Smart Sorting Based on Hyperspectral Datasets for Improved Performance 

Automated sorting technology uses sensors or cameras to gather data and distinguish between items, then a sorting mechanism to separate them. Relying on automation over manual labour offers a range of benefits.Smart sorting systems can utilise various different technologies, sensors, and algorithms to derive information about items. For example, optical sorting technology utilises simple RGB data to separate based on colour and shape. Other systems  based on X-ray or more extensive optical analyses such as infrared spectroscopy or hyperspectral imaging (HSI) provide increased differentiation for complex sorting problems.

With real-time, compact, and portable HSI cameras now available, sensing and sorting light into many more channels can enable new, more advanced smart sorting systems. This includes improving upon simple optical sorting system performance or enabling new use cases requiring more advanced optical data.

Key Components of a Smart Sorting System

Before discussing different sensing technologies and their pros and cons, let’s break down how smart sorting works. The four main components are:

1. Feed System: Often a conveyor belt or chute system that is able to move items past the sensor and through the separation process. The feed system needs to be able to evenly spread the items into a single layer so that the sensor can inspect each individual item.
2. Sensor: The technology that gathers the data to classify and sort the items. A range of possible sensors and technologies can be implemented depending on the specific attributes required to sort the items. For example, colour and shape or chemical composition.
3. Processing: Software that processes and analyses the raw data from the sensor in real-time to determine properties related to each item and classify them. Processing needs to be accurate and fast to ensure high accuracy and throughput. The level of analysis needed depends on the sensing technology and data the software has to work with.
4. Separation System: After processing, the items are sorted based on their classification. The separation process can take many different forms, requiring some sort of physical action (e.g., reject/leave, separation into different lines) to move different items. Examples include mechanical devices or air jets.

With these four components, smart sorting systems can efficiently automate the separation of items and materials, reaching throughputs of thousands per minute while maintaining extremely high accuracy.

Smart Sorting Technologies

RGB Optical Sorting

Inspects the colour and shape of each item for sorting purposes. RGB optical sorting requires simple hardware, with just an RBG camera needed, but relatively in-depth computer vision analysis to convert an array of pixel colour values into segmented items. RGB optical sorting systems are widely used for basic classification based on visual inspection. They enable simple and fast sorting decisions when items are visually disparate.

Infrared Spectroscopy

Goes beyond simple colour information to analyse objects based on reflected infrared light. Infrared spectroscopy can reveal certain chemical properties, such as density or moisture content. The technology is often used when more advanced sorting criteria are required, such as separating polymers of the same colour for recycling.

X-ray Analysis

X-ray analysis measures fluorescence and transmission to identify items based on chemical composition. As X-rays interact with different materials, they reveal specific structural properties, including density. X-ray sorting systems are used to distinguish between metals in medical applications.

Smart Sorting based on Hyperspectral Data

Hyperspectral imaging reveals spectral signatures across the near-infrared (NIR) and visible spectrum related to chemical properties. This data enables smart sorting based on in-depth spectral analysis. With much more detailed optical datasets available, HSI can improve upon RGB optical sorting and offer an alternative to infrared and x-ray approaches.

In the past, hyperspectral cameras have struggled to output and analyse the vast datasets for real-time applications. However, advances in snapshot imaging and data compression have led to the Living Optics Camera.

Capable of outputting 30 hyperspectral frames per second while being housed in a compact portable device, the Living Optics Camera comes with built-in computing for local data analysis and is capable of keeping up with smart sorting systems.

Plus, with detailed hyperspectral datasets, you can automate image processing and item classification using smaller algorithms compared to RGB datasets. With spectral information across the image, computer vision algorithms have spectral data to distinguish between items rather than relying solely on colour and shape. 

Applications of Hyperspectral Smart Sorting Systems

Hyperspectral-powered sorting systems have applications across a range of different industries:

• Food Processing: Sorting food products to guarantee quality and prevent defective items from making it to the shop shelves. With in-depth spectral datasets, users can also grade food products for dynamic pricing or improved shelf-life estimates.
• Recycling: Rapidly distinguish between different recyclable materials (   and even differentiate between quality levels to make recycling more efficient and effective.
• Agriculture: Remove defective seeds and identify weeds to guarantee plants’ health.
• Pharmaceuticals: Guarantee quality and correct labelling  by imaging tablets for potential defects and verifying packets correspond with the correct medication.

The Living Optics Camera

With real-time hyperspectral data and local compute for immediate image segmentation and item classification, the Living Optics camera enables smarter sorting systems. The snapshot technology simplifies the camera’s optics for a robust and portable device that can be easily integrated into existing sorting systems.

Upgrade your smart sorting systems and incorporate real-time spectral insights with the Living Optics camera. Get in touch today and learn more about how next-generation hyperspectral cameras are democratising the technology.