Machine Vision Technology Forum 2017

NORDIC MACHINE VISION

TECHNOLOGY FORUM 2017 >

24 October I Stockholm

Read more

HOW TO SOLVE COMPLICATED

ILLUMINATION SET UPS >

Learn more about intelligent lighting | Part 2

Read more
Improve your level of machine vision know how

OUR KNOWLEDGE

Helping you get it right

LEIA explains: What is the effect of flat field correction and why is it absolutely essential for the use of line scan cameras?

The invitation to our Machine Vision Technology Forum included the following quiz question. Let us give you the answer now!

LEIA provides the answer

What kind of problem, that usually occurs with line scan cameras, cannot be solved by using flat field correction (FFC)?

The possible answers:

a) Inhomogeneity of the sensor chip
b) Distortion caused by the optics
c) Intensity falloff towards the edges
d) Lens vignetting

LEIA provides the answer

CEOwl of the European Imaging Academy

The flat field correction (FFC) is the calibration of the camera sensor which results in an optimised recording enabling rapid and reliable evaluation.

"On each sensor there is a certain degree of inhomogeneity in the brightness output of each pixel due to natural manufacturing tolerances“, explains Lars Fermum, European Imaging Academy chief instructor at STEMMER IMAGING.

"Due to these tolerances each pixel can react differently to the same amount of light. The resulting differences in brightness within the image do not have much of an impact when using area scan cameras, as the differences occurring within the entire image are minor. The overall image is barely affected and is usually sufficient for most applications. However, when using line scan cameras where the height of a line scan camera sensor is only one pixel means that any pixel error will be repeated in each recorded line at the same position. This image error might occur, for instance, in the form of vertical stripes and will have significant impact on the recorded image data.“

These image errors can be compensated for by means of a flat field correction which comprises two steps: the Fixed Pattern Noise (FPN) calibration and the Photo Response Non Uniformity (PRNU) correction.

The first step of the flat field correction – the low dark noise calibration – is used to reduce the Fixed Pattern Noise (FPN) – also known as Offset Noise or Dark Signal Non Uniformity (DSNU) – as far as possible. For this purpose, the lens needs to be covered. Then, all pixel values are standardised by using an offset. This process compensates for inhomogeneity of the sensor chip.

In the second step, the Photo Response Non Uniformity (PRNU) calibration, an analogue gain is used to obtain a certain target value for all pixels in uniform lighting conditions. This process eliminates the intensity falloff towards the edges. At the same time – as a side effect - lens vignetting can be compensated, too. This results in images which show uniform brightness over the entire width.

After the flat field correction the line scan image is free from stripes and shadings. This makes image analysis easier, more convenient and reliable since image errors have been minimised and you do not have to perform any subsequent corrections by using a software.

However, a distortion caused by the optics cannot be eliminated by a flat field correction. Distortions like these can only be eliminated from the recorded data by applying a software fix. In order to keep the effort as low as possible, we recommend the use of high quality lenses, since they show lower distortion.

So, the right solution to our quiz question is b

The problem of distortion caused by the optics cannot be solved by flat field correction. Congratulations: Many of you knew the right answer. The lucky winners of the Machine Vision Technology Forum VIP package have already been notified.

By the way: You can learn how to perform a flat field correction step by step. Just watch our video!