WInspect Textile Web Inspection Framework

Getting rid of inefficiencies in traditional industrial processes are costly in terms of time, money and consumer satisfaction. The global economic pressures have gradually led manufacturers to seek for improved quality standards of their products, in order to become more competitive, or even to maintain their original market assets. As a result, the demand for intelligent inspection systems, that ensure high quality of products over the production line, is increasing rapidly. However, automatic inspection in the textile sector is difficult and costly, and finally the types of cameras and devices that can be used for quality control.

Automation of quality control in the textile sector is possible with effective machine vision solutions. This is a quite challenging task due to many factors, among which the most critical ones are the variety of materials and the range of possible defects in these materials. This highly challenging tasks have not been achieved completely yet. Our confidence in our experience and expertise in image processing, we were willing to achieve this task with right partners. CrossTexNet 2010 call was the opportunity we were waiting for. We found our partners from Italy (Magenta srl and University of Florence), Romania (INCDTP and SC Matiro SA) and Turkey (BOSSA T.A.S.) and started WInspect Textile Web Inspection Framework (WInspect) as the coordinator of the consortium.

The major objective of WInspect is to develop an automatic inspection system based on real time image processing for quality control of textile fabric. The software part will be developed as a framework which is compatible with all “web” materials which are usually processed in the form of a continuous stream – so called “web” – of material. Capra Image Processing Platform is used as the core of the system. So, the framework is based on a pipeline structure, which uses the parallel operating principles of today’s central and graphic processors, capturing images from multiple cameras and processing these images in accordance with the quality control needs of the users.

WInspect is developed to have the necessary effectiveness in real time defect detection and classification with our image processing and machine vision expertise. Besides high defect detection performance, WInspect will have several advantages compared to existing solutions. In contrast to existing machine vision solutions on the market, key aspects that we take into consideration in design stage are the ease of adaptation to different production lines with different width and speed. It will also be compatible with different types of cameras, the GeICam standard, which will be the most commonly used standard by camera manufacturers in the near future, and different illumination systems.

This structure will eliminate the hardware-dependency. Existing automated inspection systems are compatible with only one camera model (or a limited number of models) or hardware configuration which was used in the development process. However, textile manufacturing processes are not fixed. The width and speed of the web increase in parallel with the technological developments. It is costly to adapt existing systems to these changes. Users of these systems continue using ineffective products not to burden those high costs or they pay high amounts for what they once purchased for excessively high amounts. Sector’s needs for a cross-camera, cross-platform and easily adaptable software solution will be fulfilled by WInspect.