In recent years, with the increasing population, the acceleration of the pace of life and the iterative update of daily necessities, the output of domestic waste has increased rapidly, and the composition of garbage has become more complex and diversified. The emergence of this situation increases the complexity and difficulty of garbage classification and **. The existing various methods for the detection and classification of garbage have the problems of long detection time, low classification efficiency, and the inability to quickly sort a variety of garbage at the same time.
Among the many different types of garbage, the classification of waste plastics is an important and critical part of garbage classification at this stage. Plastics can be found everywhere in our lives, and most of the waste plastics at this stage are not being ** themselves, but ending their lives in landfills or incinerators. There is also a small percentage of plastics that are valuable to the industry, but the cost and time to separate these plastics are high. At present, most of the domestic plastic sorting is done manually, and there are also problems such as long detection time, low classification efficiency, and a variety of plastics cannot be quickly sorted at the same time.
How to sort plastics quickly and efficiently
Waste plastics can usually be labeled for macroscopic identification, but when the sorted plastic is small or the label is wrong, the sorting effect will be poor. In these cases, it is incorrect or infeasible to classify manually, by means of color visualization.
Fig.1 Materials of various plastic products.
How to sort waste plastics safely, quickly and efficiently has become one of the important topics in the renewable resources industry.
Hyperspectral technology classifies plastic products by detecting their chemical composition, and has been applied in the industry in recent years. However, the limitations of some simple hyperspectral detection instruments also hinder the specific implementation of this technology in the plastics industry, such as:
Slow detection and low ROI.
The classification effect of black plastics is poor.
*Expensive. Sorting plastics using a hyperspectral camera
As a brand focusing on hyperspectral imaging technology, SPECIM, a subsidiary of Konica Minolta Group, is one of the global leaders in hyperspectral imaging technology, and is also a hyperspectral camera manufacturer dedicated to covering the full spectral range of 400-5300nm for the industry. Especially in the plastics** industry, several hyperspectral cameras of the Specim FX series have been used in renewable resource factories to sort different types of waste plastics.
Measure, analyze, and classify mostly white or transparent samples such as PE, ABS, PVC, PS, PA, PP, PC, and PET using the FX series hyperspectral cameras.
Fig.2 Image of the plastic sample to be tested.
The data were processed using Specim Insight software to build a PLS-DA model for each sample data cube. We limit the spectral range of FX10 to 700-1000nm to eliminate any deviation in sample color.
Fig.3 Pseudo-color images (green PE, red ABS, blue PVC, yellow PS, white PA, sea green PP, cyan PC, purple PET) tested by different SPECIM hyperspectral cameras
As you can see from the figure above, the effect is better for conventional sorting.
Use FX10
We can classify most of the samples. Where the PC is usually transparent, so only the edges can be classified, and the center is incorrectly classified as the background.
Use FX17
We produced an accurate and robust model. Also since PCs are usually transparent, only the edges can be classified, while the center is incorrectly classified as the background.
Use a SWIR camera
We produced very accurate models. All samples were well classified, including those made of PC.
Use FX50
The model of the data is not as robust and accurate as that established with SWIR and FX17 cameras, but the correlation is still acceptable, and transparent PCs can also be classified.
Fig.4 Spectral reflectance curves of common plastics (PE, ABS, PVC, PS, PA, PP, PC, PET) measured by SPECIM hyperspectral camera
As we all know, black plastics are now widely used in the automotive and electronics industries, so sorting and ** are crucial. However, the use of traditional NIR industrial cameras for sorting does not give good results, because the carbon-based pigments in the black plastic absorb almost all the NIR light, making it difficult to distinguish and identify. In 2019, SPECIM developed the FX50 hyperspectral camera, which can cover tests from 2700 to 5300 nm. The FX50 was used to test the black plastic samples made of PS, ABS, PE, PA, and PC, and then the PLS-DA model was built by Specim Insight software for data classification, and the effect could reach the purpose of sorting.
Fig.5. Hyperspectral test image of black plastic SPECIM.
In summary, it can be seen that the SPECIM FX series hyperspectral cameras are suitable for a wide range of plastic classifications, regardless of their color, and can be coped with different models of cameras depending on the application requirements.
Suitable for different plastic SPECIM hyperspectral camera models (*PE, LDPE and HDPE cannot be separated, ** black plastic is applicable).
Efficient and accurate** conserve natural resources, reduce greenhouse gases and pollution, and use fossil fuels in energy production. SPECIM** solutions help the renewable resources industry better sort and sort waste plastics.
About Specim
As one of the leading companies in the hyperspectral imaging (HSI) industry, SPECIM products cover all wavelength bands from visible to thermal infrared, providing users with comprehensive hyperspectral imaging solutions to meet the diverse needs of industrial, scientific and research users. The products include industrial hyperspectral cameras, laboratory hyperspectral cameras and airborne hyperspectral cameras, which are adapted to customer-specific use scenarios. SPECIM focuses on sorter customers and has a broad customer base in the sorting**, food and pharmaceutical industries. Last year, Specim officially launched the new SpecimOne Hyperspectral Imaging Platform, which makes it easier and faster to integrate hyperspectral imaging technology with sorters. From 2021, deliveries of the Specimone platform will begin. With the Specimone platform, Konica Minolta will further expand its hyperspectral imaging business in the industrial sector.