Objective
One important aspect in the production of textile fabrics is the quality of end-products, which will determine price, customer satisfaction, and applications possibilities. They also suffer from the same problem: human subjectivity. A human judgement is influenced by eyesight, concentration and personnel preferences. This is certainly not an objective and cost effective method. Today, defects detection in the textile printing process is carried out in three different stages: first in the act; second when surveyed and finally before the fabrics became a ready-made article. The machine operator address to detect defects in the printing table is more or less between 10 and 30% at a machine speed of 30m/min. Human defects detection can be influenced by many factors, as the pattern complexity and pattern type (there are some patterns that creates optical illusions, e.g. stripes patterns), number of colours to print, type of tissue to print, etc. As the machine speed increases, the ability of the operator to detect defects decreases. According to textile machine manufacturers, today's machines speed can vary between 6-80m/min. However, currently the maximum value (80m/min) is rarely achieved. Normally, it can go from 30 to 60m/min. On the other hand, the flaws are often so subtle or sporadic that detecting them by human eye is almost impossible. Nevertheless. a blemish can be the difference between first and second quality. If the flows work through their entire textile system, eventually reaching the marketplace, unnecessary cost mount. Retailers lose through returns and mark-downs, and when retailers move to recoup the losses, consumers may see the price of products higher. Although, image processing in the textile industry is still in its infancy, in the past few years have seen an increasing desire by the textile market, where companies who do not manage to meet the customers demands quickly enough, are rapidly going out of business. The textile printing process requires today setting up of real-time on-line fabric inspection. The use of machine vision (that involves a computer image processing and a pattern recognition) to inspect materials produced in the form of a continuous web or strip is increasing rapidly. The range of applications is very wide, and includes metal strip, wood, textile fabrics and photographic film. Vision machines can see anything that the human eye can see; being faster than a human operative they are more cost effective. They are also more efficient since they are not subject to boredom and fatigue.' The aim of this project is to build a prototype capable of detect and classify defects in real-time "on-line" inspection during the textile printing process and to implement a prototype within a textile printing plant. The main advantage of the system proposed in this project is to allow an automatic and objective inspection without human interference. This advantage became even more important when we talk about 100% product quality. Besides enabling a 100% inspection, the system will permit de recuction of nonconforming product and therefore a substantial increase of textile product quality. L, Norton-Wayne: M. Bradshow and C.Sanby: "Machine Vision for the automated inspection of web materials"
Fields of science
- engineering and technologymaterials engineeringcolors
- natural sciencescomputer and information sciencesartificial intelligencecomputer vision
- engineering and technologymaterials engineeringtextiles
- natural sciencescomputer and information sciencesartificial intelligencepattern recognition
- agricultural sciencesagriculture, forestry, and fisheriesforestry
Call for proposal
Data not availableFunding Scheme
CRS - Cooperative research contractsCoordinator
4765 Vila Nova de Famalicão
Portugal