A Neural Network based system for woollen clothes colour classification

In the textile industry recycling can be applied to a wide range of products, since the final product is often obtained using only one bulk material. Wool industry is a clear example of such a condition: pullovers, sweaters and many other clothes are made almost exclusively by treated and colored wool. In this case recycling is a straightforward approach to the supply of a bulk material very similar to the original one. The common practice for companies performing wool recycling consists of providing the customer with a list of the available colors; according to the customer’s choice, the companies start to seek its depot for properly colored clothes. The clothes are selected on the basis not only of the similarity of their color to the desired one, but also of the knowledge of the recycling process. This implies that the company operators demanded to classify the clothes, called “pickers”, are aware of the way the different colors combine into a new one. Accordingly, differently colored clothes can be grouped together in color classes, while apparently similar ones can be separated. As a result highly specialized and well-trained operators must perform the selection process.