Benutzer:Dirk Hünniger/bin

Why binning is not needed

a discussion for aerial imaging purposes

Dirk Hünniger , Gordon Freeman

Wikibooks

Abstract

In this short text we will discuss the advantages and drawbacks of binning in our aerial imaging application. We will come to the conclusion, that the same effect as binning can be reached by modifying other parameters of the system in a suitable manner. We will show that these methods lead to better results than binning.

Introduction

During the flight the camera takes images of the ground. These images correspond to a certain width $w$ and height $h$ on the ground given in meters. Usually ${\frac {h}{w}}\ll 1$ holds. The device usually moves in the direction of the height. The goal of the application is to stitch individual images taken at slightly different places over an area together in order to get a highly resolved image of the full area. Therefore it is required that two images taken directly after each other overlap slightly or at least have no gap between them. In case of a limited rate of exposures per second and a given altitude of the imaging system as well as given parameters of camera and lens it is therefore necessary that a certain limit for the speed (given in meters per seconds) of the devices is not exceeded.

To Bin or not Bin

Binning is interesting to consider since it reduce the amount of data needed by a single image and thus might help to increase the frame rate. The disadvantage of binning is that it reduced the spacial resolution. The question to look at is how both of theses effects contribute to the overall result an see if there is any advantage or disadvantage caused by the decision for or against binning. So lets assume we want to bin by a binning factor of $b$. That means we take squares of adjacent pixels of width $b$ and height $b$ and combine them to a single pixel, thus reducing the number of pixels in the image by a factor of $b^{2}$. Assume that we want to scan the area with the same spacial resolution as before. This way we will have to lower the altitude of the device by a factor of $b$. This means that $h$ also goes down by the factor $b$. The same holds for $w$. Assuming that we were going at the maximum speed before we now also have to reduce the speed also by a factor of $b$ since $h$ has decreased. The width of strip we acquire in a single straight flight has also decreased by a factor $b$ owed to the decrease of $w$. So the overall time needed to scan the whole area did increase by a factor of $b^{2}$. In this calculation we did not take into account any increase in the frame rate caused by the binning. It is quite realistic to increase the frame rate also by a factor of $b^{2}$ since the amount of data per frame decreased by that factor as explained above. But you also have to take into account shutter and readout effects which will cause an imbalance rendering binning less attractive. A similar argument shows that the amount of light hitting the detector is equal in both cases. The disadvantage of binning here is that more physical pixels are involved leading to a higher amount of thermal noise.

Conclusion

We showed that binning is not useful for aerial imaging applications with respect to the parameters we considered. To put it simple if you feel like you need to bin, you simply need to fly higher.