Thanks for all the good reads in the past threads patiam, DJ4Survey and all who contributed!

As mentioned in an earlier post, we are trying to measure the size of whales (i.e length and width) in order to research body condition.

If an RTK photo is opened in Notepad++ , a whole string of metadata is received, and among that Absolute altitude and Relative altitude. Is the given Relative Altitude data exact enough to be used, or does this raw data needs additional processing? We technically need only 1 good photograph per flight / whale / day-week (e.g. time interval) to track changes in size over time.

What kind of corrections or additional parameters are required to obtain optimal RTK height data – if any?

Good work @Strannik!
If you have good RTK fix (RtkFlag="50") and RtkStdHgt looks good then your relative altitude should be good to use without any further processing. Start measuring  whales & seals!

Thank you @patiam!

From  a cold and foggy Sea of Okhotsk greetings! Please allow me to ask you some advise:  We are trying to use the formula H= 36 X GSD  (H=Height; 36= a constant factor according DJI's manual instructions and GSD as Ground Sample Distance /  pixel unit (Photoshop).  Let me show you an example from our practice





The calculation would be:        GSD=   19.88 m (height)     :    36 (DJI factor)         GSD= 0.552

Size of measured object:  Quantity pixels -  148    X  - 0.552  (GSD) = 81.73cm
That seems to work well.
But with the next example it goes a bit different:



By using the same formula H= 36 X GSD we get an incorrect length: the whale as photographed should be > 12 meter.
Does the position of the RTK2-Mobile ground station(i.e. height and distance ) can influence the relatively height readings?

And is the formula the right on to be used here?

Many thanks!

Strannik Wolkov Posted at 6-19 03:51
Thank you @patiam!

From  a cold and foggy Sea of Okhotsk greetings! Please allow me to ask you some advise:  We are trying to use the formula H= 36 X GSD  (H=Height; 36= a constant factor according DJI's manual instructions and GSD as Ground Sample Distance /  pixel unit (Photoshop).  Let me show you an example from our practice

@Strannik Wolkov-

Good to see you making progress! A few thoughts:
H is not the only thing at play here; the camera angle as well as lens distortion also affect GSD across your image. The formula you're using is fairly accurate at the center of the CMOS sensor for an exactly orthagonal image, but that's not what you have. The appropriateness of the formula decreases with increasing camera angle, height, and and horizontal distance from the image center.
If you're just using jpegs straight from the drone with no distorion correction I'm not sure how consistent your results will be. Image processing built into photogrammetric software generally tries to account for these distortions to produce corrected images.

Further, you'd be better off with your images in an application made for measurement such as a GIS or NIH-Image/Image-J rather than P-shop.

Be sure your H is really the H above the object (not just ATO, which may or not be the same elevation as your object)... If your object being measured is below the surface of the water then it's more complicated. Also if it is bent (arched back) so that caudal fin and rostrum are farther away from camera than middle of back, you will underestimate length. Without orthorectification, measurements only work within a single plane parallel to image sensor.
I know you'd rather not hear it but you may really end up needing something for scale in every image (laser dots or a known object).

@patiam

Rain and fog and time to go through test results. We have been working between 10 - 15 meter with 1-meter increments , trying to evaluate the RTK accuracy, which seems pretty decent. Mind you, dry-flying - over land, as the whales seem currentlyvery focussed on food, emerging for two breaths and gone they are.
A table here below shows its accuracy which is between 0.36 - 1.03 cm. Looks decent enough  to me. But you are right: we should work on distortion correction, as the little table deck we are photographing is neatly placed in the centre of the picture where the whale will be larger and the body will be closer at the edges.



Which version of GIS or NIH-Image/Image-] you would suggest? Reading up Burnett, Dawson, Christiansen and Bejder: they mention the same as you. I think the laser idea is a very simple and effective tool that helps scaling within the work distance, as you mentioned in an earlier post.  And the simpler the equipment or method,  the less possibilities for mistakes. I fully agree. Any leight-weight, durable green, double- or tripple laser set-up you would recommend? I had an improvised double-laser connected to a P4P but it still a 250 g. pay-load and the drone was not too fond of it. Looking seriously further in to that option but smaller and lighter.

Water transparency, sea state, behavior of the individual whale - all come in to play and much more.

@Strannik-

Looks like its working pretty well for you in testing, at least for something ~ 1m long and flat

I don't have any specific recommendations on software versions, I just tossed Image out there as I have used it before for measuring critters in video frame grabs (long ago). As for lasers, I think you should be able to do better than 250g for a pair, but haven't sourced them myself recently so I'm afraid I'm no help there either. I'll see if I can find anything though.

Nice E. robustus pix, keep 'em comin! We have them here as well but lately everyone has been going gaga over the Orcas & Humpies.

I love your research. Great efforts. Sorry I'm not the brightest kid on the block to help you. So here goes something stupid from my mouth. Hand guns have very small lasers.  I mounted one on my HS2000. They are very light and 2 could be mounted on the legs easily.

Hey @Strannik:

Good paper here on volumetric & weight estimates of baleen whales using sUAS photos:

https://besjournals.onlinelibrar ... 111/2041-210X.13298

(they used a laser rangefinder to measure altitude)

Cheers!
-pat