I have this new DMK51 camera that I have been using to capture comets with the ED80 telescope or a 26mm lens, lots of fun, see Panstarrs blog entries. The camera is connected to a tablet via USB and the other USB port is connect to a 1 Tb drive.
So, the concept here is big sky imaging using small equipment that can be easily setup on the car roof at a dark location. No need to do tracking just yet. Attaching, a small fish eye lens to the DMK has proved interesting. I can see a few satellites, and the occasional small meteor. Here's a good one on the edge of the frame from a few nights ago, Lyrid ?? may be.
The software that comes with the camera allows you to make 100 frame AVI's. I have to manually look through each avi to find the good frames with meteors, or satellites. I am quite lazy about this method, and wanted an auto triggered frame grab when a bright meteors happen. The processing.org software I have been using seemed promising again. It has video capture and play back functions. Also image processing functions. Could it do the job?
Well, yes as it turns out. Under windows 7 the camera is detected, and frames can be saved on a trigger event. So, I made a little program using the processing examples as a starter. Using a laser pointer as a meteor simulator. I can track the brightness of the light and see it at speed. It has a grid and a compass the rotates so that north setting can be made correct.
One of the tricks that makes this work is that the DMK camera remembers its last settings, so you can set the exposure time to get the best frame rate. I used an exposure of 1/6 second and gain and gamma set to 255. Then turn of the DMK Icapture software and let processing.org take over the camera.
I will try it out when the clouds clear away in the next few days. Hope it works......then i share the code, as i did with the radio meteor software. 8)
Wahooo.... its working , and can detect car lights pass along the edge of the FOV. This AVI has mostly clouds, so the next clear night will be the big test......
Here is the code for this version, DIY-MeteorCamera , copy the text into a new processing sketch and run.
You should see a list of available cameras. Make sure you can see your DMK camera. Change the code at this location,
// The camera can be initialized directly using an element
// from the array returned by list():
cam = new Capture(this, cameras[29]); // <<<<<
//29 name=The Imaging Source Europe GmbH DMK 51AU02.AS,size=1600x1200,fps=12
// set gamma = 255 set exposure = 1/6 sec set gain = 255
It should also work with built in cameras or most any that can be detected in the list, you will need to tweak some of the code to make it look good for your camera. Have fun..........8)
Upate 20130430 - Cars, Planes, but no meteor trains.....there's never a bright fireball when you want one....tracking bright aircraft ok... may have a big problem with a bright moon...hmm ..
Update 20130503 - While waiting for a meteor, i put the 26mm lens back on the camera and took a quick few frames of Crux, good depth for 1 sec frames.
Update 20130510 - No good meteors detected, so a few more tests reveal good satellite tracks.
Attached the camera to an old ETX90 mount I had lying around not being used. Now, I can track the imager all over the sky.
Update 20130524 - Added a bit of plumbing and gutted one of those imitation security camera shells to make an all-weather housing, don't know how water tight it will be, the plumbers out there will recognize the bits from the local hardware store, under $50 of bits.
Also changed the trigger method to frame to frame brightness difference that is more sensitive It captured this satellite passing over head, a comos I think....
Update 20130526 - I'll call this my proof of concept image. A small meteor captured over 4 frames with the trigger detection square indicting that it was a detected event.
 |
| small meteor detected near the pointers |
Update 20130622 - Had three clear nights to run a full twilight to dawn scan for events. Generating about 3.5Gb of images, as I take one image on every 55th second and any detected events. So just running for detected events should produce less data. Caught some good small examples and can get smaller spot meteors. Here is a good couple,
Update 20130704 - Generating more than 8GB of data as clouds roll across the sky, will need to work out a cloud detection method. A busy sky during the evening of 20130703, then morning of 20130704 clouds skirted by, with a few good detects mixed in, one very good longer duration event in the morning sky..
