Slashdot Mirror
Drones Could 3D-Map Scores of Hectares of Land In Just a Few Hours
sciencehabit writes: Unmanned drones aren't just for warfare. In recent years, they've been used to map wildlife and monitor crop growth. But current software can't always handle the vast volume of images they gather. Now, researchers have developed an algorithm that will allow drones to 3D-map scores of hectares of land in less than a day — an advance that is important for cost-effective farming, disaster relief, and surveillance operations.
Their computer program directly projects the points from each photo onto a 3D space without knowing the exact shape of the land or the camera positions. As a result, the tie points don't necessarily match up, which means the same corn plant can have two projections on the model. When that happens, the algorithm automatically takes the middle point between the two projections as the more accurate location and adjusts the camera position accordingly, one image at a time. Because the algorithm tweaks far fewer things at each step, the shortcut drastically speeds up calculations. Once the software has adjusted the camera positions for all the photos, the software repeats the entire process — starting from projecting the points to the 3D space — to correct for any errors.
Their computer program directly projects the points from each photo onto a 3D space without knowing the exact shape of the land or the camera positions. As a result, the tie points don't necessarily match up, which means the same corn plant can have two projections on the model. When that happens, the algorithm automatically takes the middle point between the two projections as the more accurate location and adjusts the camera position accordingly, one image at a time. Because the algorithm tweaks far fewer things at each step, the shortcut drastically speeds up calculations. Once the software has adjusted the camera positions for all the photos, the software repeats the entire process — starting from projecting the points to the 3D space — to correct for any errors.
what does it mean scores of hectares? [checks google]
For the 'mericuns, one score hectare is 49 acres. for the people who insist on SK/mks, one score hectare is 200,000 m^2. For the left-brain thinkers, there are 17 score hectares in Central Park in NYC or 64 football fields in one score hectare.
Scores of Hectares? Really guys?
Why, with our revolutionary fly-a-micating devices, which are capable of travelling at dozens of furlongs per fortnight, we will be able to monitor the Aether, and map the location of the peasants houses to within a few rods, thus ensuring we can maximize tithing.
A spokesman for the government was overcome with the vapors at the excitement of it all, and needed to be leeched lest her spleen overtake the rest of her humors and leave her dyspeptic and the evil spirits sway her from her normal temperament.
Off the record, a spokesman was hopeful that the new phrenology module would be available in version III.V, and evil people will be easily spotted from the air, and can then be rounded up for burning at the stake.
Goode Frye was optimistic this would remove the threat of the witches which have been stealing the souls of babies.
Lost at C:>. Found at C.
Such a bizarre quantification. Why not just say "roughly 50 acres" or "200 square meters"?
no kidding. This is how DEM's have been made for decades.
the good ground has been paved over by suicidal maniacs
If you actually learn to read, you'll discover that the new thing here is a newer algorithm that's significantly faster than the current "standard".
Depending on the accuracy obtained, such mapping would be highly useful on our farm for figuring out drainage. Some areas of a field might drain better with only a very small slope, if we knew where to put the channel. Currently the only real option is to drive over the field with an RTK GPS receiver and make a GIS map of elevations. Which works well enough (depending on the grid resolution; can get really old driving every 10 feet over 160 acres), but takes quite a long time to do.
So it makes it better, then makes that better?
Support my political activism on Patreon.
Talking with a large farm owner over the weekend who is ready to start using Drones to cover about 5500 acres as a supplement to their current sat. imagery. They were told no by the FAA, which says they will not be able to accommodate farming drones until 2016 at the earliest.
So instead of figuring out how it's going to benefit the farmer and working through the leading edge adoption issues with farms who are motivated to give it a try, they are regulating them out of existence, before it can even get started.
I can certainly understand the need for regulations as problems arise, but to write the regulations before there is an issue is just a typical stupid gov response. The FAA will make the entire process more expensive, more cumbersome and more restrictive than it needs to be without any real world experiences upon which to regulate.
It would be simple to set a weight, height and radius limit for farming drone use. Could be done in just a couple of days with 2-3 smart people but that is just too easy for gov work.
I've always been fascinated when software makes dramatic enhancements to the capabilities of existing hardware and data. Like a few years ago there was the release of the TLD algorithm which suddenly turned my old webcam into an futuristic object recognition/tracking device!
What I wonder is, when these software enhancements are made, does hardware usually evolve to converge with the software? Meaning, in this example, if the software is using a new method for processing point data, does that not mean the hardware could be made to collect point data in a way more conducive to this new method?
And is this kind of progress a common thing? Is it common or rare this leapfrog progressive dance of hardware and software?
Drones Could 3D-Map Scores of Hectares of Land In Just a Baker's Dozen of Milli-fortnights
Fixed for consistency.
How many Libraries of Congress is that?
-Rick
"Most people in the U.S. wouldn't know they live in a tyrannical state if it walked up and grabbed their junk." - MyFirs
The images are projected into 3D space to find overlaps, but from reading TFA, it sounds like the output is still a good old fashioned 2D photo, just one covering an awfully big area.
TFA states that the old algorithm breaks down once the number of source images exceeds a few hundred, at which point it can take thousands of hours to process. The new algorithm can accommodate over a thousand images and process them all in about four hours.
John
Note that there are about 260 hectares in ONE square mile.
3200.4 LoCs Yours truly, Unapologetic Pedant
200 square kilometers rather
I take that back, it is actually roughly equal to 1 LoC. My original metric conversion was off by a factor of 1000.
I don't see your problem. Hectares are a fairly common unit for measuring land areas, essentially it's the SI analogue to the acre - a unit of area that's readily applicable to most land-working uses, without introducing clunky customary-unit conversion factors. At 10,000 times smaller, measurements in square meters have far too many trailing zeros to be convenient, and at 100x larger you'll usually be dealing with small fractions of square kilometers. And at 4046.86 square meters, do you really expect any scientist or engineer to futz around with acres if they can avoid it?
Meanwhile in modern usage "scores" is a conveniently vague number that's a bit larger and less specific than "tens" or "dozens", while still being being considerably smaller than "hundreds"
--- Most topics have many sides worth arguing, allow me to take one opposite you.
More relevant units for those unaccustomed to SI: One hectare = 2.4711 acres (international presumably, there are a number of slightly different definitions of acre)
"Score" meanwhile seems to have largely lost it's historical specificity of 1 score = 20, or at least I've only ever heard modern usage in the indeterminate plural, wherein "scores" is a quantity larger than "dozens" but smaller than "hundreds"
--- Most topics have many sides worth arguing, allow me to take one opposite you.
Acres? What is this, the 1700s? Please use a modern SI unit such as hectares.
--- Most topics have many sides worth arguing, allow me to take one opposite you.
40 rods to the hogshead.
Many (most?) surveys are historical - until such time as all surveys have been updated to high-precision GPS coordinates you'll need traditional surveyors.
And GPS surveying isn't without it's own issues - traditionally land divisions tended to follow natural boundaries - rivers, ridgelines, etc. And the land does not stand still - in geologically active regions surface features, including your house, may move a few feet per year. Build a fence along a GPS-surveyed boundary today, and in a few decades it may no longer be anywhere near your property line. So, should everyone move their fences (and houses) on a regular basis, or should we draw the boundaries based on surface features and relegate the GPS coordinates to a short-term convenience?
--- Most topics have many sides worth arguing, allow me to take one opposite you.
Of course, this was exactly the kind of legitimate purpose people wanted drones to have all along, so actually much older than the drone issues people are usually excited about.
But the 3-D modelling math sounds very cool.
Exactly, in 1890 we used modern units, and 124 years later you still are using old ones (acres?)
Uh, isn't there 24 hours in a day?, so if 40 hectares were done in 4 hours, that would be 240 hectares done in 24 hours which is 2.4 square kilometers.
Still wrong. 50 acres is about 0.2 square kilometers.
when you were already out plowing or harvesting or fertilizing the fields? Why wouldn't you just do this continually over time to improve your maps, once you made the investment in the GPS receiver?