Calibration of scanned objects and of camera photos

Last update 30 Oct 2013.
When dpi resolution is not known you can calibrate your picture if  you know the exact distance between two points on the image.

After you clicked the Calibrate button a message pops up instructing you to set your two points. Set your first and second point! When done a new small window pops up with information about the distance based on the current dpi setting for the image.

Change the distance to the correct value. That will update the dpi resolution.

Using the calibration button to measure distances on a map.

If you have scanned a map with a scale on it, you can then calibrate it accordingly. Later you can use CooRecorder to look at different parts of the map. If you want to quickly measure the distance between two points, click on the Calibrate DPI button and set your two points. The pop up window will show you the distance between the two points according to your previous calibration!


Calibration of camera photos

When scanning an object with your desktop scanner, the pixels of the resulting picture will be spread out over the picture area almost exactly as the corresponding points on the object. I.e. you can safely use a ruler to measure either on the scanned picture or on the object itself!

When taking the picture of the object with a camera you have to be more careful if you need exact measurements.
Even when the photo is taken "at right angles" with your camera placed centrally above your sample, only the central part of your photo will have an approximately linear picture representation of your sample. This means that the dpi value varies over your photo with e.g. 10% lower dpi values at the outer borders than at the centre of the photo.

Accordingly your measurments with CooRecorder will then underestimate the ring width thickness in these outer areas. This probably does no matter if your data is intended only for crossdating your sample - i.e. finding the cutting year of the tree.

Though if you intend to use your ring width data to study tree growth then such measurements may be very misleading.
So be careful!

An example:

A camera photo of a ruler taken with a 5 Megapixel digital camera. The high resolution version of the photo above has 2560*495 pixels.
A calibration with CooRecorder from 0 to 45 mm gave a resolution of 1390 dpi.
A measurement serie of the millimetre marks is created with CooRecorder.
The millimetre marks can be seen as rings of a tree sample with exactly 1 mm between every yearly ring.

After creating that coordinate file with CooRecorder, the file is read into CDendro and the "ring-width-curve" above is plotted.

The curve should be a straight line if dpi value were constant along the measurement line!
Here the dpi value varies some 10% over the picture! The leftmost and rightmost 5 mm sections of the ruler has an actual 1290 dpi resolution while the central part of the picture has a resolution of 1460 dpi!

This is surely not a problem when crossdating - but it will be a problem to handle if you study actual growth rate!

Pictures created with a high resolution flat bed scanner will not give you this type of problem!



Copyright 2015, Cybis Elektronik & Data AB, www.cybis.se