Hi chris mills,
Looks like I was writing my post while you were writing yours.
A situation I have run into doing multiple flights to cover an area is, if you are creating orthophotos there is a noticeable change in shadows. I had a project where I tried to get around this by doing 2 to 3 flights a day, at the same time of day, over a period of a few days.
Here's a link to an interactive map of the orhto and DSM for that project:
http://www.i-maps.com/3d-garage/al-baleed/al-baleed4.html
You can still see some variation in shadow, mainly along the sea wall.
I am really so glad this thread has grown and continues to be populated. In light of the P4Pro discussion I have discovered that the latest camera offered is essentially the Zenmuse x4s, which is a great little mapping camera - which explains the increase of cost - and the increase in your capabilities.?ÿ
Keep it coming as as Mr. Ott says - I love this Bar.
Hi chris mills,
Looks like I was writing my post while you were writing yours.
A situation I have run into doing multiple flights to cover an area is, if you are creating orthophotos there is a noticeable change in shadows. I had a project where I tried to get around this by doing 2 to 3 flights a day, at the same time of day, over a period of a few days.
Here's a link to an interactive map of the orhto and DSM for that project:
http://www.i-maps.com/3d-garage/al-baleed/al-baleed4.html
You can still see some variation in shadow, mainly along the sea wall.
Yes, one advantage of the UK is you can pick days when the sun isn't out. As long as there is only a short gap between the flights then the shadows shouldn't move around too much and if you are flying middle of the day then they should only be very short.If you are flying large areas then the change in shadows does show over the whole site, but generally they creep round gradually so it isn't too noticeable. We see it quite often on sites of over 500 ha. but clients rarely notice. It doesn't affect the DEM.
Fixed wing has a duration advantage over a rotary. A single flight covering 70 ha. will only take around 20 minutes and if the wind conditions were calm there is enough power left in the batteries for a further flight.
Incidentally, you can extend the area of your coverage by walking along the grid as you fly, trying to keep up with the aircraft. That way, on a 15 minute flight you have extended the area by however far you can walk in, say, 10 minutes - try not to trip over things as you walk!
I like your example - it must be nice to fly just plain sand with no vegetation to worry about. I'll post separately on that one.
Sea Walls - yes they are interesting. We did 5km. a few weeks ago and had the usual problem of the software trying to smooth the ground up and over the wall. We have an experimental site lined up for a few weeks time to try and resolve the best way of minimising that problem. Currently we run a GPS line along the root of the wall and add that in to the DEM.
?ÿ
Since when did the sun come out in UK? Huh, news to me... ??ÿ
My first drone project was 220 acres with a P4P. There was waist high sage brush and grass over 30% of the job. We were able to extract the data we needed and generate a better surface than the test areas using only gps. The key was exracting data, not trying to manipulate and use the entire dsm. We augmented with one guy hitting things we thought would be hard to pick in photos or those critical for elevation. The profit from that job paid for the entire program up to that point.
?ÿ
We allow for some field work to check and "calibrate" the vegetation. The guy fixing the GCP also records as he walks between those locations picking typical vegetation as he does so. We then have a method of comparing the DSM vegetation heights with the ground; usually we generate a high density grid as this gives good average values from the DSM. General area corrections are then applied to the vegetation types. Yes, you need to spend a bit of time putting in rough edge strings to mark each area, but the end result is normally at least as good as the results of Joe wandering around occasionally pressing the button on his logger. Stringing of the vegetation changes are done over a backdrop of the orthophoto direct into our survey software.
It's worthwhile walking critical features like road edges.
Some software vendors have developed solid algorithms to extract ground points in vegetation. Adding in ground truthing is a must.
The best part of all this is, it's only going to get better...
We use Photoscan - to date only LIDAR seems to be able to get really consistent extraction of ground from under vegetation (note I said consistent rather than good).
Feature extraction will also come with time at this "budget" end of the market. Its a choice at the moment of time spent manually extracting or knowing somebody with a spare Cray - and you still need to give it a visual to make sure the automated system has worked.
Since when did the sun come out in UK? Huh, news to me... ??ÿ
It's out now! ??ÿ
It looks like the UK had plenty of sun and droughts this summer. I seen a couple news reports where old (ancient) walls and earth works were visible from the air because of the drought conditions over there. Climate change... what climate change. ??ÿ
Here is an image of a color relief image with contours from one of our drone projects focused on stream bank erosion. The water was clear enough that we actually picked up the relief in the channel. If you look at the channel on the left side you can see how the shadow from the tree in the stream introduced noise.
Here are links to a 3d model of the project and an interactive map with higher resolution imagery:
http://www.i-maps.com/nature-conservancy/elk-river-streambank-stabilization-map.html
We use Photoscan - to date only LIDAR seems to be able to get really consistent extraction of ground from under vegetation (note I said consistent rather than good).
Feature extraction will also come with time at this "budget" end of the market. Its a choice at the moment of time spent manually extracting or knowing somebody with a spare Cray - and you still need to give it a visual to make sure the automated system has worked.
I've been extracting ground data under vegetation from photogrammetry reliably for over a year. 3d feature lines are still a bit hit and miss in the less expensive software but it's here.
Nice image Bisenberger. We've found similar results. See article starting on page 23 in Geomatics World.
https://www.geomatics-world.co.uk/magazine/september-october-2017
The tree "shadow" effect is probably because the trees cut off view from some of the images, so that a bias is introduced in the intersection ray bundle. (Similar, I think, to the fillet at the bottom of walls, as the images are only seen from one side.)
