GCP or just a ruler ?

[birdofprey]

I do not have the need that the model is georeferenced.

Instead I have the need that the model can be measured in 3 dimensions (x,y,z) and that these measures are accurate to the centimeter (an error of like 10 centimeters is acceptable).

It's necessary, in any case, spreading and measuring (with a total station) a dozen of GCPs along the area to be mapped or I can obtain a precise model (centimeter) simply by placing a ruler (scale bar) along the ground ? If the second one, just one ruler or more rulers ?

Thanks.

[Alexey Pasumansky]

Hello birdofprey,

If you are interested in dimensions only, then you can use scale bars to set up the reference distances. I suggest to use multiple scale bars with the dimensions comparable to the object size (I mean not use 1 meter distance to scale the football pitch).

[birdofprey]

(EDIT)

Thanks for your answer.

Today I did a lot of testing and what I have seen is that with my drone (a Phantom 4 with some "flight customizations" but no RTK) I get the correct RELATIVE measures (in the xy plane and along z axis) without GCPs or scale bars...

How possibile ?

[SAV]

Hi birdofprey,

You should achieve quite good results by using the geotagged pictures from your Phantom 4 drone in combination with a few scale bars (try to place them throughout your survey area). As already mentioned by Alexey, the scale bars need to be long enough, probably 10s of meters. I generally lay out some markers (ground control points) and measure the distance between them using a tape measure.

The scale bars will allow you to get accurate distance measurements in your model. However, be aware of the general error of your model that is inherited from the relatively inaccurate camera locations (geotagged images). This means that your whole model is not only shifted but also tilted. For example if you would like to do some surface water flow simulations, a tilted model is not that great. But if you for example would like to simply measure the distance between 2 objects, the general tilt of the model might not be that critical.

Regards
SAV

[birdofprey]

With just the drone I can get an error between 10 and 20 cm (maximum).

Its an errror which is still good for my purposes.

Anyway I think that a ground support survey with at least 5 points (4 corners + 1 central point) is necessary for verification. For this purpose I am trying to get a low cost RTK GPS system... I've found I can build it buying only 2 chips...