Remote Sensing Tools

Two of the core value-added plug-ins that come with Remote Sensing Tools, allow for the ability to convert false-color composite imagery into pseudo natural-color imagery. Whether it’s Landsat imagery composited using the 7,4,2 bands as RGB, or CIR imagery, Quickbird 4,3,2 or Aster 3n,2,1 band composited imagery — the algorithms in these plug-ins will allow anyone to convert to pseudo natural-color within the Photoshop® image editing environment.

In this example, I’ll walk through a typical conversion to demonstrate the value of having these plug-ins at your disposal. Most imagery requires either a pre-processing step or post-processing steps in order to acheive the best results for your application. Each case varies, but this example will give you an idea of a typical case where some pre-processing, as well as some post-enhancements were needed to produce a good conversion-product.

Download the project file, which contains a CIR aerial image and a quick-guide readme.txt file with the steps listed here. The Geotiff metadata header has been stripped-out, so this is not a Geotiff image and will not be viewable in a geospatial environment. The image has also been saved as a .tif with lossless LZW compression, to preserve bandwidth without compromising image quality.:

CIR to Naturaltone Tutorial, Project Files (~2.6 MB download)

Converting the Image to Pseudo Natural-Color

Follow the steps below to convert the image from CIR to pseudo natural-color.

1. Levels Pre-Processing Adjustment — ‘Rectify’ luminence and detail. Click through Image → Ajustments → Levels to open the Levels dialogue. Place in the following settings:
   • RGB:
     • Input Levels: 0, 0.80, 255
     • Output Levels: 23, 255
   • Red Channel:
     • Input Levels: 16, 0.75, 255
     • Output Levels: 14, 255
   • Green Channel:
     • Input Levels: 10, 0.82, 255
     • Output Levels: 0, 255
   • Blue Channel:
     • Input Levels: 12, 0.71, 255
     • Output Levels: 0, 255
2. Now convert the image using the CIR to Naturaltone & Vegetation conversion plugin. Click through Filter → Remote Sensing Tools - Conversions → CIR to Naturaltone & Vegetation to open the plug-in dialogue and use the following setting:
   • Vegetation: 255
3. After the conversion, we’re going to enhance the image by ‘boosting’ its color using another plug-in from Remote Sensing Tools — Colour Boost. Simply click through the following to run the filter: Filter → Remote Sensing Tools - Enhancement → Colour Boost 1.00
4. For the final step, we’ll make a hue/sat adjustment to specific colors, globally. Click through Image → Adjustments → Hue/Sat and use the following settings:
   • Greens: -50 Saturation, -25 Darkness

You should now have a converted image that has the richness of a satellite image, and has a very natural look suitable for visualization and cartographic applications. Additional editing can be done to enhance specific details or to ‘fix’ anomolies present for a final image product.

Breaking down the approach. What did we do?

Step 1, Levels (Histogram) Adjustment

Manually ‘rectifies’ the image by adjusting its luminosity values in each channel. The settings try to ‘correct’ for processing or lack of detail present in the source image. This helps to generate a more ‘rich’ image, when additional processing methods are applied.

Step 2, Conversion

Simply converting the image to a pseudo natural-color image. I set the slider to its maximum due to understanding a typical end-use for the imagery, where vegetation is often wanted to appear like the middle of the summer season. The goal here is to leverage the NDVI (Vegetation) calculation, in order to add as much vegetation as possible so that a more ‘work-able’ or visually appealing image can be acheived.

Step 3, Boost Image

As you probably noticed, the first two steps left the image appearing ‘dead’ and unlike what you’d see on ground-level. The color details are inherently present, however — and ‘boosting’ their values helps to bring out, or ‘enhance’, what already exists. Due to the calculation being based on intensity values, this is a more accurate method than simply using Photoshop’s base Saturation slider in Hue/Sat.

Step 4, Hue/Saturation Adjustment

I manually re-adjusted the greens to bring them back to a ‘natural’ level. Generally, vegetation is tricky to process in this type of imagery and produce consistant results everytime. This is why I took a more ‘pro-active’ approach to processing the image and ‘injecting’ as much vegetation information as possible — so that this later finishing step could be performed.

ADDITIONAL NOTES

You can also record your steps as an Action in Photoshop to automate the procedure for your entire set of images. I suggest this not only for efficiency, but primarily so you have consistent imagery across your entire data-set — every image having been processed the same, eliminating any potential for ‘mistakes’ along the way. You can also run the Action as a Batch process through the function provided in Photoshop.

From there, you can make selective adjustments to individual tiles, to touch-up any areas you’re dissatisfied with. Or, you can load up the original image alongside the processed one — run through slightly different steps until the part of the image you want looks good — then copy/paste a selection into the image to finalize.

If you’re using Geographic Imager — I do not suggest running an Action or Script, since Geographic Imager does not support Actions/Scripts. This isn’t Avenza’s fault, actually — it’s due to Adobe Photoshop’s API limitations in their development kit. In this event, you’ll be forced to run each step on every image you load.