⚡ electrical engineering

Efficient infrared radiation calibration method for wide dynamic range infrared radiation measurement systems

This paper proposes an efficient calibration method for wide dynamic range infrared radiation measurement systems that utilizes a comprehensive nonlinear response model to significantly reduce calibration time while accommodating a broader range of system configurations without requiring strict linearity in optical attenuation or integration time.

Original authors: MU GU, XIANG PENG, SUJUN LI, QIFENG REN, JINMEI ZHOU

Published 2026-09-23

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: MU GU, XIANG PENG, SUJUN LI, QIFENG REN, JINMEI ZHOU

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the invisible world of heat, where objects glow with infrared light rather than visible color, scientists rely on specialized cameras to see the unseen. These instruments are essential for spotting aircraft in the sky, checking the integrity of stealth materials, or measuring temperatures without ever touching a surface. To make these cameras accurate, they must be calibrated: a process of teaching the machine exactly how much digital signal to produce for every specific amount of incoming heat. This is usually done by pointing the camera at a "blackbody," a perfect source of heat that can be set to precise temperatures. However, when these cameras need to see targets that range from cool ground objects to scorching hot engines, they must be equipped with variable settings. They use filters to dim the intense light and adjust how long they "look" at a scene to capture the right amount of detail. The problem is that calibrating a camera with so many different settings is a slow, tedious ordeal. It requires taking thousands of pictures at dozens of different temperatures, a process that can take a full day or more, during which the laboratory's own temperature might shift and ruin the measurements.

A team of researchers from the Chinese Academy of Sciences has developed a way to slash this time dramatically without sacrificing accuracy. Instead of treating every combination of filter and camera setting as a separate problem that needs its own unique set of rules, they created a single, unified mathematical model that understands how the entire system works together. They realized that the relationship between the heat coming in and the picture the camera takes is not a simple straight line; it bends and shifts depending on how much light is blocked by the filters and how long the camera waits to capture the image. By building a model that accounts for these curves and bends, the researchers could determine the camera's behavior using just a handful of data points. In their tests, they reduced the number of temperature steps needed for calibration from twenty-four down to just five. This change cut the calibration time from over twenty-two hours to a fraction of that, allowing the system to be ready for use much faster.

The researchers tested this new approach on a large infrared imaging system with a 600-millimeter lens, capable of seeing heat in a specific band of the infrared spectrum. They equipped the camera with three different filters that let through 100%, 10%, and 1% of the light, and they tested it at four different speeds for capturing images. In the old way of doing things, to calibrate this system, they would have to take pictures of a heat source at temperatures ranging from 50 degrees Celsius to 1200 degrees Celsius, stepping up by 50 degrees each time. This required capturing hundreds of images and waiting for the heat source to stabilize at every single step, a process that took 22 hours and 10 minutes. With the new method, the team selected only five specific temperatures that provided the most useful information across all the different settings. They fed these few data points into their complex model, which used a sophisticated calculation process to figure out the behavior of the entire system at once. The result was a set of rules that could predict the camera's response for any setting with high precision.

When they compared the results of this efficient method against the traditional, slow method, the accuracy was nearly identical. The new method produced an average error of 1.81 percent, which is very close to the 1.03 percent error of the traditional approach. This means the camera is just as reliable, but it is ready to work in a fraction of the time. The researchers also tested their method against another "fast" calibration technique that had been proposed by other scientists. That older fast method assumed the camera's response was perfectly straight and predictable, like a ruler. However, real-world cameras are rarely that perfect; their response curves can be slightly bent. Because the new method accounts for these bends, it worked far better on the actual equipment. The older fast method produced errors that were far too large to be useful, proving that trying to force a simple, straight-line solution onto a complex, curved reality leads to mistakes.

The team also applied this logic to a more complex calibration technique known as "inner-outer" calibration, which is used when a single heat source cannot cover the entire temperature range needed. This method usually involves calibrating the camera with a large, low-temperature source for the cool end of the scale and a small, high-temperature source for the hot end, then stitching the two together. This stitching process is notoriously difficult and time-consuming. By using their unified model, the researchers were able to perform this stitching efficiently, reducing the number of required temperature points from twenty-six down to six. The resulting accuracy was excellent, with an average error of just 2.08 percent, vastly superior to the 25.8 percent error seen with the traditional inner-outer method. This success demonstrates that by understanding the underlying physics of how light and time interact within the camera, scientists can bypass the need for exhaustive testing.

This work matters because it changes how quickly these powerful tools can be deployed. In fields like aerospace or defense, where ground-based systems must constantly monitor the sky for targets, the ability to recalibrate a sensor quickly is a significant advantage. If a system needs to be recalibrated because the weather changed or the equipment was moved, the old method would tie up the system for nearly a full day. The new method allows for a recalibration in a matter of hours or even minutes, depending on the setup. The researchers showed that this efficiency does not come at the cost of precision; the camera remains just as accurate as before. By moving away from the idea that every setting needs its own separate calibration and embracing a model that sees the whole system as one connected unit, the team has made infrared measurement faster, more flexible, and more practical for real-world use.

Technical Summary: Efficient Infrared Radiation Calibration Method for Wide Dynamic Range Systems

Problem Statement
Wide dynamic range infrared radiation measurement systems, essential for target detection, stealth material evaluation, and non-contact temperature measurement, typically employ variable optical attenuators and adjustable integration times to accommodate diverse target radiance levels. Traditional calibration methods for these systems—specifically collimator-blackbody, close-range extended source, and inner-outer calibration—rely on linear models fitted separately for each combination of integration time and attenuation setting. This approach necessitates a large number of calibration temperature points (e.g., 24 points from 50°C to 1200°C in 50°C intervals), resulting in excessively long calibration times (over 22 hours in the reported experiments) and potential inaccuracies due to ambient temperature drift during the process. While existing fast calibration methods (e.g., by Luo, Li, Chang, and Yang) attempt to integrate these variables into a single model, they often rely on strict linearity assumptions in both the attenuation and integration time dimensions. These assumptions limit their applicability to systems where nonlinearities are present, and they frequently fail to fully integrate the inner-outer calibration concept required for high-temperature range coverage.

Methodology
The paper proposes a unified, efficient calibration framework that addresses these limitations through three primary technical advancements:

Comprehensive Nonlinear System Response Model:
The authors develop a mathematical model that accounts for the energy transfer mechanisms involving both optical attenuators and integration time. Unlike previous linear models, this approach incorporates:
- Nonlinear Attenuation Correction: A quadratic function ( $\alpha_1\tau^2 + \alpha_2\tau + \alpha_3$ ) is used to model the relationship between actual and design transmittance, addressing deviations in attenuator performance.
- Attenuator Self-Radiation: The model includes the emissivity of the attenuator ( $\epsilon = 1-\tau\sigma$ ) and its thermal radiation contribution.
- Integration Time Nonlinearity: Quadratic and cubic correction terms ( $t^2, t^3$ ) are introduced to account for nonlinearities in the system's response to integration time, particularly regarding thermal stray radiation offsets.
  The resulting equation (Eq. 5) contains 11 unknown parameters ( $R_0, \alpha_1, \alpha_2, \alpha_3, B_{out1}, B_{out21}, B_{out22}, B_{out23}, \sigma, B_{att}, B_{det}$ ) which are estimated using a nonlinear optimization algorithm (interior-point method) rather than simple linear least squares.
Integration with Inner-Outer Calibration:
The model is designed to be universally applicable, seamlessly integrating with the inner-outer calibration concept. This allows the framework to cover the full optical path and temperature range by:
- Performing efficient outer calibration (using extended sources or collimator-blackbodies) in the common temperature range.
- Performing efficient inner calibration (using cavity blackbodies) for high-temperature ranges.
- Deriving correction coefficients ( $a$ and $b$ ) in the overlapping temperature range to unify the inner and outer calibration equations into a single model for the full temperature range.
Efficient Data Acquisition Strategy:
By utilizing the unified nonlinear model, the method drastically reduces the number of required calibration points. Instead of dense sampling across all settings, the method selects a minimal set of blackbody temperatures (e.g., 5 points for collimator-blackbody, 6 points for inner-outer) that ensure valid data coverage across the linear grayscale ranges of various attenuator and integration time settings.

Key Contributions

Unified Framework: The paper presents a single calibration equation capable of handling all working settings (attenuation and integration time) simultaneously, eliminating the need for separate calibration equations for each configuration.
Nonlinearity Handling: By incorporating quadratic and cubic corrections for both attenuation and integration time, the method removes the strict linearity constraints found in previous fast calibration methods (such as Yang's method), making it applicable to a broader range of infrared radiometric systems.
Full Coverage: The approach successfully integrates the inner-outer calibration concept, ensuring accurate calibration from ambient to high temperatures (up to 1200°C) without the limitations of extended-source blackbodies.

Experimental Results
The proposed method was validated using two experimental setups: a 600mm aperture system (collimator-blackbody method) and a 450mm aperture system (inner-outer method).

Efficiency:
- In the collimator-blackbody experiment, the proposed method reduced the calibration time from 22 hours and 10 minutes (traditional, 24 points) to a fraction of that time using only 5 blackbody temperature points.
- In the inner-outer experiment, the method achieved a fourfold efficiency improvement, requiring only 6 calibration points compared to the 26 points needed for the traditional inner-outer method.
Accuracy:
- Collimator-Blackbody: The proposed method achieved an average relative error of 1.81%, which is close to the traditional method's 1.03%. In contrast, Yang's method exhibited significantly larger errors, particularly at high attenuation settings, due to its reliance on strict linearity assumptions.
- Inner-Outer: The proposed method achieved an average relative error of 2.08% for settings requiring correction. This is significantly more accurate than the conventional inner-outer method (25.80% average error) and comparable to the full-temperature-range blackbody-collimator method (1.84% average error). The conventional inner-outer method failed to provide satisfactory results for settings not explicitly used in the correction process.

Significance
The paper claims that this efficient calibration method offers a practical solution for wide dynamic range infrared systems by significantly reducing calibration time and labor while maintaining high accuracy. By overcoming the linearity constraints of previous fast methods and unifying different calibration techniques (collimator, extended source, and inner-outer) into a single nonlinear framework, the method enhances the universality of infrared radiometric calibration. The authors state that this approach holds significant scientific value and engineering application prospects, particularly in reducing system task response efficiency and mitigating the impact of environmental temperature fluctuations during the calibration process.

More like this