New! Sign up for our free email newsletter.
Science News
from research organizations

Artificial Vision Used To Improve Recycling Of Electronic Scrap Metal

February 12, 2009
Basque Research
Researchers are studying an innovative method based on multispectral artificial vision systems to enhance the value of electronic scrap which currently represent 4% of urban waste in Europe.

TECNALIA Technological Corporation is studying an innovative method based on multispectral artificial vision systems to enhance the value of electronic scrap which currently represent 4% of urban waste in Europe.

The aim of this project, known as SORMEN, is to develop a technology for the separation of scrap metal from electronic waste based on a system of multispectral vision and incorporate it into the process of a recycling plant. This new machine overcomes the limitations of current, basically manual, methods which consume a large amount of manual labour and time and which are unable to separate metals whose characteristics of colour, shape and weight are similar.

The solution proposed by TECNALIA enables the separating of elements of the same colour - such as aluminium, nickel or stainless steel -, employing the recycling of these materials to the full. It represents a highly significant advance over other techniques of separation based on colour vision and useful for other processes such as separating lead impurities, from copper for example. In the case of aluminium, for example, the system designed by TECNALIA will enable the recovery of between 30 and 40% more of this metal.

Currently Europe generates more than 6.5 million tons of electrical and electronic waste per year, of which more than 90% goes to waste dumps.

One of the current problems facing the correct recycling of electronic scrap waste is that it contains many different materials that cannot be separated with current technology. Dismantling electronic equipment requires manual tasks, making the process labour-intensive and very costly. For example, with television sets, only the cathode-ray tubes are dismantled while the rest is crushed up. With other equipment, only the larger parts made of aluminium, copper or iron are separated, while the rest may be passed on for various uses.

To make the process more economic it is highly important to have machines that can automatically identify each one of the elements, above all in a non- destructive manner. Moreover, in this way it will be less contaminating for the environment and workers will not be exposed to the emission of substances that could be damaging to health.

Investigating the possibilities of classification of different materials, such as iron, lead, stainless steel, aluminium, plastic or brass, it can be observed that, in some cases, it is possible to find a way of identifying them in the visible spectrum. Nevertheless, other metals such as aluminium or stainless steel are impossible to separate by colour.

Multispectral identification

It is necessary to find other methods and here is where multispectral identification comes in. These solutions can be based on the fact that each pure metal has a spectral reflectivity response which is unique to the element, some of the metals can be identified in the 380-740 nm visible spectrum (as in the case of lead and copper) and others outside this range.

Unlike colour cameras, multi- and hyper-spectral systems can appreciate multiple bands, from ultraviolet to infrared, with very good resolution of up to 2.5 nm between bands, for example, with the ISA camera from the Finnish company, SPECIM. This versatility makes it possible for these systems to detect, classify and identify different materials, overcoming some of the limitations of the colour cameras that operate in the visible range.

The application of this type of technology to the classification of metals is the novel approach which holds out hope for the solution to the problem. It is expected that, in 2015, although it is estimated that electrical and electronic waste generated annually will double to 12 million tons, the amount of material recycled will rise significantly (for example, in the case of aluminium, between 30 and 40% more).

Drawing up the project

Apart from Tecnalia, taking part in this project are the CSL technological centre at the University of Liège (Belgium), the recycling companies Indumental Recycling (the Basque Country) and Ige Hennemann (Germany), the recycling machine manufacturer Hevac Ambient (Barcelona), the SPECIM multispectral system designers (Finland) and the Aclima Environment Cluster (the Basque Country).

The current processes of the recycling companies have a number of samples with mixtures of different components (aluminium, copper, brass, steel, plastics, and so on) for which it is desirable to be able to separate their distinct components. Initial analyses of these samples were carried out by the designers of the CSL illumination system, by means of point to point analyses, and by SPECIM using analysis with multispectral cameras

For the design of the prototype machine, including the vision system, the targets of the recycling companies have been taken into account, with the aim of that the machine will be in the end also competitive at the level of performance, functionality and price.

The recycling companies, members of the consortium (Indumental Recycling and Ige Hennemann) have thus defined their main requirements for the first stage of the project:

  • The initial quantity they wish to identify by unit of time (the entry of the mix with all the components)
  • The components of each fraction and their proportion in the mix
  • The size, colour, shape and other physical characteristics of each component
  • The components that have to be extracted
  • The minimum size to be identified
  • The minimum and maximum size of components to be separated
  • The height of each component
  • Notable problems of certain materials
  • The average particle weight of each component, or the shape thereof, in order to estimate how many particles could pass through if a flow of 1 Tn/h is desired.

The designed system should also be insulated from water and dust because it is to be used in the recycling industry, where all the machinery has to be strongly protected from great amounts of dust as well as withstand temperatures ranging from - 5ºC to 35ºC.

Likewise, the machine must be able to work 24 hours a day, five days a week, have little maintenance and provide the possibility of using it at low speed in order to obtain material as surplus in a preliminary cycle.

The initial trials in the project were the most important to identify the particles required to identify each mix, in accordance with the interests of the recycling companies. The samples analysed followed different strategic lines: SPECIM with a hyper-spectral line-by-line camera and CSL, with point-to-point spectrometric analysis. ROBOTIKER, given the great similarity between and choice of different classes, has employed processing techniques that enable the extraction of discriminating characteristics, simultaneously taking into account the spectral and spatial characteristics. In this way, discrimination of materials is achieved with a performance superior to the discrimination obtained by both methods operated separately.

The system of illumination proposed by CSL is a new design the aim of which is to achieve uniform illumination without brilliances, and with sufficient power to illuminate in the appropriate spectrum, taking into account the initial trials carried out with the mixes supplied by recycling companies.

Story Source:

Materials provided by Basque Research. Note: Content may be edited for style and length.

Cite This Page:

Basque Research. "Artificial Vision Used To Improve Recycling Of Electronic Scrap Metal." ScienceDaily. ScienceDaily, 12 February 2009. <>.
Basque Research. (2009, February 12). Artificial Vision Used To Improve Recycling Of Electronic Scrap Metal. ScienceDaily. Retrieved April 24, 2024 from
Basque Research. "Artificial Vision Used To Improve Recycling Of Electronic Scrap Metal." ScienceDaily. (accessed April 24, 2024).

Explore More

from ScienceDaily