Controlling for different lighting in two medical images of the same subject

[badgettrg]

Does anyone have suggestions on how to use existing PainNet program or plugin features to solve the following problem?

Sometimes in publishing of medical research, we need to show the changes in a patient's appearance over time. Inevitably, the two photos will be in different lighting. For example, the current photo may be in the flourescent light of a hospital while the older photo may be in the incandescent light of a patient's home.

I have tried a method to identify the darkest pixel in a feature common to both pictures (for example, eyebrows) with the following results:

New photo: R:48 G:34 B:30

Old photo: R:45 G:43 B:40

So, how should I adjust the old photo? Simply add either the relative or absolute difference between the channels above to the entire photo? Is there a plugin that supports this? I looked at ColorMatch and ScriptLab, but did not see a solution.

Thanks

[pdnnoob]

Wrong section to post this in (shoulda gone in general discussion)

Anyhow, there are a few things you could try:

1. whitepoint correction

2. color balance

3. curves+ http://forums.getpaint.net/index.php?/topic/7291-

[Ego Eram Reputo]

<moved>

[delpart]

I think I know what you are describing, but I'm not sure if the darkest common is really where you would focus. Skin tone is more likely to change from a source light difference. Could you post some samples (clipped even to avoid issues of propriety and idenity of course) to better "see" what you are running into?

The stated example is pretty common correction for image blending, but the direction of the source light and a ton of other things will affect how much of a change to the primary subject is happening. Whitebalance, cross-processing and few others might help, but it depends on how much color difference there is. At least for me, fluorescent hues and incandescent hues tend to vary too much for there to be "one method" really.

Not trying to nitpick it to death either or be dismissive. Examples would help though to probably get more engagement as well.

[badgettrg]

Thanks for the ideas. Due to privacy issues, I cannot put the complete photos in public. However, a link to extracts of the photos and my initial, primitive test with color balances and brightness is at:

http://sumsearch.org/photos.shtml

The values for the darkest pixel of the eyebrow are:

• Before photo: 43 30 21
• Current photo: 48 43 43
• Before photo with adjustments to balance and brightness: 47 40 41

Thanks

I think I know what you are describing, but I'm not sure if the darkest common is really where you would focus. Skin tone is more likely to change from a source light difference. Could you post some samples (clipped even to avoid issues of propriety and idenity of course) to better "see" what you are running into?

The stated example is pretty common correction for image blending, but the direction of the source light and a ton of other things will affect how much of a change to the primary subject is happening. Whitebalance, cross-processing and few others might help, but it depends on how much color difference there is. At least for me, fluorescent hues and incandescent hues tend to vary too much for there to be "one method" really.

Not trying to nitpick it to death either or be dismissive. Examples would help though to probably get more engagement as well.

[delpart]

I'm afraid I lack the knowledge of how to correct for something like that after spending some time trying what I thought might work (most of which is comprised off your efforts and the tools listed by Noob.)

The samples should help though. Thanks for posting those.

Best regards,

[badgettrg]

The examples I used are at http://sumsearch.org/photos.shtml . The photos are only a segment of the face to protect the patient's privacy. In addition, the server with the photos is having some trouble and at times the page is not display today. Any help is appreciated. Thanks.

Not trying to nitpick it to death either or be dismissive. Examples would help though to probably get more engagement as well.