.::APHESA SPRL - imaging, electronics and software consulting::.

Why using average instead of variance in EMVA1288 dark current measurement

The EMVA1288 standard proposes two methods for dark current measurement. The first method uses the average variation of dark images when exposure time is increased, while the other uses the variation of the variance of the dark images.

EMVA1288 logo

Both methods should be as robust as they are based on the measurement over a large number of pixels and an interpolation of the measured data by a linear fit. So why are there two methods and when should we use each of them?

Many image sensors or cameras have a feature called dark current compensation. This feature measures the dark level based on dark reference pixels or use calibrated dark current information in order to compensate the offset drift due to pixel leakage. If this feature works perfectly, then the dark level does not depend on the exposure time and does not depend on temperature. It is therefore impossible to measure dark current based on the average of dark frames. However, because such reference pixels are still slightly sensitive to some light, especially longer wavelengths, and do not exactly behave like the real pixels because of difference capacitance over the shielded pixels, the method is improved by various simple algorithms that sometimes can make the offset decrease with exposure time, wrongly behave at very short or very long exposure times or slowly oscillate. These effects can sometimes be seen in the dark plot.

Using the variance instead of the average gives a result that is not affected by the dark current compensation as the variance does not depend on the average output level but only on the average dark current. As dark current is random, it exhibits a shot noise that grows proportionally to the dark level itself and is therefore proportional to the dark current and the exposure time. As this is a noise, it is not corrected or filtered and is therefore always available.

However, even if the sensor or camera does not have a dark current compensation feature, we still see discrepancies between the average and the variance measurement and these discrepancies are still being investigated. It is currently recommended to report both values when available.

It is also important to note that a camera or a sensor should always give access to an analog offset that allows the user to shift the dark level high enough above the 0DN value so that the full distribution of the pixels can be seen, otherwise the EMVA1288 measurements are not possible, not only for dark current, but for all other parameters as well and the sensor or the camera will not be compliant.


About us

Aphesa develops custom cameras and custom electronics including FPGA code and embedded software. We also provide EMVA1288 test equipment and test services as well as consulting and training in machine vision and imaging technologies. Aphesa works in several markets including industrial, medical, oil&gas and security.