Mission accuracy results

[Theo450]

I flew a 16 acre photogrammetric topographic mission the other day with my phantom 4. I had traditionally surveyed as built data to utilize for ground control points. I used Pix4D Capture to plan and fly the mission and ESRI's Drone2Map to process the data. There were 306 images with 80% overlap flown at an altitude of 160 ft agl. The processing was pretty easy, but it did take a few tries for me to learn how to use the software. Each run at it took about 30 minutes for setup and about 3 hours of processing time. I used 5 ground control points, and related each one to about 20 individual images. I ended up with a great orthodontic photo, a hillshade, and a digital surface model (raster surface) that have about 8 inch horizontal accuracy according to the as builts. I need to check the vertical accuracy, but judging from the post processing report it seems about the same.

I thought those results might be of interest in this sub forum.

 

[Pharm]

An orthodontic photo, eh? Sorry, no offense. Hope I didn't hurt your fillings. Sounds like the processing took a big bite out of your time! To tell you the tooth, it sounds like quite a job. Guess it beats a hard day at the orifice anyway. Ok, enough! I won't grind on. I know you meant "orthographic". Again, I apologize for poking fun. I'm just a bit abscessive/compulsive. Autocorrect?

 

[Theo450]

An orthodontic photo, eh? Sorry, no offense. Hope I didn't hurt your fillings. Sounds like the processing took a big bite out of your time! To tell you the tooth, it sounds like quite a job. Guess it beats a hard day at the orifice anyway. Ok, enough! I won't grind on. I know you meant "orthographic". Again, I apologize for poking fun. I'm just a bit abscessive/compulsive. Autocorrect?

LOL, gotta love spell check. Actually I meant orthomosaic.

 

[Pharm]

LOL, gotta love spell check. Actually I meant orthomosaic.

Ah! Orthomosaic! I got so wrapped up in my litany of dental puns, I forgot THAT term!

 

[Theo450]

Vertical accuracy looks to be about 1 foot.

 

[Kevinusma]

How did you set up your GCPs? I'm a total noob at GCPs so if you can, explain how you did it as simply as possible. Thanks!


Sent from my iPhone using PhantomPilots

 

[Silversand]

....interesting. Thank you. The whole world of *very low cost SfM* seems to be just emerging, with the inexpensive aerial / camera technology coming out (like the P4P, with its new 1 inch sensor) ! I've been doing some investigating into the current limitations of SfM (I always compare SfM with LiDAR, but LiDAR of course is more robust, because of the ability to gather intensity measures, and multiple returns, etc).

I do raw LiDAR processing myself (using LAStools), but I am going to process the structure-from-motion orthogonal images using a software bundle called VisualSFM (high density point clouds processing).

Looking at SfM theoretically, I think that the best success for a high quality data set would require a mission with multiple flight elevations using extreme overlap (direct over-head mission; then ever decreasing elevation missions...working inside small quads; my preference would be to mission 4 acre quads at a time, then process, then repeat...and then stitch the entire tile-set together as the last process)...this wouldn't take place till spring (we have at least 5 months of hellacious winters up here LOL).

Following this tech closely...

 

[Theo450]

How did you set up your GCPs? I'm a total noob at GCPs so if you can, explain how you did it as simply as possible. Thanks!


Sent from my iPhone using PhantomPilots

In the photogrammetric post processing software I use, I can import a table of x,y,z coordinates that I pick from a traditional survey. I make sure the GCPs are a visible feature on the ground like the corner of a concrete pad.

 

[Theo450]

Silversand has inspired me...

What is accuracy? Quantification of error based on another fallible way of collecting measurements? Yes...

The earth is not actually a sphere. It is a big lumpy rock. It all depends on where you are in the "geoid", which datum you select, what precision you can achieve with your instruments, and what you quantify your accuracy against.

The absolute truth in these matters is elusive.

 

[Silversand]

....an interesting example: ....Florida is just about done (may be done now?) migrating to the NAVD 88 standard....all Fed agencies are required to upgrade NAVD 88. Inside the boundary of Florida water management, the elevation numbers from NGVD 29 to the new standardized datum shift the elevation down appx ~~ 0.7 to ~~1.1 feet lower (this varies geographically)...