The View Up Here

Random scribblings about kites, photography, machining, and anything else

  • Flickr Gallery

  • Advertisements

Posts Tagged ‘ImageJ’

More Fun With False Color

Posted by Tom Benedict on 06/07/2009

Most of the time I’m doing kite aerial photography, it is to create a picture I like; that I would want to hang on my wall.  But on occasion I’ve done KAP strictly as a method of remote sensing.  The pictures are data rather than an attempt at art.  I’ve flown at Kiholo Bay several times now, usually in the former category of creating pleasing images.  This most recent trip was strictly in the latter sense: to take data.  But any time I’m out in the field doing KAP, I try to have fun and to stretch what I can do with what I’ve got.  This outing was no exception.

The requirements for the flight strictly called for oblique angles.  I settled on 45 degrees for most of them, though occasionally I used a steeper or a shallower angle as best suited the subject.  But I took some ortho images anyway.  It’s always fun to compare against what Google Earth shows for a particular region, and ortho can be useful when I’m going through the pictures later to judge distance and height.  The field of view of my camera is such that the horizontal width of the frame corresponds to the height of the camera to within 5%.  By taking the occasional orthogonal image, I can typically go back through the images with a map or Google Earth to figure out how high the camera was.  Besides, it’s always fun to compare the KAP image to the Google Earth image, if for no other reason than to point out that comparing satellite imagery and low altitude aerial imagery is seriously comparing apples and oranges.

Kiholo Bay - Google Earth Imagery

Ever since doing some work for an archaeologist from Oahu, I’ve been trying various false color techniques to try to get information out of my images.  Most of the time this has been done with orthogonal pictures, though as you can see from my previous post, the same tricks can be used on practically any picture.  But it works well with the orthos:

Kiholo Bay Ortho Composite

There are a number of features in this picture that are worth trying to isolate.  There are kiawe trees, a patch of fountain grass, pahoehoe lava from a Mauna Loa lava flow, aa lava from a Hualalai lava flow, water, coral rock, and even a swimmer in the water.

Kiholo Bay Ortho Composite - False Color

The first approach I tried was to use ImageJ with the DStretch plug-in.  It’s a really good false color filter that can be used to boost any manner of color combinations in an image.  In this case I used the YDS setting with a 315 degree rotation in hue, and managed to isolate most of the features listed above:  Kiawe trees are rendered as a combination of yellowish green in the upper branches, and red for the dead undergrowth.  The fountain grass is rendered as a bright red.  Pahoehoe lava is rendered as a blackish purple, and the aa is rendered as a reddish purple, as are the cracks in the pahoehoe lava.  Coral shows up as  a very light blueish purple, almost white, and the water is rendered as antifreeze green.  Lurid though the colors may be, it makes picking out individal elements in the image quite easy.

Another approach that has worked well in the past is to separate the R, G, and B channels in the image, and treat them as components in a mathematical expression.  This is a tried and true technique that’s been used in astronomy for ages.  B-V (or in the RGB world, B-G) images can be used to judge the temperature of a star, for example.  I wasn’t taking pictures of stars, but there’s still validity in the idea.  Let’s say I want to find the redder aa lava, but don’t want to get a false positive from the coral rock in the frame.  Subtracting the blue channel from the red channel picks up primarily red objects, in this case aa lava and cracks in the darker pahoehoe lava, though this catches the green vegetation as well:

Kiholo Bay Ortho Composite - Red minus Blue

Similarly, green vegetation can be isolated by subtracting red or blue from the green component, and in this case it doesn’t do such a good job of picking up the rocks:

Kiholo Bay Ortho Composite - Green minus Blue

Slightly more complicated expressions can be used to isolate other colors in the image, or to further separate colors.  And likewise, this trick can be used on false color images that have been produced through some other tool, such as DStretch.

Despite having resources like Google Earth available to us, kite aerial photography still offers a great deal of utility as a remote sensing platform.  But I still like making pretty pictures best of all.  The next time I grab my bag and head out the door, it’ll be to go somewhere nice and take pictures I like: the kind I want to hang on my wall.

— Tom

Advertisements

Posted in Hawaii, Kite Aerial Photography, Photography, Software | Tagged: , , , , , , | Leave a Comment »

Stereo Aerial Photography and False Color

Posted by Tom Benedict on 26/06/2009

It’s always fun to see what can be done with aerial photography: large stitched panoramas, Photosynths, 3D modeling, 2D maps.  One technique that KAP lends itself to particularly well is stereo aerial photography.

With an airplane, stereo photography is typically done by pointing the camera 90 degrees to the direction of flight, and taking a succession of pictures as the airplane flies across the landscape.  Careful choice of frame rate, airspeed, and altitude yields good results.

With a kite, a similar technique can be used:  Point the camera 90 degrees to the kite line, start the shutter going, and carefully walk backwards.  The kite will quickly settle into a stable flight angle with the increased apparent wind speed from the walking, and a nice steady stream of pictures is the result.  This is an example from a recent flight over the contact between two lava flows on the Big Island of Hawaii:

Lava Stereo Pair - True and False Color

The top two pictures are the natural color images as they came off the camera.  The bottom two require more explanation:

Another technique I’ve used with aerial photography is to apply false color techniques to boost certain details, certain colors, or to change the contrast of the image so that particular features will catch the eye.  Most of my experimentation along these lines has been done using an image manipulation program called ImageJ, and a plug-in called DStretch.  ImageJ is a general purpose image manipulation program written in Java.  DStretch is an implementation of the principal component algorithm for contrast stretching that was written by Jon Harman.  It was originally written for bringing up faint details in pictographs, but it has proven to be useful with aerial imagery as well.

In order to get a sense of scale of the image, the thin yellow line in the top pair is a three-section 25′ painter’s pole thath as been collapsed to its shortest length.  My flying partner and I were using it to lower a camera into one of the holes we found in the lava in order to explore the inside.

This probably isn’t the best example of either technique, but it’s the first time I used them in combination.  At some point I’ll write a more in-depth article about building stereo pairs, and a second article about the use of ImageJ and DStretch with aerial photography.  In the meanwhile, enjoy.

Tom

Posted in Kite Aerial Photography, Photography | Tagged: , , , , , , , , | 1 Comment »