Building a UAV for photo mapping - Camera Selection

I have bought a Pentax Optio A30 to use as my aerial camera. It is similar to a number of the latest generation of sub-compact 10Mpixel cameras and has good reviews, especially in terms of its image quality. It weighs about 160gm with battery and SD memory card and at 58mm, has one of the smallest widths I have found – important for fitting into the Cularis body.


The highest resolution image is quoted as 3648 x 2736 pixels. Based on the handbook, a 4Gbyte SD memory card at the best quality jpeg setting, should give about 1200 images. The camera has a continuous mode where images are taken and downloaded to the SD card as fast as possible as long as the shutter is pressed. One option for using the camera in the UAV, is to set it running in continuous mode and then select 1 in n of the images produced to make the photo-composite.

With a fully charged battery, I tested the camera in continuous mode while photographing a clock. The camera managed to capture 1613 images at full resolution in high quality mode in 23 minutes before filling the 4Gbyte SD card. The battery indicator showed about half full at the end of the run. Although the average interval was about 1.2 seconds, there were some gaps between images of up to 8 seconds and in other cases, 2 sequential images showed the same time. Both results suggest that the timing is not constant, but is a function of the amount of data that needs to be stored on the card. The review also states that the file size varies a lot depending upon the image content, so this make matters worse.


I then tried taking photos at timed intervals. I could get down to about a 4 second interval with only an occasional missed image, although this may have been finger trouble. So if I want images taken at regular, known intervals, I will have to use a servo driven IR interface such as the PRISM.


Now to look at flying the camera. There are two limiting scenarios.


  1. Fly at 120m (400ft) altitude. The ground area covered with the A30 lens set to wide angle (40mm equivalent in 35mm film terms) would be 108m x 72m. This gives a pixel resolution of about 3cm per pixel or 3 times the target of 12cm per pixel. Assuming a 30% overlap between images in the direction of flight, then the distance between images is 75m and the maximum speed of the plane taking an image every 4 seconds is 19m/sec or about 68kph (about 40mph).
  2. Fly at 480m (1600ft) Now the ground area is 432m by 288m which gives the required 12cm resolution. The distance between images to give a 30% overlap would now be 302m

Comments and Questions.


  1. Assuming a 20-30min flight, the memory size and battery power for the camera are not limiting factors.
  2. For regular images at known times and or positions, the camera will have to be fitted with an IR trigger.
  3. What should I aim for as a cruising speed for my Multiplex Cularis as this will impact the number of photos taken during each flight?
  4. Is it better to fly lower and use the additional resolution in post processing of the images to end up with corrected 12cm resolution map?


Building a UAV for phot mapping - Previous Posts

  1. The Goal
  2. Camera Selection
  3. Packing it all in

Views: 4336

Comment by Jeffrey Johnson on December 21, 2007 at 10:34am
Paul, You may want to check into something like a URBI interface http://blip.com.au/item.aspx?itemID=14 to control the shutter or physical triggering with a servo, but in general, it is easier just to leave the camera in continuous mode and filter the resulting images for quality in post processing.

Please keep us all up-to-date with your progress, and we (Pict'Earth) have some software for you to beta test as soon as you get in the air.

Also, we will should be making a blog post during the holidays about the Cularis setup that we use, and this may be helpful as you begin to build.

Have fun!

Jeff
Comment by James Ross on December 21, 2007 at 11:32am
The formula for calculating any one value, given the others would be something like this:
Altitude*sin(VFOV*(1-overlap)) = Speed/CamFrequency

The cruising speed will just depend on the other values in that formula and the performance capabilities of your UAV.

So if Altitude = 100m, VFOV = 50º, overlap = 0.5, and CamFrequency = 0.25 Hz, you'd need a speed of 10.6 meters/second

I imagine you'd want to fly high enough so that changes in terrain elevation won't require you to scale your images as much. The area you're photographing looks mountainous.

I've worked on code that projects digital images onto a terrain map. The USGS has all the elevation data over in this part of the world...I'm not sure about Italy
Comment by Paul Gregory on December 21, 2007 at 1:17pm
Matt

I haven't quite got round to hacking the camera apart yet - that's a bit of lateral thinking I may need in the future.

Currently I am thinking of putting the camera in a fixed position inside the foam body of the Colaris without any stabilized mount and rely on the autopilot to keep the plane stable.

In my next post, I plan to look at the issues of camera mounting, weight and COG.

Paul

3D Robotics
Comment by Chris Anderson on December 21, 2007 at 5:33pm
Paul,

Two things:

1) As the moderator, I'd say you're spot on. Your posts are just the right length and you might just want to put links at the bottom to the previous posts so people coming from a Google search can come up to speed. If you do indeed suceed in making it a proper UAV with an autopilot etc, I'll happily feature a overview post on it on the front page, with my GeoCrawler posts.

2) As for hacking the A30, there's really no need. We recommended that camera because it has a good IR interface, and you've got the PRISM so you can access that nicely. The Cularis can handle the weight of the the LCD, etc, fine. The only thing is how to trigger the PRISM to fire every 4 seconds, or whatever, which might require programming another board, but frankly I think you're better off just putting it into continuous mode and letting the post-processing software (like Pict'Earth) sort it out, assuming you're also carrying a GPS logger like the i-Blue 747.

The project sounds better and better!

Chris
Comment by problah on January 8, 2008 at 2:57pm
Paul,

I agree with chris on setting continuous mode. I'm sure you will be making multiple runs, so as long as you grid out your nav points, and keep them constant, you should be able to overlay images captured in one run to the master image map that has those sections missing. It will probably take some hemming an hawing to get it lined up, but that's what makes it even more fun! 0]

I would really love to see video of this bird, as well as video from it's eye looking down. This is such a great idea! Good luck!
Comment by James Harwood on October 8, 2009 at 12:50am
what happened with this project? I am in the process of doing the same thing - still trying to decide on the platform to use. How did yours turn out? I have seen others with good video from the cularis, but I was just wondering why you discontinued updating here.
Comment by Paul Gregory on October 8, 2009 at 8:50am
Hi

I am afraid I got sidetracked into trying to build an autopilot - a great danger. The Cularis is neatly packed away ready for another period of spare time - which never seems to happen. Maybe next year.

Sorry I cannot be of any further help at the moment.

PJG
Comment by Pedro on January 22, 2015 at 10:58am

Hi,

Sorry but how can i know the aerea covered on the ground by the camera if im flying at 90 meters above the ground i´m guessing that is has to do with the lenses of the camera.

Thanks!

Comment

You need to be a member of DIY Drones to add comments!

Join DIY Drones

© 2019   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service