Orthorectification digital camera - Revision history

Neteler: +Orthorectification chapter from the GRASS GIS book

2014-05-10T16:29:10Z

+Orthorectification chapter from the GRASS GIS book

← Older revision		Revision as of 16:29, 10 May 2014
Line 254:		Line 254:
	</pre>		</pre>

	=== See also ===		== See also ==

			* M. Neteler, D. Grasso, I. Michelazzi, L. Miori, S. Merler, and C. Furlanello, 2005: An integrated toolbox for image registration, fusion and classification. International Journal of Geoinformatics, 1(1), pp. 51-61 [http://www.grassbook.org/neteler/papers/neteler2005_IJG_051-061_draft.pdf PDF]
			* Orthorectification chapter from the GRASS GIS book: http://www.grassbook.org/sample2nd/grassbook_2nd_2004_chapter10_aerial.pdf

	* M. Neteler, D. Grasso, I. Michelazzi, L. Miori, S. Merler, and C. Furlanello, 2005: An integrated toolbox for image registration, fusion and classification. International Journal of Geoinformatics, 1(1), pp. 51-61 [http://www.grassbook.org/neteler/papers/neteler2005_IJG_051-061_draft.pdf PDF]
	[[Category: Documentation]]		[[Category: Documentation]]
	[[Category: Image processing]]		[[Category: Image processing]]
	[[Category: Orthorectification]]		[[Category: Orthorectification]]

Neteler: /* See also */ +cat

2013-03-24T20:26:27Z

Neteler at 09:57, 17 May 2011

2011-05-17T09:57:49Z

← Older revision		Revision as of 09:57, 17 May 2011
Line 1:		Line 1:
	== Geocoding of photographs taken with digital handheld camera ==		== Geocoding of photographs taken with digital handheld camera ==

	To ~~geocoding~~ oblique photographs taken with a digital handheld camera,		To geocode '''oblique''' photographs taken with a digital handheld camera,
	we need a digital elevation model (DEM) and a topographic reference		we need a digital elevation model (DEM) and a topographic reference
	map. The elevation model is required to normalize the terrain		map. The elevation model is required to normalize the terrain
Line 17:		Line 17:
	system (like UTM or Gauss-Krüger).		system (like UTM or Gauss-Krüger).

	Metadata like time stamp, are taken from EXIF data (e.g. with "jhead",		Metadata like the time stamp or the focal length are taken from EXIF data (e.g. with "jhead",
	http://www.sentex.net/~mwandel/jhead/ ).		http://www.sentex.net/~mwandel/jhead/ ).

Line 254:		Line 254:
	</pre>		</pre>

			=== See also ===

			* M. Neteler, D. Grasso, I. Michelazzi, L. Miori, S. Merler, and C. Furlanello, 2005: An integrated toolbox for image registration, fusion and classification. International Journal of Geoinformatics, 1(1), pp. 51-61 [http://www.grassbook.org/neteler/papers/neteler2005_IJG_051-061_draft.pdf PDF]
	[[Category: Documentation]]		[[Category: Documentation]]
	[[Category: Orthorectification]]		[[Category: Orthorectification]]

Neteler: updated to 6.4.x

2011-05-17T09:12:39Z

updated to 6.4.x

← Older revision		Revision as of 09:12, 17 May 2011
Line 63:		Line 63:
	=== i.ortho.photo ===		=== i.ortho.photo ===

	Menu ~~4 Camera~~:		Menu 5. Compute image-to-photo transformation

			Some considerations:

	<pre>		<pre>
	Calibrated Focal Length mm.: guessing 50		Calibrated Focal Length mm.: guessing 50 for a digital handheld camera (airborne cameras have often a focal length of 150 mm)

	Fiducial marks (positive values, use g.region -p):		Fiducial marks (positive values, use g.region -p):
Line 124:		Line 126:
	Then use menu 5 in i.ortho.photo. It's a challenge to hit the center		Then use menu 5 in i.ortho.photo. It's a challenge to hit the center
	of the image borders... try smallest RMS		of the image borders... try smallest RMS
			</pre>

	~~###################~~

	~~Menu 6~~

			Menu 6. Initialize exposure station parameters
			<pre>
	Initial Camera Exposure X-coordinate Meters: 664194________		Initial Camera Exposure X-coordinate Meters: 664194________
	Initial Camera Exposure Y-coordinate Meters: 5104463_______		Initial Camera Exposure Y-coordinate Meters: 5104463_______
Line 135:		Line 137:
	Initial Camera Phi (pitch) degrees: 20____________		Initial Camera Phi (pitch) degrees: 20____________
	Initial Camera Kappa (yaw) degrees: -45___________		Initial Camera Kappa (yaw) degrees: -45___________
			</pre>

			Menu 7. Compute ortho-rectification parameters
			<pre>
			Use mouse now

			Search ground control points, minimize RMS (see Analyse menu entry) to
			ideally half a pixel.
			</pre>

	~~###################~~		Menu 8. Ortho-rectify imagery files
	~~Menu 7~~		<pre>
			Enter an extension to be appended to rectified maps:
			.ortho______________

	~~Use mouse now~~
			Compute local camera angle? (y/n) [n] y

			Enter a name for the camera angle map:
			22_1684.camera_angle__________


	~~###################~~		Overwrite maps in target location/mapset? (y/n) [n] y

	~~Menu 8:~~

			Please select one of the following options
	1. Use the current window in the target location		1. Use the current window in the target location
	2. Determine the smallest window which covers the image		2. Determine the smallest window which covers the image
	> > 2		> 2

			Enter desired target resolution, or
			RETURN to determine it automatically:
			> 2.0

			Please select one of the following interpolation methods
			1. nearest neighbor
			2. bilinear
			3. bicubic
			4. bilinear with fallback
			5. bicubic with fallback
			> 2


			Enter amount of memory to use in MB, or
			RETURN use 100 MB
			> 1000

	~~Done.~~
	</pre>		</pre>

			Now the orthophoto is generated and will be stored in the target location.

	=== Hints: How to calculate Fiducial marks, if no DPI known ===		=== Hints: How to calculate Fiducial marks, if no DPI known ===

Neteler: +cat

2009-05-25T18:56:07Z

+cat

← Older revision		Revision as of 18:56, 25 May 2009
Line 219:		Line 219:
	- average over 2 longitudinal and transversal oriented objects		- average over 2 longitudinal and transversal oriented objects
	</pre>		</pre>

			[[Category: Documentation]]
			[[Category: Orthorectification]]

Neteler: draft

2008-01-05T13:20:14Z

draft

New page

== Geocoding of photographs taken with digital handheld camera ==

To geocoding oblique photographs taken with a digital handheld camera,
we need a digital elevation model (DEM) and a topographic reference
map. The elevation model is required to normalize the terrain
undulation. The raster resolution of the DEM should be similar to the
resolution of the aerial photo. Usually such high-resolution DEMs
(raster cell length below 1m) are not easily available. Therefore,
interpolation of a given DEM to a higher resolution is recommended to minimize
displacement effects. The topographic reference map is needed to find
corresponding ground control points (GCPs) between aerial photo and
this map for geocoding. The map scale of the reference map should at
least match the average scale of the aerial photo (for example
1:5000), but should be probably larger to provide more details for
improved geocoding accuracy. The ground control points are required to
register the image xy coordinate system to a georeferenced coordinate
system (like UTM or Gauss-Krüger).

Metadata like time stamp, are taken from EXIF data (e.g. with "jhead",
http://www.sentex.net/~mwandel/jhead/ ).

== PROCEDURE ==

=== Ortho-rectification of oblique photographs ===

In projected location:
<pre>

- get approximate camera position:

d.where/d.what.rast:
EAST: NORTH:
664194.71461534 5104463.03343466

height: 450

Pitch: 20
Roll: 0
Yaw (against north): -45
</pre>

See also: http://en.wikipedia.org/wiki/Pitch_%28flight%29

<pre>
######################################

g.region rast=cf83 -p
projection: 0 (x,y)
zone: 0
north: 798
south: 0
west: 0
east: 1200
nsres: 1
ewres: 1
rows: 798
cols: 1200

-> img size in Pixel (NOTE: newer cameras provide this in exif metadata!)

</pre>

=== i.ortho.photo ===

Menu 4 Camera:

<pre>
Calibrated Focal Length mm.: guessing 50

Fiducial marks (positive values, use g.region -p):

y
| 2
+--------+--------+
| |
| |
| |
1 + + 3
| |
| |
| |
+--------+--------+------> x
(0,0) 4

also: xmax = 1200/2 = 600
ymax = 798/2 = 399

Digital cameras have a resolution
Fiducial marks have to be given in millimeters (so, to calculate)
and not in pixel:

first calc. DPI, then derive "pseudo''millimeters from that
for Fiducial marks:

E.g.: Lenght of an object (eg bridge) in pixel (use d.what.rast + Pytaghoras)
compared to length of bridge in meters on reference map (d.measure).

eg 226 pixel versus 75 meter in UTM map

[ DPI = pixel/inch = pixel / 2.54cm ]

-> 226/7500 * 2.54cm = 0.07653867 DPI (check)

Since we calculate in *millimeter* and not DPI, we calc. for Fiducial marks:
226/7500 * 0.1 = 0.003013333

Multiply all pixel with this value:

1200 * 0.003013333 = 3.616000
1200 * 0.003013333 /2 = 1.808000
798 * 0.003013333 = 2.404640
798 * 0.003013333 /2 = 1.202320

Ergo:
Fiducial marks:
1 1_____ 0_________ 1.2______
2 2_____ 1.8______ 2.4______
3 3_____ 3.61_____ 1.2______
4 4_____ 3.61______ 0_________

(be a bit more precise...)

##################
Then use menu 5 in i.ortho.photo. It's a challenge to hit the center
of the image borders... try smallest RMS

###################

Menu 6

Initial Camera Exposure X-coordinate Meters: 664194________
Initial Camera Exposure Y-coordinate Meters: 5104463_______
Initial Camera Exposure Z-coordinate Meters: 200___________
Initial Camera Omega (roll) degrees: 0_____________
Initial Camera Phi (pitch) degrees: 20____________
Initial Camera Kappa (yaw) degrees: -45___________

###################
Menu 7

Use mouse now

###################

Menu 8:

1. Use the current window in the target location
2. Determine the smallest window which covers the image
> > 2

Done.
</pre>

=== Hints: How to calculate Fiducial marks, if no DPI known ===

<pre>

- Pixel extent: g.region rast=oblique

y
| 2
+--------+--------+
| |
| |
| |
1 + + 3
| |
| |
| |
+--------+--------+------> x
(0,0) 4

e.g.
xmax = 1200/2 = 600
ymax = 798/2 = 399
</pre>

Important: while digital handheld cameras provide DPI and chip size,
one has to estimate everything if it is unknown.
Fiducial marks have to be given in millimeters, not in pixel!

<pre>
Calculation of Pseudomillimeter:

- measure in XY location length of an object in pixel (with d.what.rast +
Pytaghoras)
- measure in UTM location length of same object in reference map (d.measure)
in meters

Example: bridge 226 pixel versus 75 meter in UTM.

[ DPI = pixel/inch = pixel / 2.54cm ]

-> 226/7500 * 2.54cm = 0.07653867 DPI

BUT: We want *millimeter* and not DPI for the fiducial marks:
226/7500 * 0.1 = 0.003013333

Finally multiply pixels with this factor:

1200 * 0.003013333 = 3.616000
1200 * 0.003013333 /2 = 1.808000
798 * 0.003013333 = 2.404640
798 * 0.003013333 /2 = 1.202320

Ergo:

Fiducial marks:
1 1_____ 0_________ 1.2______
2 2_____ 1.8______ 2.4______
3 3_____ 3.61_____ 1.2______
4 4_____ 3.61______ 0_________

Suggestions:
- take long objects,
- average over 2 longitudinal and transversal oriented objects
</pre>

← Older revision		Revision as of 20:26, 24 March 2013
Line 258:		Line 258:
	* M. Neteler, D. Grasso, I. Michelazzi, L. Miori, S. Merler, and C. Furlanello, 2005: An integrated toolbox for image registration, fusion and classification. International Journal of Geoinformatics, 1(1), pp. 51-61 [http://www.grassbook.org/neteler/papers/neteler2005_IJG_051-061_draft.pdf PDF]		* M. Neteler, D. Grasso, I. Michelazzi, L. Miori, S. Merler, and C. Furlanello, 2005: An integrated toolbox for image registration, fusion and classification. International Journal of Geoinformatics, 1(1), pp. 51-61 [http://www.grassbook.org/neteler/papers/neteler2005_IJG_051-061_draft.pdf PDF]
	[[Category: Documentation]]		[[Category: Documentation]]
			[[Category: Image processing]]
	[[Category: Orthorectification]]		[[Category: Orthorectification]]