Calibrating Uncooled Thermal Cameras for Accurate Temperature Measurements in Agricultural Applications

Sunday 06 April 2025


The uncooled microbolometer-based thermal focal plane arrays (UC-FPAs) have revolutionized the world of thermography, offering affordable and compact solutions for a wide range of applications, from agricultural monitoring to environmental surveillance. However, these cameras are not immune to the challenges posed by environmental factors such as wind velocity, radiation levels, and temperature fluctuations. The result is a persistent drift in offset and gain, making it difficult to extract accurate thermographic information.


In recent years, researchers have been working on developing scene-based nonuniformity correction (NUC) methods that can eliminate the need for continuous calibration. One of the most promising approaches involves using a sequence of shifted images, with relative shifts caused by realistic drone hovering modeled by homography transformation. This method has shown great potential in correcting not only offset but also gain and thermographic values.


The key to this approach lies in its ability to jointly estimate unknown gains and offsets from a series of images taken at different blur levels. By using a minimum likelihood estimator, which is found using alternating minimization, the algorithm can accurately recover both gain and offset profiles, even when they are space-variant. The registration process is carried out using the generalized Lucas-Kanade method, allowing for accurate alignment of the shifted images.


In simulations, the proposed algorithm has shown remarkable accuracy in restoring thermographic values, with a mean Pearson correlation greater than 0.9999998 between ground truth and restoration. Under ideal assumptions, this translates to an equivalent mean restoration error of less than 0.01°C. The results are impressive, especially when compared to previous methods that have achieved significantly larger errors.


One of the significant advantages of this approach is its ability to simplify the requirements for hardware implementation. By using a series of shifted images rather than focusing and defocusing kernels, the algorithm can operate with a much simpler setup, making it more practical for real-world applications.


The future of scene-based NUC looks promising, with potential applications in various fields such as agriculture, environmental monitoring, and even medical imaging. As researchers continue to refine this approach, we can expect to see even more accurate and efficient methods for correcting nonuniformity in thermal cameras. For now, the prospect of using affordable and compact UC-FPAs without the need for continuous calibration is an exciting one, with significant implications for a wide range of industries and applications.


Cite this article: “Calibrating Uncooled Thermal Cameras for Accurate Temperature Measurements in Agricultural Applications”, The Science Archive, 2025.


Microbolometer-Based, Thermal Imaging, Nonuniformity Correction, Scene-Based, Nuc, Offset Correction, Gain Correction, Thermographic Values, Image Registration, Homography Transformation.


Reference: Peretz Yafin, Nir Sochen, Iftach Klapp, “Scene-based nonuniformity correction with homography transformation” (2025).


Leave a Reply