PixelSense 742
Point and click processing of Landsat 742 imagery into ready-to-use high-quality natural color GeoTIFF images.
PixelSense 742 is the most efficient and cost-effective way to turn Landsat 742 remote-sensed images into natural-color georeferenced TIFF images for use in any mainstream GIS or visualization software. Simply feed the input Landsat 742 image into PixelSense 742 and in a very short time, you will have a detailed and realistic output image. No algorithms to mess with. No settings to tweak. No World Files to detatch and re-attach. No multi-thousand-dollar GIS programs to buy. No Photoshop plugins needed. No channel math. No hand-tuning Levels and Curves.
Fast, easy, Inexpensive.
In-Depth Explanation:
What is a Landsat 742 image? Simply, it is an image using band 7 (Mid-Infrared), band 4 (Near Infrared) and band 2 (Green) as the Red, Green and Blue channels. Images of this type contain no actual Red or Blue data, making it very difficult to make them look the 'natural' way a camera or human eye would see them. This 742 band combination is used in the well-known GeoCover mosaicked dataset.
The challenge for PixelSense 742 is to fabricate reasonable new Red and Blue channels that will allow presentation of a natural-appearing image utilizing the remaining available data. PixelSense 742 calculates facsimile Red and Blue channels using the relationships between the other channels (including the two Infrared channels, 7 and 4).
Drawing on our PixelSense CIR experience (reconstructing the missing Blue band), we set out on a quest to discover the magic remixing ratios. In the end, we ended up with three algorithms: One hand-made experimentally, one "evolved" by endless trial and error by a genetic program and one master algorithm that decides which areas should be treated by which, and which areas should be left alone entirely.
The result is surprisingly similar to an actual RGB image, and very useful for visualization purposes. 742 imagery is often available in pre-made large-area mosaics (GeoCover) that are often conveniently available where normal Landsat 321 imagery is not.
Source Image and Output Image (approximately 800 pixels wide):
->
Detail Comparison (Source and Output):
 | |
| Left: 742 | Right: PixelSense Natural Color Output |
Validation (Output [left] versus true RGB image [right]):
|
 |
| PixelSense 742 Output | Actual 321 RGB image |
Validation summary: On the left is the natural-color image generated by PixelSense 742 from Landsat 742 Imagery. On the right is a Landsat 321 RGB made from different bands of the exact same Landsat scene (using PixelSense LS). Because they were imaged by Landsat 7 at exactly the same time, the images are very comparable. While the color cast is not identical, the PixelSense 742 output is very usable as a natural-color image. Some difference in red tints is to be expected, because the 742 image completely lacks a red band, and no suitable data exists in the 742 bands to reconstitute the red component..
Caveats: Not all 742 images are mixed the same (any more than all 3,2,1 images are mixed the same). You may find some images that have had exaggerated contrast boost, strange color skews, Pan merging (GeoCover) or the like. PixelSense was designed to take an undoctored 742 image and remix it to match as closely as possible, the equivalent 321 image made from the exact same Landsat scene. Your mileage may vary, depending on how your 742 image was produced.
See the Gallery for more examples of PixelSense 742 output.
The effort you save using PixelSense 742 the very first time, will pay for the program.
PixelSense 742 costs only USD $199 and can be purchased online. A Demo version is available. Order now!
Current PixelSense 742 Version
12/12/2005 Version 1.01 released.
- Fixed problem where GeoTIFF tags weren't always preserved in output file.