Calibrating each individual bolometer of an uncooled microbolometer focal plan array is mandatory to achieve a uniform image when looking at a uniform reference such as a black body. This is due to the behavior of each bolometer (or pixel) being inherently different from one another. The non-uniformity of a LWIR detector is impacted by various parameters such as the sensor temperature, the sensor gain, the sensor exposure time, the sensor frame rate, the optics used etc.

Different methods of calibration can be used depending on the application. A classical 2-point NUC (non-uniformity correction) consists of adjusting each pixel value by an offset (acquired using a ‘cold’ reference from a black-body) and a multiplicator/gain value (acquired using a ‘warm’ reference from a black-body) such that after the correction is applied, the response of the entire array is homogenous and the image quality optimal. A 2-point NUC correction produces an RNU (residual non-uniformity) that is optimal for the calibration temperature (T0) but that quickly degrades as the detector temperature moves away from T0 as shown in the figure below:

Variation of image quality/non-uniformity as a function of FPA temperature with a 2-point NUC calibration only

Depending on the application, there are several ways to compensate for the changes in image quality as the detector temperature drifts due to environmental conditions changing (daytime vs nighttime etc).

Method 1: The shutter-based approach

The use of a mechanical shutter introduced either between the sensor and the lens or in front of the lens allows additional 1-point NUC corrections, thus increasing the latitude around the original T0 . Additional 1-point NUC corrections enables operation at other sensor temperatures and can be done as many times as necessary. Though efficient, mechanical shutters have several drawbacks:

    • Loss of information as the camera is blinded during the shutter period

    • Additional bulk and weight to the camera

    • Additional cost

    • Additional power consumption from the electric motor

    • Decreased reliability inherent to the mechanical moving parts

    • Noise as the shutter opens/closes.

Method 2: Shutterless operation

comparison between thermal images with and without shutterless function
During shutterless operation, there is no need for a uniform thermal reference such as NUC or mechanical shutter. Instead, a factory calibration consisting of a 2-point NUC at a temperature T0 is performed with the lens attached. The shutterless algorithm natively expands the latitude around T0 as shown in the figure below:
Variation of image quality/non-uniformity as a function of the FPA temperature for a shutterless operation with a single 2-NUC factory calibration
For cases requiring an even wider range of detector operating temperature, an additional 1-point NUC can be performed at either a higher temperature, lower temperature or both. As a result, a large range of FPA temperature is covered without the need to acquire new 1-point NUC while using the camera:
Variation of image quality/non-uniformity as a function of the FPA temperature for a shutterless operation with a 2-NUC at T0 and two 1-point NUC (Tlow and Thigh) factory calibration
 In shutterless mode, the algorithm compensates for any change of the detector temperature without user intervention. Optionally, additional 1-point NUC with the help of an external mechanical shutter (in front of the lens) can be used to further improve the image quality and reduce the use of the shutter (thus improving reliability and decreasing power usage). 

 

All our LWIR thermal cameras, wether with processing on the host controller (thermal modules) or with on-board processing (engine cores), are able to store a number of shutterless calibrations (up to 8 depending on the models). 

IrLugX1M3â„¢ | SXGA Thermal Camera Module

 

Scroll to Top