Orthomapping using fixed wing aircraft and panasonic DMC-ZS20

[hootener]

I've inherited a project that requires the use of Agisoft PhotoScan to generate terrain maps (orthophoto and DEM generation) of an archaeological dig site from aerial images collected using a fixed wing aircraft flying roughly 100ft above the terrain.

I'm completely new to all of this, but I've been browsing this forum heavily and I've read all of the PhotoScan documentation/tutorial information I can get my hands on. Based on forum poses, I've deduced that the Hero GoPro series of cameras (which our fixed wing currently utilizes) are terrible for mapping with PhotoScan. Am I correct in this deduction?

With that said, we're looking into replacing the GoPro on our aircraft with a panasonic DMC-ZS20. I've seen the Panasonic GF series recommended on this forum once or twice, but cannot find any technical information to help me determine whether or not the DMC-ZS20 is applicable. Does anyone have any experience using this camera to generate overhead maps? If so, would you mind sharing your experiences and commenting on the quality of the end result? Also, this camera has embedded GPS, does anyone know if that GPS data is encoded into the EXIF data of the captured images?

I can provide more details about the area being mapped if necessary, but I'm more curious about this camera in general as a tool for image capture and mapping (e.g., if the camera is complete rubbish for this application, I'd rather find out now than after I've wasted time and money using it).

If this camera is not acceptable, can anyone provide some recommendations? I've already seen the Canon D10 and A2200 mentioned positively, as well as the Panasonic GF series. Are there any others I can consider?

Thanks!

[RalfH]

I haven't used the DMC-ZS20 myself, but some of the reviews don't look too promising (http://reviews.cnet.com/digital-cameras/panasonic-lumix-dmc-zs20/4505-6501_7-35133414.html). You're not planning to use it in video mode, are you? Because that would be a bad idea. If you are using it in photo mode, does it have an interval function, or how are you planning to trigger it? Is it possible to turn off the display to save battery?

I have been using Canon point-and-shoots for the simple reason that they are usually acceptable to very good quality and (more importantly) because I can run scripts on them (interval timer etc.). If you don't know CHDK (http://chdk.wikia.com/wiki/CHDK) yet, you should check it out. I've been using the Powershot A3000IS (very low-cost and light-weight, not very good quality) and G12 (somewhat pricy, heavier, much better quality). Just check out which camera supports CHDK and choose one which would work best for you in terms of weight, resolution, image quality, battery...  The D10 is perhaps not optimal in terms of image quality but would be great in terms of being dust-proof and not having its lens exposed to the elements (I should have used one of those when doing KAP over the sand dunes...). There now is a D20 which is a bit slimmer (but more expensive).

Regarding built-in GPS, I assume that GPS data would be saved in the EXIF. Also, I assume that using the built-in GPS will further cut into battery time. Depending on the maximum load on your plane it may be better to use a separate GPS logger and later synchronise GPS data and images. Still, normal GPS does not have the precision and accuracy required to correctly georeference your model; good ground control will be necessary.

P.S.: Using a fixed-wing plane to document small features on the ground means that you need fast shutter speeds: if you are flying at (guessing) 10 m/s, a shutter speed of 1/1000 second results in 10 mm of motion blur. I don't think that this would be acceptable for documenting archaeological excavations. Try this with realistic numbers for your plane and desired resolution. Shutter speeds faster that 1/1000 second are restricted to very good light or come at the price of having to increase ISO (and hence image noise). So, comparing cameras in terms of very good image quality also at ISO 400 or higher will have to be an issue. If your excavations are relatively small, PAP might be preferrable - my experience is that you can document on the order of 1000 m2 in 1-2 hours, with subsequent processing in PhotoScan taking much more time.

[hootener]

RalfH, thanks for your timely response.

Based on what I've read on this forum, Canons do seem like the way to go because of the convenience of CHDK. I was asking about the Panasonic because that camera had already been purchased for the task by a project manager (without the proper due diligence of course, of course  ...)

I will look more into the features you've asked about, interval timing, battery life, etc. However, I feel like the Panasonic camera will be limited compared to what I can do with, say, a D10 and the CHDK.

I tend to agree with you concerning GPS logging separately. I think I can cross reference time stamped GPS data from the plane with timestamps from the image capture and synchronize the image and GPS information post-hoc. Concerning ground control: Do you have any advice/sources you can point me to which discuss good ground control methods for this kind of work?

Like I said, this project unsuspectingly fell into my lap due to unforseen circumstances, so if you know of any reference material or resources that are essential to performing good orthomapping that you could point me to online, I would appreciate it. I've got about a six month lead time before the system will be deployed in the field. So I have time to make use of good learning material if it's available.

Thanks for all your help!

[RalfH]

I think the most important things about ground control are in the PhotoScan tutorials and have been touched upon in this forum. Basically, follow the same rules as as with ground control in any other remote sensing project:
- use enough GCPs (better have too many than too few)
- GCPs should be evenly distributed in the area (not just around the edges, not in a single line)
- make sure to measure GCP coordinates with high accuracy (use cm-resolution differential GPS or a total station)

Standard GPS will help you to know approximately where the picture was taken but is not good enough for georeferencing if you want to get better than a few metres. There are ways in which standard GPS data can be somewhat improved if you have a receiver that saves raw satellite data (using PPP and/or doppler correction), but that can be quite a hassle and still doesn't make it much better than a metre.