Research

Onloom

In cooperation with:

[ Detailed english version on the project website ]

The Fabric Enhancement System is a prototype realisation of a fully automatic inspection system for defect control in textile web material. It is a joint project of the institutes ITA, WZL and LfB at RWTH Aachen University. The project was funded by the AiF gefördert.

The system is based on a traverse that guides a matrix camera over the fabric during production. The system can be mounted on any loom. Defect control takes place immediately behind the weft insertion, so that the time between a defect occurring and its detection can be reduced to a minimum. The resolution of the system is over 400 pixels/mm, so that individual threads can be found in the image and measured in terms of position, appearance and shape. Robust detection rates, defect classification and real-time processing during production are hallmarks of the system.

Real-time defect control with 400+ pixels per millimetre

The core of the system is an industrial camera that captures images with a resolution of 2456×2058 pixels at a frame rate of 15 images per second. The working distance of about 10 cm between camera and fabric allows spatial resolutions of 400+ px/mm. Plain, twill and satin weaves were examined with weft densities between 8 and 30 picks/cm. The system design places strict demands on processing time, since 5 megapixel images have to be processed in real time and only a time window of 1/15 s is available for image processing. The spatial resolution of the system is currently state of the art and is unique in the field of automatic quality control for woven textiles.

A SLIDING ACQUISITION SYSTEM with vibration damper

The traverse was designed with minimal complexity to allow easy installation on any loom. Transmitted light is used as the light source because it gives the best results in terms of thread visibility and contrast. The transmitted light is mounted on a further traverse beneath the cloth and moves synchronously with the main traverse. The traverse travels at about 1 m/s across the 1.80 m width of the loom. More than 32 images are captured during one acquisition cycle. To reduce the negative influence of machine vibrations on image acquisition, a vibration damper was designed and integrated into the system design.

Yarn Tracking – Independent of Weave Pattern

The high spatial resolution of the raw input data allows to distinguish single wefts and warps within the image. Since wefts and warps are braided among each other, some parts of each yarn are visible only from the fabric’s top side while other parts can only be seen from the bottom side. This characteristic motivates the novel concept of using fabric specific feature points as landmarks to detect yarn trajectories. Extracted feature points are structured and arranged in a yarn map, which allows to assign each feature point to a specific warp or weft. Subsequently, the trajectory of each yarn can be derived. The tracking of yarns has been successfully realised for plain, twill and satin fabrics. The framework has been further extended for pickage measurement. Within this scope, the high precision in locating single yarns is used to measure the material density on-line for loom feedback control .

THREAD MEASUREMENT with millimetre accuracy

Knowing the position of each yarn, it can be measured with regards to size, shape and texture consistency. Aberrations in yarn spacing, width, curvature and textural appearance are strong indicators for present defects. Evaluation results proved that the prototype system is able to reliably detect +98% of all occurring defects in the material with millimeter precision.