A new era for machine vision: VTT’s breakthrough broadens hyperspectral imaging to exciting, mainstream applications

Historically, hyperspectral imaging was developed for remote sensing applications, such as satellite imagery and aerial surveys. In these scenarios, the sun was the only practical light source available.

The historical burden of using the sun as a light source has long limited the application of hyperspectral imaging in the wider industrial context. Hyperspectral imaging systems have been optimised for scenarios where controlling the light source wasn't an option.

“We had been researching hyperspectral imaging for a long time and noticed that to fulfil its true potential, the technology needed a fundamental rethinking,” says Jussi Soukkamäki, VTT’s Research Lead in Hyperspectral and Imaging Technologies. “The potential of using an artificial light source with a high-speed camera had been widely overlooked.”

VTT’s new approach combines an adjustable, artificial light source with a high-speed linear camera. This combination offers unprecedented speed, resolution and flexibility in spectral imaging. 

By breaking free from the constraints imposed by reliance on solar illumination, VTT’s innovation opens new possibilities for hyperspectral imaging in industrial and close-range applications, where close control over lighting conditions is required.

Soukkamäki believes that the expected breakthrough of hyperspectral imaging is near: “This is an exciting time for hyperspectral imaging, as we are constantly drawing closer to a wider acceptance of the technology. We believe that our game-changing solution is the key to a final breakthrough.”
 

Some of the most promising industries include: 

Food processing
In the food industry, this technology can revolutionise quality control and sorting processes. High-resolution imaging enables the detection of contaminants as small as individual rice grains, ensuring food safety at an unprecedented level. The system can perform real-time quality grading of fruits and vegetables, identifying subtle differences in ripeness, bruising, or disease that might be invisible to the naked eye. Moreover, it can quickly detect foreign objects in food production lines, such as small pieces of plastic or metal.

Recycling and waste management
The recycling industry stands to benefit greatly from this advanced imaging technology. It allows for precise sorting of different plastic types in recycling streams – a crucial step in improving the efficiency and effectiveness of plastic recycling processes. The high-resolution capabilities make it possible to identify and separate even small plastic flakes that were previously too difficult to sort accurately.

Agriculture
In agriculture, this technology can be integrated with drones or machinery to provide detailed crop health monitoring. Farmers can gain insights into plant stress, nutrient deficiencies, or pest infestations allowing for early intervention. The system's ability to operate in low-light conditions also enables night-time monitoring. Additionally, hyperspectral imaging can enhance precision agriculture applications, such as targeted weed detection and treatment. For instance, when integrated with spraying equipment, it could significantly reduce herbicide use by applying treatments only where needed.

Industrial quality control
Across various manufacturing sectors, this hyperspectral imaging technology can enhance quality control processes. It can detect subtle defects or contaminants that might be missed by conventional imaging systems, ensuring higher product quality and reducing waste. Real-time material composition analysis could be used for process optimisation and product verification in industries ranging from pharmaceuticals to electronics manufacturing. 

 

Traditional hyperspectral imaging technology performs the spectral separation within the camera itself, using complex optics to break down the incoming light into its constituent wavelengths. This approach works well for large-scale, outdoor applications but results in expensive, bulky systems with limited speed and resolution when adapted for industrial use.

The new approach completely reimagines the role of the light source in hyperspectral imaging. “The key innovation lies in shifting the spectral separation from the camera to the light source, a radical departure from traditional hyperspectral imaging systems”, Soukkamäki explains. 

Instead of relying on ambient light or broad-spectrum illumination, the new solution uses:

• A customisable supercontinuum laser light source developed by VTT
• A high-speed filter for rapid wavelength selection
• A standard high-speed line scan camera.

The innovation has clear benefits and broadens the potential use cases for hyperspectral imaging. By controlling the light source, this configuration allows for:

• Simpler, more robust camera design. Traditional hyperspectral cameras require complex optics to separate different wavelengths, making them expensive and sensitive to environmental factors. By moving spectral separation to the light source, the new system can use a standard line scan camera with normal optics. This not only reduces costs but also increases durability and reliability in industrial environments. The simpler camera design also means easier integration into existing machine vision systems.
• Optimised data collection, capturing only necessary wavelengths. In many applications, only specific wavelengths are needed for analysis. Traditional systems capture the entire spectrum, leading to excessive data and processing requirements. The new technology can selectively illuminate only the required wavelengths, dramatically reducing data volume. For instance, if only 10 wavelengths are needed to distinguish different plastic types in recycling, the system can capture just those 10, instead of hundreds of spectral bands. This optimisation speeds up processing and reduces storage needs, making real-time analysis more feasible and reducing energy and material requirements.
• Improved dynamic range through individual exposure time adjustment for each wavelength. Different materials reflect light differently across the spectrum, often resulting in some wavelengths being overexposed while others are underexposed in traditional systems. The new technology allows for individual exposure time adjustment for each wavelength resulting in a significantly improved dynamic range. This allows for more accurate spectral measurements across a wider range of materials and conditions.
• Easier calibration and maintenance. Traditional hyperspectral cameras require precise alignment and calibration of their internal optics, which can drift over time or due to temperature changes. The new system's simpler camera design and external spectral filtering significantly simplify calibration by enabling software adjustments to the light source rather than requiring physical adjustments to the camera. This reduces downtime and maintenance costs, making the system more practical for industrial applications.
  
   

The potential applications of this revolutionary hyperspectral imaging technology are vast, and we've only scratched the surface of what's possible. We now invite businesses to explore how this technology can be applied to solve challenges in their respective fields.

If you have ideas for novel applications or would like to discuss how this technology could address specific challenges in your industry, please don't hesitate to contact us. 

“Let's work together to push the boundaries of what's possible with hyperspectral imaging and create innovative solutions that can make a real difference in the world“, Soukkamäki concludes.