false color imaging

lalalandrus

2

false color imaging

« on: 24 / February / 2016, 04:14:05 »

Advertisements

I am trying to do some false color imaging by combining the UV and the IR. my proposal is to take two pictures and have the camera combine them on the fly.

This will be done by taking one IR picture in greyscale and assigning it a color say blue and taking one UV picture again in greyscale and assigning it another color say red. Then merge the two pictures into one. Is this possible?

Logged

reyalp

14000

Re: false color imaging

« Reply #1 on: 24 / February / 2016, 13:20:28 »

Quote from: lalalandrus on 24 / February / 2016, 04:14:05

I am trying to do some false color imaging by combining the UV and the IR.

I may be mistaken, but AFAIK these cameras are not very sensitive to UV, and unlike IR there isn't a separate UV filter you can remove.

Quote

my proposal is to take two pictures and have the camera combine them on the fly.

This will be done by taking one IR picture in greyscale and assigning it a color say blue and taking one UV picture again in greyscale and assigning it another color say red. Then merge the two pictures into one.

It is possible to process raw images on the camera (see modules/raw_merge.c for example), or read/modify raw data before it is written (see http://chdk.wikia.com/wiki/Lua/Raw_Hook_Operations) but it's quite slow. The memory available to CHDK is also much less than a full raw image, which makes some things complicated. There is currently no code to load or manipulate jpeg image, and no one has figured out how to program the digic image processing engine directly.

Logged

Don't forget what the H stands for.

Hardware_Hacker

322

Re: Super Blue Infrared Filter and false color imaging

« Reply #2 on: 24 / February / 2016, 18:49:30 »

Quote from: lalalandrus on 24 / February / 2016, 04:14:05

"...I am trying to do some false color imaging by combining the UV and the IR..."

Is this possible? = "Yes, see below"

Search Google for "Super Blue Infrared Filter"
Also Google Images for the results of "Super Blue Infrared Filter"

Ref: http://www.lifepixel.com/infrared-photography-primer/ch4-internal-filters-for-modified-cameras-super-blue-infrared-filter

You should also note that the Canon HS series ("BSI") might have a better UV response.

This is probably a very expensive, but it's also a "One Shot" "Option". [{(

)}]

Another Option is the, Canon EOS M3 and M10 Cameras, which might also have "BSI imaging Sensors".

H-H

« Last Edit: 24 / February / 2016, 19:01:47 by Hardware_Hacker »

Logged

lalalandrus

2

Re: false color imaging

« Reply #3 on: 26 / February / 2016, 13:48:48 »

thanks for the info

1) looks like the raw hooks only process pixels at a time, it is unfortunate that we are unable to access the digic processor

2) yes i have some sort of filter to let in only UV, my main issue is the image processing itself however. the camera still has the IR filter in so the UV insensitivity + the IR filter works out quite closely. i just need to use really long shutter speeds but i am shooting still objects so this is not a problem.

if anyone has some genius way to do the combination (without pixel by pixel manipulation) on camera this would be great as my project prevents me from accessing a computer for long periods of time and a preview of the results would be great.

Logged

reyalp

14000

Re: false color imaging

« Reply #4 on: 26 / February / 2016, 16:42:21 »

Quote from: lalalandrus on 26 / February / 2016, 13:48:48

if anyone has some genius way to do the combination (without pixel by pixel manipulation) on camera this would be great as my project prevents me from accessing a computer for long periods of time and a preview of the results would be great.

There is currently no way to do it other than pixel by pixel in raw data, either as a batch process like raw average, or using raw hooks or a C module like curves. Figuring out whether it is even theoretically possible to use the digic image processing engine for this would require a lot of reverse engineering.

Logged

Don't forget what the H stands for.