Symposium International de Cartographie Thématique Dérivée des Images Satellitaires, Saint-Mandé 2-4 octobre 1990 0 Conception cartographique et intégration dans de nouveaux documents
Cartographie design and integration into new documents
SATELLITE - AND TH EMATIC ORTHOPHOTO - MAPS BY MEANS OF IMAGE OPTIMIZATION
AND DIGITAL CARTOG RAPHY
par Gerhard SCHWEINFURTH
Institut für Photogrammetrie und Fernerkundung, Universitat Karlsruhe
Abstract
Combinations of image data and cartographie information are most/y unsatisfactory and do not utilize the full spectrum of spectral image contents and avai/ab/e tools.
This lack is removed by a new strategy, based on an association of two different methods of image processing: Image Optimization and Digital Cartography. This combination offers optimum results by sole/y digital process and is demonstrated using LANDSA T-TM data and digital orthophotos.
Besides a typical image enhancement method, namely the co/our-transformation, the developed cartographie image processing system is described.
The digital/y constructed maps provide a wide variety of applications for different user groups. By use of special opera tors, cartographie details can be handled in a flexible way and produced economical/y at resolutions that satisfy cartographie requirements.
+ Introduction
Duetothe improving radiometrie and geometrie resolution of present sensors, remote sensed data are increasingly used in different disciplines: geography, geology, forestry, planning etc.
Simplified remote sensed data can be characterized by two facts: data represent spectral information and are coded in rasterformat. The spectral attributes of remote sensed data are the most important for many users and are used in the different ways:
• landuse classification
• image enhancement
• colour transformation for special applications
• etc.
+ Digital Cartog raphy
From the cartographie point of view it is ali the same which type of maps being derived from remote sensed data:
land use maps, spectral maps or plain image maps. The image data are always coded in rasterformat and symbolize the map surface.
63
Cartographie systems to handle these data are normally rastersystems and consist of following modules (Schweinfurth 1985):
• map surface (satellite data, orthophoto etc.)
• mapframe
• map lettering
The cartographical process is very similar to the production of analogie maps, but is performed in a solely digital way. He runs as follows:
• generation of the map frame
• combination of the map frame with the map surface (satellite data etc.)
• inserting names into the map surface
• inserting of further information (e.g. isolines, special signs etc.)
• sheet assembly
An often used modul in this process is «map lettering ... Map lettering has to explain the map contents. While in analogie maps the particular letters haveto be inserted cartographically, they must be co pied by computerwithin digital maps.
Letters or characters, which should be reproduced digitally, are stored in a symbollibrary. ln the developed symbol generation and manipulation program SCRIBO (SESTER 1989) the characters are stored as scanning patterns. Each letter is described by box coordinates, font baseline, body heigt and width. Letter size, inclination and direction can easily be ha nd led by a few parameters. Characters, however, cannat be reduced at will. Due to the limited size of pixels, particular letters will merge into one another.
The limited size of characters in digital images is indeed a severe problem. If you choose a high resolution to achieve an excellent graphie quality, the method is no longer economically acceptable duetothe need of large storage.
On the ether hand, type face looks stepped, if the resolution is low. By means of special antialiasing operators it is possible to process characters in such a way that their stepped structure almost disappears (Fig. 1) (Schweinfurth, 1986).
Fig. 1: Left: binary letter; right: result of antialiasing
+ Image Optimization
Merging of remote sensing imagery and thematic information is handled in many different ways. One, often practised approach is to underlay sail, geologie or otherthematic maps by the bright ness variations derived from optical imaging systems. This relief-information should make orientation easier for the observer.
Another idea is to create colour composites in a way that structural information is preserved and spectral differences are simultaneously enhanced. The resulting high saturated colours are now useable for constructing a readable legend (Kaufmann, 1988).
ln a first step the covariance matrix is calculated to select three significant bands of the image data corresponding to the respective application. These bands are needed for additive colour coding.
Because the saturation cannat be tully optimized by means of additive (Red, Green, Blue) colour processes e.g.
in the photolaboratory, the following procedure is used. Precondition is a digital transformation into three new mutually independant components called lntensity (1), Hue (H) and Saturation (S).
64
ln arder to en ha nee subtle variations in structural contrast, highpass-filtering is applied to the 1-component using a 3x3 matrix and added back. The disadvantageous effect of reduced spectral differences by filtering the RGB components can be avoided. After a linear shift of saturation levels to the high saturation part within the given 8-bit range, the modified intensity and saturation components and the original hues are retransformed into RGB-outputs (Bodechtel, 1985, Haydn, 1982). The last steps are geometrie corrections, histogram computing and stretch for display on screen or transmission via a photowrite system (Fig. 2) (Kaufmann, 1988).
-
PROCESSIH6 CONCEPT general task
hipass-fi 1 tering applled ta I-canponent
increase
~ of saturation CO!IIPOnent
color dtsplay
pnotowrite systs
+ Applications
• Spectral Map
selection of three relevant spectral bands for RGB color-codtng by corre-
lation techniques and Quantitative parameters
intens i ty ( l) ..._ calculation of the three hue (H) - mutually independent campo- saturation (S) - nents from inputs ( 1.2.3)
contrast
retransfonnation of l,H,S
tc R, G .a-outputs
- -
r---,
1 radiometrie 1 1 corrections 1
1 1
L---~
input 1 (B)lue - Input 2 (G)reen
input 3 (R)ed
-
( 8 )lue output 1 (G)reen output 2 ( R )ed output 3
r ---1
1 geometrl r
-
histograms ~-- 1 11 corrections 1
1 1
enhancement
L -- - -- - .J
Fig. 2: Processing flow chart (Kaufmann, 1988)
Comparative, quantitative and qualitative results have shawn, that combinations of LANDSAT TM-bands 1, 4, 7 is the most relevant for rock and sail discrimination where vegetated areas can be separated too (Bodechtel, 1985).
Figure 3 shows a result of digital cartography and image enhancement using these bands. lt offers optimum conditions for lithological and structural mapping within one single image product. Cause the legend explanations depend on surface categories and not on stratigraphie units, the resulting product is called «Spectral Map .. (Kaufmann, 1988).
• Thematic Orthophotomap
Figure 4 shows a superimposition of a land use plan (FNP) and a digital orthophoto (Schulz, 1987). The FNP is based on the German base map (DGK 5) at the scale of 1 :5.000, whereas the orthophoto is derived from an aerial photograph at 1:13.000 and a DTM of a 50 m raster. The cartographie result- the Thematic Orthophotomap-is produced by image optimazition techniques and satisfies high cartographie requirements.
Difficult to see by the black-white reproduction there are conflicts between orthophoto and the FNP especially in streets. Whereas the positional errors in small streets are caused on height differences of houses and on radial displacements, the origin of the eminent differences in the main roads only can explain by inaccuracies of the DTM and the computation of the orthophoto (Schweinfurt, 1988).
65
O'l O'l
ce· .,
(,)
,
(/) ID (')- ..
Dl3: Dl
,
o. ID
..
:c:·ID o.
-
0 3,...
)>
z
(/) )>-;-1 -4 3:
SPECTRAL MAP
processed on the basis of TM-safeUite data
As Siraf-Area (SW-Saudi Arabia)
Legend
- Th1Basal!
ci!W>e-rlci>
tJrdisiJo Th3 Tl1 Kno!!nl!e T!2 LaterU&
..
~,.IIi, •. gg!Granlte, G<anodlo<it&
..: gg2 (T161SSic ciai('O oo3 pegr.alite dl O!t~~Ua, Aod<»llta d2
Pw1 Wa,ikl -'Sndstooo Pw2 N.Pian tyJe
sb1 Schis!o.w CO<Il>lo>
llndesiJG Sb2 cPf)b..~Jsk
gpl Granite, S!1•n~e
l~ gp2 pl}( ei<IJ/ic gr1 Grenl!o
Ct1k:8hF:JIÎc gr2
omphiboH!&
-?Schis!
Vegetation
Malo roadt'\,~' Reservoir VUioge Syncllne 'J: Aoliclloa 7/app.horitontai
/ L Ot.splacoment
Data Source ' fOS AT oN ASA LAND SA T -5 The matie >Jloppe<
Acquislllorl Oof8 : 17.04.85, Polh f(Jl Rori' 48 Otbll Ali/lude 705 lm~ Ground Pixo/ Slze 80x80 rn Usod Speclro/Bands : 1: 0.45·0.52 um Œl/oe) cod: B
4 : 0.78-0.00 urn (NJR) cod.: e
7 : 2.08-2.35 ..., (S\Wi') cod: R Source of geologie ù>loml8!1oo :
MAP GM-217A U.S.G.S. (DGMR)
....-r===: r.
-
!0 km0 1 ,
lmeQ< l'rocoS!IOf!
Md!l!g!loiC.rlo{j•ophij:
R.&dw,Ji!fti,- (i&il-rwfiiW.Wtli:""""
Jnatltut. o1 P1wl0(,1t'lltmtlti'1J
•nd Top~
t..ftWor&liJOI~dF.R.
U!ed progt"ft1ll aljStom:J : O!OAli.sœJBO
Fig. 4: Thematic Orthophotomap (Landuse)
+ Conclusions
THEMATISCHE ORTHOPHOTOKARTE
I7!J!!/J,
-
a
'fRf!'
- œ
Wo!H\tHir,ll!!l''''''ll
Sor.d!:rl=j!:'!:r.iel·~
fJ!IiJll Üb•.::rf-.rliJt;·r.:o:l Sl·;~~nnr:ei~
îl!lfJI ~'!i~:~(!'l 1•)\ ~;~cr. Srtdf'l'\I"'W.'!''i\{''"'
Combinations of image data and cartographie information are mostly unsatisfactory and do not utilize the full spectrum of spectral image contents and available tools. By means of a new strategy, base on an association of two different methods of image processing nam ely Image Optimization and Digital Cartography the se deficiencies can be overcome. This combination offers optimum results by solely digital process. Combining data of different origin by these methods, there may be occur geometrical conflicts, which risk the satisfying results.
References
BODECHTEL J. and KAUFMANN H. -Future MOMS Development- Discussion of lmprovements and Possibilities ta be exspected for Geologie Applications. - Proc. ESA-EARSeL Symp., ESA SP-233, Strasbourg 1985.
HAYDN R., DALKE G.W.,HENKEL J. and KAUFMANN H.- Multidisciplinary investigations on HCMM data over Middle Europe and Marocco. - NASA-HCMM type Ill Final-Report, HCM-053.
KAUFMANN H. and SCHWEINFURTH G.-An Interactive Approach to Geologie Remote Sensing by Means of Image Optimization and Digital Cartography.- Geologisches Jahrbuch, A 104, Hannover, 1988.
SCHULZ E. - Digitale Erstellung einer Thematischen Orthophotokarte unter Verwendung eines Flachennutzungsplans. - Unpublished diploma thesis, IPF Universitât Karlruhe, 1987.
SCHWEINFURTH G.- Orthophotomaps from Digital Orthophotos.-Photogrammetria, 40, Amsterdam, 1985.
SCHWEINFURTH G.- Projekt: Digitale Orthophotokarte.- Nachr. aus dem Karten-und Vermessungswesen, 197, Frankfurt a. M., 1986.
SCHWEINFURTH G. - Errors due ta Integration of Image Data ta Geographical Information Systems. - lnt. Archives of Photogrammetry and Remote Sensing, Vol. 27, Part B 10, Kyoto, 1988.
SESTER M. - SCRIBO- ein Programm zur Plazierung von Schriften. - Nachr. aus dem Karten - und Vermessungswesen, 1 103, Frankfurt a.M., 1989.
Co/our photocopies of the original documents can be obtained on request from the Comité Français de cartographie
67