Hyperspectral Camera Price vs. Performance – Finding the Right Balance

Next-generation hyperspectral imaging cameras promise to get the technology out of the lab and into the hands of people who can use it, creating a wide range of use cases. New cameras offer real-time hyperspectral data in a more robust and portable form factor while also reducing the price. 

This makes upgrading from RGB computer vision to hyperspectral-powered systems viable, delivering advanced insights in real time with locally run computer vision algorithms based on hyperspectral datasets. It is also possible to replace expensive hyperspectral imaging cameras with more cost-effective systems depending on the required performance. 

With more hyperspectral camera options now available, what do you need to know about price vs performance when it comes to finding the right balance for your use case? In this blog, we will discuss hyperspectral camera performance and the key factors impacting hyperspectral camera pricing, before providing advice on finding the best camera for you. 

Hyperspectral Imaging Camera Performance Factors 

There are many factors that define how a hyperspectral imaging camera performs. These factors quantify different aspects related to the quality of the final datasets and the practicality of deploying the camera. 

Spectral range  

Spectral range describes the range of wavelengths the hyperspectral camera is able to capture. While a traditional RGB camera covers just the visible part of the electromagnetic spectrum, hyperspectral cameras often extend into the infrared spectrum. By capturing light beyond the visible spectrum, it is possible to reveal new spectral signatures related to different material properties, enabling new use cases. 

Listed below are typical spectral ranges for hyperspectral imaging cameras with an example of the properties they can detect. 

• Visible (VIS) ≈400–700 nm: Detecting visual blemishes or in-depth colour information. 

• Visible-Near Infrared (VIS-NIR) ≈400–1000 nm: Returning new insights related to medical diagnostics, plant health, and minerals. 

• Short-Wave Infrared (SWIR) ≈900–2500 nm: Increased sensitivity to moisture, chemical bonds, and material types. 

• Mid-Wave Infrared (MWIR) ≈3–5 µm: Absorption features in the MWIR range can be used to identify certain gases for environmental monitoring or tracking industrial processes. 

• Long-Wave Infrared (LWIR) ≈8–12 µm: Captures thermal radiation to detect temperature variations. 

• Full-spectrum, Combines VIS, NIR, and SWIR: Provides detection capabilities from all three spectral ranges. 

While VIS and VIS-NIR wavelengths can be captured with low-cost and readily available silicon imagers, such as CMOS or CCD sensors, capturing longer infrared wavelengths requires more expensive detectors. Some of these also need cooling to reduce noise levels, leading to a significant jump in price and affecting how they can be deployed. 

Spectral resolution 

Spectral resolution determines the hyperspectral camera’s ability to distinguish between different wavelengths. A vital parameter, spectral resolution defines whether or not the camera can capture specific signatures and reveal new information about the materials present in the image. 

High spectral resolution cameras output a large number of narrow bands more accurately, recreating the original wavelength of light hitting the sensor. In contrast, low spectral resolution cameras utilise fewer, wider bands that lose much of the original wavelength data. 

Spectral resolution is defined by the spectral range, the number of wavelength bands, and the width of those wavelength bands. Typically, the width of the camera’s wavelength bands changes across the spectral range, and manufacturers quote values in nm at different points throughout the spectrum. 

The final spectral resolution for the camera depends on the technology used to capture light. Generally speaking, higher spectral resolution is beneficial and leads to higher prices. However, there are some trade-offs with higher spectral resolution leading to larger datasets and increased noise levels as each spectral band contains fewer photons. 

Spatial resolution 

How many pixels the final image contains. Spatial resolution determines the smallest detail that can be distinguished in the image. Typically higher spatial resolutions lead to higher hyperspectral camera prices. 

Depending on the imaging technology, the spatial resolution might be more complicated than a simple value for the number and size of pixels. Some snapshot hyperspectral cameras utilise two sensors to output both RGB and hyperspectral images. In these instances, the hyperspectral data will be sampled from across the image, rather than delivering pixel-by-pixel analysis provided by scanning-based cameras. 

Frame Rate 

While snapshot cameras provide comparatively lower spatial resolution data, they can output that data much faster. Frame rates determine how quickly the camera outputs hyperspectral images. These are typically measured in frames per second or Hz.  

New snapshot systems can deliver video-rate frames to enable a range of real-time monitoring systems capable of capturing hyperspectral data for moving objects. In contrast, scanning hyperspectral cameras are significantly slower, combining many frames to build up all of the hyperspectral data. This makes them unsuitable for imaging dynamic scenes or live analysis. 

Sensitivity 

Sensitivity measures the probability that a photon entering the camera is recorded by the sensor. This parameter is dependent on the wavelength of light, the sensor, and the optics. Higher sensitivity cameras allow the camera to be used in lower light conditions as each photon is more likely to be recorded. Sensitivity is critical at higher frame rates when the integration time to capture light making up the image is significantly shorter. 

Sensitivity also impacts parameters related to image quality, such as 

• Signal-to-Noise Ratio (SNR): How strong the signal is relative to the background noise. 

• Dynamic Range: The ratio between the biggest and smallest measurable intensity of light. Wide dynamic ranges mean the camera is able to capture both bright and dark regions accurately. 

Illumination 

The illumination conditions required to generate quality hyperspectral data. To make use of a hyperspectral camera’s entire spectral range typically requires a broadband light source. However, for use cases with a smaller spectral range of interest, more focused light sources with a narrower range of wavelengths can be utilised. 

Some hyperspectral cameras require fixed lighting conditions in terms of strength, wavelength range, and uniformity to output meaningful data, while others are more flexible. A typical source used for VIS, VIS-NIR, and SWIR is halogen illumination, while capturing shorter wavelength infrared radiation requires thermal illumination. Again, illumination is critical for faster frame rates as there is a shorter integration time to capture light. 

Ease of Use 

There are a number of challenges to overcome when utilising hyperspectral imaging cameras, including: 

• Calibrating these intricate devices to return accurate and reliable data that allows for easy comparisons across frames. This process is time-consuming and requires technical expertise in the field and knowledge of the specific camera in use. 

• Dealing with significantly larger datasets than RGB or multispectral systems, hyperspectral cameras need high bandwidth, processing power, and significant storage. 

• Interpreting large and complex datasets to perform meaningful analysis requires advanced algorithms and an understanding of spectral analysis. This makes most hyperspectral cameras inaccessible to non-specialists. 

• Controlling environmental conditions (lighting, background, etc.) to deliver reproducible datasets that are free of artefacts and aren’t affected by high levels of noise. 

• Slow framerates, particularly for cameras utilising scanning techniques, make hyperspectral systems unsuitable for real-time analysis or capturing dynamic scenes. Acquisition speeds also affect calibration. With extended wait times between frames, it takes significantly longer to check performance, test new parameters, or adapt the experimental setup.  

A hyperspectral camera’s ease of use defines how it performs in the real world and who the technology is available to. The overall ease of use is determined by a combination of hardware (imaging method, optical setup, local compute, etc.) and software capabilities provided by the vendor. Often, hyperspectral imaging vendors provide tools to simplify a range of functions, including calibration and data analysis. 

Size, Weight, and Power (SWaP) 

The camera’s size, weight, and power (SWaP) are related to ease of use, determining how easy it is to move and run a hyperspectral device. These factors can be particularly important for flight systems where resources are more constrained. However, SWaP also plays a vital role in the portability and deployment of a hyperspectral camera. Whether its use is restricted to the lab or if it can be moved to the field. 

Key Factors Affecting Hyperspectral Camera Price 

As with most industries, when it comes to hyperspectral imaging, you get what you pay for. The better the performance or the higher the specs, the higher the price of the hyperspectral camera.  

While it is simple to say higher performance equals higher prices, there are nuances related to certain factors that significantly impact pricing due to how they affect the hyperspectral camera. These include the scanning method, sensor, and software ecosystem. 

Imaging Method 

The imaging method refers to how the image is captured and how the camera converts incoming light into hyperspectral frames. Hyperspectral imaging methods include: 

• Push-broom (Line-scan): The most prevalent method that captures one line of pixels at a time. While push-broom cameras provide high spectral resolution, they do this by scanning across the image. This leads to slower frame rates and more complex optical designs that are less portable. 

• Snapshot: Captures all of the data in a single shot, enabling real-time, live spectral analysis. Plus, with less delicate optical arrangements, snapshot cameras are more robust and portable for use in the field. However, the speed of snapshot hyperspectral cameras comes at the expense of spectral resolution. Snapshot cameras output fewer wavelength bands compared to equivalent push-broom systems. 

• Tunable Filter: Integrates electronically controlled filters to capture a single wavelength at a time. Common filters include AOTF (Acousto-Optic Tunable Filter) and LCTF (Liquid Crystal Tunable Filter). While this method is relatively affordable and compact, capturing only one wavelength at a time leads to slow frame rates. 

• Whisk-Broom (Point-scan): A precursor to push-broom cameras, the whisk-broom method is now seen as incredibly slow as it scans the scene pixel-by-pixel. It does , generate high spectral resolution images but is rarely used in modern applications. 

• Fourier Transform Imaging: Uses interferometry to capture spectral data, similar to FTIR (Fourier Transform Infrared Spectroscopy). Hyperspectral imaging based on Fourier transform imaging requires bulky cameras that are not yet suitable for use outside the lab. 

Sensor 

The sensor defines the spectral range of the hyperspectral camera. While VIS and VNIR can rely on cheap CMOS sensors, extending spectral range further into the infrared region requires more expensive sensors based on semiconductors that aren’t silicon. Examples include Indium Gallium Arsenide (InGaAs) and MCT (Mercury Cadmium Telluride) sensors. 

Detecting lower wavelength light deeper in the infrared spectrum can also require cooling. These photons have lower energy and, therefore, are harder to detect above the thermal noise present. To reduce thermal noise you have to introduce cooling. While this improves performance, it complicates camera design, significantly increases price, and potentially limits how the hyperspectral system is deployed. 

Some snapshot cameras now utilise dual sensor designs that output both an RGB frame and a hyperspectral dataset. This allows users to quickly visualise and interpret the image while making it easier to adjust and focus on the specific areas of interest for spectral analysis. Additionally, computer vision systems can be trained on both colour and spectral data from the same scene.

Software Ecosystem 

The software ecosystem that accompanies a camera governs how you can actually use it in the real world. Specific data processing and analysis are required to convert raw sensor outputs into spectra and then to build automated computer vision models on top for different use cases. These models need to be able to understand or at least classify spectral features to deliver meaningful computer vision capabilities. 

Hyperspectral cameras can come with varying levels of software capabilities to simplify their integration into different use cases. This includes closed-off proprietary camera-specific software with limited functionality vs open software development kits with APIs that allow developers to build their own tools and integrate them into existing programs. Additionally, more expensive systems often deliver turn-key software capabilities such as real-time classification or sorting models that simplify camera use.  

Finding the Right Hyperspectral Camera for Your Needs 

Hyperspectral cameras are complex devices with many factors affecting performance and price. However, when it comes to finding the right option for your use case, you can simplify the decision by performing a “needs-based analysis.”  You want to identify the most important performance factors that will enable you to achieve your goals. This includes considering the: 

• Spectral features you are looking for and the resolution required to distinguish them 

• Speed at which you need to perform this analysis, and the associated frame rate required 

• Conditions in which the images will be captured, such as the environment and lighting conditions. 

The outcome of your needs-based analysis should return performance floors required for your use case. With this in mind, you can explore the trade-off between performance and price to find the optimal choice of hyperspectral camera. It is easy to aim for the highest-spec hyperspectral camera available. However, this often leads to wasted funds, with cameras delivering performance beyond what is necessary. 

Examples of hyperspectral camera performance factors to consider for different industries include: 

• Agriculture: Moderate spectral resolution required. While flight instruments can utilise push-broom cameras, direct field measurements are improved by faster readout and portable systems. This favours snapshot hyperspectral cameras. 

• Quality Control: Prioritise high framerates and real-time processing over spectral resolution for quality control and sorting use cases across various industries. The hyperspectral system will likely only need to distinguish between a fixed number of known spectral features but must make real-time decisions. 

• Healthcare: Still in the proof-of-concept research phase, however the risk of drawing wrong conclusions from hyperspectral datasets could have dire consequences. Therefore, healthcare use cases will typically require higher spectral and spatial resolution. 

• Defence: Wide spectral range and high spectral resolution to detect camouflaged threats and high frame rates for real-time analysis. 

Where The Living Optics Camera Fits in the Hyperspectral Marketplace 

Balancing performance vs pricing is critical to finding the best hyperspectral camera for your use case. Identifying the solution to this trade-off requires an understanding of spectral imaging, including the factors that: 

• Affect performance and the practicalities of using it in the real world 

• Determine camera pricing, including both hardware and software capabilities 

Thankfully, the next generation of hyperspectral cameras is democratising the technology by offering high performance without the high prices. In particular, advances in snapshot image capture deliver video-rate imaging for real-time monitoring while retaining spectral resolution that is more than capable of enabling use cases across a range of industries. 

This advance is demonstrated by the Living Optics Camera, a VIS-NIR video-rate hyperspectral system capable of video framerates up to 30 fps. Its small and portable hardware, combined with the Living Optics Python SDK and advanced software capabilities make it easy to use across different environments for real-world analysis. Plus, with tethered edge compute and dual sensor outputs you can perform real-time processing and analysis to build models that combine RGB and hyperspectral datasets. 

At a significantly lower cost than the competition, Living Optics offers a unique value proposition in the hyperspectral market. The device is comparable in price to multispectral cameras while offering 96 spectral bands across the visible and infra-red spectrum. 

Find out more by getting in touch with the Living Optics sales team. We can discuss the Living Optics camera and accompanying software capabilities in-depth, or we can help you perform your own needs-based analysis to determine whether the Living Optics camera is suitable for your needs.