GRASS GIS manual: i.fusion.hpf

NAME

i.fusion.hpf - Fusing high resolution panchromatic and low resolution multi-spectral data based on the High-Pass Filter Addition technique (Gangkofner, 2008).

KEYWORDS

imagery, fusion, sharpening, high pass filter, HPFA

SYNOPSIS

i.fusion.hpf

i.fusion.hpf --help

i.fusion.hpf [-l2c] pan=filename msx=filename(s)[,filename(s),...] suffix=suffix string [ratio=rational number] [center=string] [center2=string] [modulation=string] [modulation2=string] [trim=rational number] [--overwrite] [--help] [--verbose] [--quiet] [--ui]

Flags:

-l: Linearly match histogram of Pan-sharpened output to Multi-Spectral input
-2: 2-Pass Processing (recommended) for large resolution ratio (>=5.5)
-c: Match color table of Pan-Sharpened output to Multi-Spectral input
--overwrite: Allow output files to overwrite existing files
--help: Print usage summary
--verbose: Verbose module output
--quiet: Quiet module output
--ui: Force launching GUI dialog

Parameters:

pan=filename [required]: High resolution Panchromatic image
msx=filename(s)[,filename(s),...] [required]: Low resolution Multi-Spectral image(s)
suffix=suffix string [required]: Suffix for output image(s); Names of Pan-Sharpened image(s) will end with this suffix; Default: .hpf
ratio=rational number: Custom ratio; Custom ratio overriding standard calculation; Options: 1.0-10.0
center=string: Center cell value; Center cell value of the High-Pass-Filter; Options: low, mid, high; Default: low
center2=string: 2nd Pass center cell value; Center cell value for the second High-Pass-Filter (use -2 flag); Options: low, mid, high; Default: low
modulation=string: Modulation level; Modulation level weighting the HPF image determining crispness; Options: min, mid, max; Default: mid
modulation2=string: 2nd Pass modulation level (use -2 flag); Modulation level weighting the second HPF image determining crispness (use -2 flag); Options: min, mid, max; Default: mid; min: Minimum: 0.25; mid: Mid: 0.35; max: Maximum: 0.5
trim=rational number: Trimming factor; Trim output border pixels by a factor of the pixel size of the low resolution image. A factor of 1.0 may suffice.

DESCRIPTION
- Background
NOTES
EXAMPLE
TODO
REFERENCES
SEE ALSO
AUTHORS

DESCRIPTION

i.fusion.hpf is an implementation of the High Pass Filter Additive (HPFA) Fusion Technique. It combines high-resolution panchromatic data with lower resolution multispectral data, resulting in an output with both excellent detail and a realistic representation of original multispectral scene colors. The process involves a convolution using a High Pass Filter (HPF) on the high resolution data, then combining this with the lower resolution multispectral data. Optionally, a linear histogram matching technique is performed in a way that matches the resulting Pan-Sharpened image to the statistical mean and standard deviation of the original multi-spectral image.

Background

Computing ratio of low (Multi-Spectral) to high (Panchromatic) resolutions
High Pass Filtering the Panchromatic Image
Resampling MSX image to the higher resolution
Adding weighted High-Pass-Filetred image to the upsampled MSX image
Optionally, matching histogram of Pansharpened image to the one of the original MSX image

Figure:
 ____________________________________________________________________________
+                                                                            +
| Pan Img ->  High Pass Filter  ->  HP Img                                   |
|                                      |                                     |
|                                      v                                     |
| MSx Img ->  Weighting Factors ->  Weighted HP Img                          |
|       |                              |                                     |
|       |                              v                                     |
|       +------------------------>  Addition to MSx Img  =>  Fused MSx Image |
+____________________________________________________________________________+

Source: Gangkofner, 2008

NOTES

Grasping and testing the various parameters that define the High-Pass filter's kernel size and center value is a matter of short time.
Works with any number and type of raster imagery (8-bit, 16-bit)
The "black border" effect, possibly caused due to a non-perfect match of the high vs. the low resolution of the input images, can be trimmed out by using the trim option --a floating point "trimming factor" with which to multiply the pixel size of the low resolution image-- and shrink the extent of the output image.

EXAMPLE

The module is fairly easy to use. Arbitrary examples: Pansharpening of one band:

i.fusion.hpf pan=Panchromatic msx=Red

Pansharpening of multiple bands:

i.fusion.hpf pan=Panchromatic msx=Red,Green,Blue,NIR

Various illustrated examples detailed in the document i.fusion.hpf, implementation of the High Pass Filter Additive (HPFA) Image Fusion Technique (PDF)

TODO

Go through Submitting Python
Access input raster by row I/O
Support for parallel processing
Proper command history tracking.
Add timestamps (r.timestamp, temporal framework)
Deduplicate code where applicable
Make verbose level messages shorter, yet more informative (ie report center cell)
Test if it compiles in other systems
Check options to integrate in i.pansharpen. Think of FFM methods vs. Others?
Improve Documentation.lyx

REFERENCES

Gangkofner, U. G., Pradhan, P. S., and Holcomb, D. W. (2008). Optimizing the high-pass filter addition technique for image fusion. PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 74(9):1107--1118.
"ERDAS IMAGINE." Accessed March 19, 2015. ERDAS IMAGINE Help.
Aniruddha Ghosh & P.K. Joshi (2013) Assessment of pan-sharpened very high-resolution WorldView-2 images, International Journal of Remote Sensing, 34:23, 8336-8359

AUTHORS

Nikos Alexandris
Panagiotis Mavrogiorgos

Last changed: $Date: 2018-08-22 09:41:47 +0200 (Wed, 22 Aug 2018) $

SOURCE CODE

Available at: i.fusion.hpf source code (history)