. In operation, the fabric web traverses the printing system at speeds in excess of 70 yards per minute and requires color-to-color registration accuracy of approximately 20 mils.

Oct 1st, 1999

Textile-inspection system measures colors accurately

Andrew Wilson

Rotary screen printing is a complex, high-speed method that applies up to 20 separate colors to a continuous textile web. In operation, the fabric web traverses the printing system at speeds in excess of 70 yards per minute and requires color-to-color registration accuracy of approximately 20 mils.

To automate the inspection of this process, the Image Science and Machine Vision Group (ISMVG) of Oak Ridge National Laboratory (Oak Ridge, TN) has teamed with Sandia National Laboratories (Albuquerque, NM) to develop an on-line inspection system designed to characterize this rotary screen-print process. Designed for multiple web-inspection systems, the system was integrated into the manufacturing process using an inspection frame designed and built for the three sites where the system was deployed.

"With any on-line inspection system, the stability of the illumination is critical for maintaining image parameters such as uniformity across the field of view and uniformity of intensity over time," says Ken Tobin, the group leader for the ISMVG. Because of this, Tobin and his colleagues chose to base the illumination subsystem around a high-frequency lamp controller and four aperture-fluorescent tubes from Mercron (Richardson, TX). "The aperture-fluorescent tubes were chosen and fixured to deliver a maximum amount of light to the fabric surface with a minimum number of elements," says Tobin. These tubes operate similarly to a standard fluorescent tube except that the tube has a nonphosphored slit that runs along its length. Because the Mercron controller drives the tubes at 50 kHz, the imaging camera can integrate over several lamp cycles eliminating banding or flicker in resultant images.

To image the fabric, Tobin chose the CL-G1 three-color linescan camera from Dalsa (Waterloo, Ontario, Canada) for the imaging front-end. "Because of the nature of the continuous-web manufacturing process, a linescan camera was needed that could independently capture red, green, and blue channels of image data," Tobin says. Dalsa`s CL-G1 uses a Kodak silicon linescan sensor composed of three independent 2096-pixel element lines separated from each other by an eight-pixel distance. Each linescan element has a red, green, and blue thin-film filter deposited on the silicon surface to generate the camera`s tristimulus response.

To synchronize the image-acquisition system to the moving web, the system uses a rotary incremental encoder from Eaton Corp. (Cleveland, OH). Mounted on the inspection frame, the encoder rolls along the edge of the vinyl mat that transports the fabric past the rotary screens. With a resolution of 150 pulses/in., halved to 75 pulses/in., the encoder is synchronized with the image sampling frequency of 75 pixels/in.

To restart image-data acquisition at the beginning of each repeated pattern, once per rotary screen revolution, a mark-sensor is used as a reset timing signal. For portability, an optical photodiode device called a Scan-O-Matic mounted to the print range views the rotating collar of the nearest print screen.

After image data are detected, 24-bit color RGB data generated by the color linescan camera are clocked into a MaxVideo MV200 pipelined image processor from Datacube (Danvers, MA) for further processing. While Terry Stalker at Sandia designed the image-processing algorithms to analyze the quality of the printed pattern, Tobin and his colleagues developed a semi-independent imaging tristimulus colorimeter system to analyze the quality of the color of the printed textiles.

"High-speed production of textiles with complicated printed patterns presents a difficult problem for a colorimetric measurement system," says Tobin. "Image-based color sensors used for on-line measurement are not colorimetric by nature and require a nonlinear transformation of the component colors based on the spectral properties of the incident illumination, imaging sensor, and textile color," he says.

Consequently, the ISMVG textile-inspection system implements an algorithm for mapping component color measurements to standard tristimulus values. The algorithm also incorporates structural and color-based segmentation for improved precision and accuracy. "Determining the constraints of the true tristimulus values based on measured imperfect values," says Tobin, "allows accurate color measurements to take place."

More in Factory