mosaicking raster geodata

Mosaicking Raster GeodataGetting Started

MosaickingRaster Geodata

with

TNTmips®

MOSAICKING

Mosaicking Raster Geodata

Before Getting Started

It may be difficult to identify the important points in some illustrations without acolor copy of this booklet. You can print or read this booklet in color fromMicroImages’ web site. The web site is also your source for the newest GettingStarted booklets on other topics. You can download an installation guide, sampledata, and the latest version of TNTlite.

http://www.microimages.com

This booklet introduces procedures for creating raster image mosaics in TNTmips®.The Mosaic process provides two modes for assembling images: automatic posi-tioning of georeferenced images, and manual positioning by placing tie pointsbetween pairs of overlapping images. The exercises in this booklet introduce theMosaic interface and show you how to use the many tools for creating a nearlyseamless mosaic, including integrated masking with processing areas, contrastand color matching, and varied overlap processing options.

Prerequisite Skills This booklet assumes that you have completed the exercisesin Getting Started: Displaying Geospatial Data and Getting Started: Navigating.Those exercises introduce essential skills and basic techniques that are not cov-ered again here. Please consult those booklets and the TNTmips reference manualfor any review you need.

Sample Data The exercises presented in this booklet use sample data that isdistributed with the TNT products. If you do not have access to a TNT productsCD, you can download the data from MicroImages’ web site. In particular, thisbooklet uses sample files in the MOSAIC data collection.

More Documentation This booklet is intended only as an introduction tomosaicking raster objects. Consult the TNTmips reference manual for more in-formation on the Mosaic process.

TNTmips and TNTlite® TNTmips comes in two versions: the professional ver-sion and the free TNTlite version. This booklet refers to both versions as“TNTmips.” If you did not purchase the professional version (which requires ahardware key), TNTmips operates in TNTlite mode, which limits object size anddoes not allow export.

The Mosaic process is not available in TNTview, TNTedit, or TNTatlas. All ofthe exercises can be completed in TNTlite using the sample geodata provided.

Randall B. Smith, Ph.D., 24 August 2001


The Mosaic process in TNTmips lets you combineraster images of varied types into a single mosaicimage. You can mosaic grayscale rasters, RGB colorraster sets, or color composites. The input compo-nents do not have to be of the same type; you canmosaic different types of grayscale rasters, RGB ras-ter sets with color composites, or even color withgrayscale rasters.

Image positions in the mosaic can be established intwo ways. If all input images are georeferenced, usethe Automatic positioning option to automaticallyplace the images in the specified map projection. Ifsome or all of the images lack georeferencing, usethe Manual positioning mode to place tie points be-tween pairs of overlapping images. Manual modeuses a bundle adjustment algorithm to compute aleast-squares best fit for all tie points and any avail-able ground control points.

A number of options allow you to fine-tune the ap-pearance of the mosaic. You can define thegeographic extents of the mosaic manually by draw-ing an extents box or match the extents to a referenceobject. If you want to exclude parts of the inputimages from the mosaic, such as the fiducial marksand marginal data blocks on scanned aerial photo-graphs, there is no need to crop the images prior toentering the mosaic process. Simply define Process-ing Areas to automatically mask unwanted portionsof each input object (or object set) as the mosaic isprocessed.

The Mosaic process allows you the option to applycontrast enhancement to each input raster in con-structing the mosaic and to set up contrast matchingof input objects. You can specify a reference formatching or match all objects to a model histogram.There are also a number of options for processingoverlap areas to produce nearly imperceptible tran-sitions between input images.

� launch the Mosaicprocess (Process /Raster / Mosaic)

� press [Close] on the Tipof the Day window

� from the Layer Menu onthe Mosaic window,choose Default Name /File and Object Name

� choose File / Exit� reopen the Mosaic

process

The exercises on pages 4-9introduce the Mosaicprocess interface, andillustrate how to mosaicgrayscale rasters, set theoutput cell size, and definethe geographic extents of themosaic. Pages 10-11 showyou how to createprocessing areas to crop theinput images. Some causesof spatial mismatch inmosaics are reviewed onPage 12. The exercises onpages 13-16 introducecontrast matching forgrayscale and colormosaics. Mosaic layouts,raster overlap operations,and trend removal arediscussed on pages 17-19.Pages 20-22 introduce themanual mosaic mode, andpage 23 summarizes theallowed types of input andoutput rasters.

Welcome to Mosaicking Raster Geodata


As an introduction to the Mosaic process, we mo-saic four georeferenced images using the Automaticpositioning option. The images are segments of ad-jacent (and overlapping) digital orthophotoquadrangles covering part of Lincoln, Nebraska. Thegeoreference information for each raster is used au-tomatically to determine its position in the mosaic.These relative positions are also automatically shownin the View window.

STEPS� click the Add icon

button on theInput tabbed panel

� use the standard SelectObjects window tonavigate to the LINCOLN

Project File in the MOSAIC

data collection andselect raster objectsNW_DOQ, NE_DOQ, SE_DOQ,and SW_DOQ (in thatorder)

� click the Run iconbutton and usethe File / ObjectSelection procedure tocreate a new Project FileMOSAICS and a new rasterobject

Making Your First Mosaic

The View window automaticallydisplays georeferenced input objects intheir correct relative positions.Overlapping objects are stacked in theorder in which you added them (last ontop). The extents box and object labelfor the active raster are drawn in red.

The completed mosaic is automaticallydisplayed in the Mosaic Result window.

The input rasters are listed on the input panelof the Mosaic window. The last raster addedis the default active raster; its listing ishighlighted in black.


The manner in which input objects are displayed iscontrolled by the Display Options settings on theParameters tabbed panel. The default settings (usedin the first exercise) display each input ras-ter with a colored box outlining the objectextents (Show Object Extents), and a la-bel with the name of the Project File andobject (Show Object Labels). These andother display options can be turned off oron using the corresponding toggle buttons.Changes in these settings do not take ef-fect until you redisplay the View window.

If your input rasters are very large, you maywish to turn on the Wire Frame DisplayMode (with Object Extents and Object Labels alsoturned on). In Wire Frame mode, only the extentsbox and label are displayed, which speeds upredisplay while still allowing you to see the spatialrelationships between the different input objects.

Explore Display Options

Default Display Optionssettings on the Parameterstabbed panel.

Choose Color from the Options menu to openthe Color Editor window and adjust thebackground color of the View and MosaicResults windows.

The Mosaic window input list uses thelayer name option that is set from theLayer / Default Name menu: objectdescription (the default), object name, orfile and object name. The illustrations inthis booklet show the file and objectname (the selection you were instructedto make when you first opened theMosaic process). A change in theDefault Layer Name setting takes effectin the next Mosaic session.

� turn off the ScanlineTransparency togglebutton on the DisplayOptions panel

STEPS� click on the Parameters

tab to expose theParameters panel

Choose Wire FrameDisplay Mode to displayonly the extents box andlabel for each input image.


The raster cell size of the output mosaic is controlledby the values in the Line and Column text boxes onthe Cell Size portion of the Output tabbed panel. Thedefault values are provided by the input raster withthe smallest cell size (and therefore the highest spa-tial resolution). The four Lincoln DOQ input rastersall have a line and column cell size of 2 meters, sothe mosaic produced in the first exercise also has a2-meter cell size.

If rasters in an input set have differing cell sizes,you can choose any one of them to control the cellsize of the mosaic by selecting the raster from the

list in the Auto-Update menu on the Cell Size por-tion of the Output tabbed panel. You can also enteran output cell size manually, as in this exercise.

STEPS� expose the Output

tabbed panel� in the Cell Size controls

change the Line andColumn values to 3.0

� click the Run iconbutton and directthe output rasterto the MOSAICS ProjectFile

Change Output Cell Size

You can close the Mosaic Result window at the end of each exercise (by choosingClose from the window’s File menu) to reduce screen clutter as you move to thenext exercise. Another Mosaic Result window opens automatically each timeyou generate a new mosaic.

Default Cell Size and RasterSize settings used in theprevious exercise.

The Raster Size values update automaticallywhen the output cell size is changed.Increasing the mosaic cell size reduces thenumber of lines and columns in the outputraster, reducing its spatial resolution.

Output mosaic with reduced rastersize and reduced spatial resolution.

If the mosaicked images do not fill the fullextent of the output mosaic raster, “blank”cells (such as the black edge cells in themosaic to the right) result. Blank cells arenormally assigned a value of 0. Changingthe Null Value option to Set designates any“blank” cells in the output mosaic as null,using the value shown in the text field to theright of the option button. If any of the inputobjects have null cells, they are automaticallyrecognized and converted to the null valueyou have set for the mosaic. Null cells in themosaic are displayed transparent by default.


The Output Area option menu gives you several waysto control the geographic extents of the output mo-saic image. The previous exercises used the defaultTotal Extents option, wherein the mosaic assumesrectangular extents equal to the total geographic ex-tents of the set of input objects. The User-Definedoption matches the output raster extents to a rectan-gular box that you draw in the View window usingthe Output Area tool.

STEPS� in the Output tabbed

panel reset the Line andColumn Cell Size fieldsto 2.0

� select User-Defined fromthe Output Area optionbutton

� click the OutputArea icon buttonon the Viewwindow

� place the mouse pointernear the upper left cornerof the image area

� click and hold the leftmouse button as youdrag the mouse pointertoward the lower rightcorner of the image areato create an extents box,then release the mousebutton

� drag an edge or cornerof the box to resize it ifnecessary, keeping thebox within the imagearea

� click the right mousebutton to accept theoutput area

� run the Mosaicprocess

Define Output Extents Manually

Mosaicked image. The geographicextents and cell size together deter-mine the size of the output image.

Select User-Defined from the Output Area option menu.

Once you have drawn and accepted the Output Areabox, its extents are shown in the Output Extents textboxes.

Click on the Output Areaicon button to draw anoutput extents box.

When you click the rightmouse button to accept theextents box, the selectedarea is shaded.


You can also match the extents of the output mosaicto the first (lowest) object in the input list. Moveany desired object to the bottom of the list, thenchoose the Match First Layer option from the Out-put Area menu on the Output tabbed panel. If youdon’t want this reference object to become part ofthe output image, press the Reference icon button tochange its status from Mosaic (used to create theoutput image) to Reference.

STEPS� click the Add icon

button on theInput tabbed panel

� select the MAP objectfrom the LINCOLN ProjectFile

� click theReference iconbutton; the status of theMAP object changes fromMosaic to Reference

� click the ToBottom iconbutton above theinput object list to movethe highlighted MAP

object to the bottom ofthe list

� on the Output tabbedpanel, select Match FirstLayer from the OutputArea option menu


Move an object to thebottom of the input list touse it to control the extentsof the mosaic.

Use the Reference iconbutton to toggle objectstatus between Mosaicand Reference.

Reference MAP object displayed on top of theinput DOQ images before being moved tothe bottom of the input object list.

Mosaicked image with extentsmatched to the MAP raster object.

Match Extents to a Reference Object


The default method of handling overlapping imagesin the Mosaic process uses the topmost raster in eachoverlap area for the output image. (We will exploreother options in a later exercise). In the Lincoln DOQimages, for example, the high school building lies inthe overlap between rasters SW_DOQ and SE_DOQ. Inthe previous exercises the image of the high schoolin the output mosaic came from SW_DOQ, which over-lies SE_DOQ. This overlap order was determined bythe order in which you added the input rasters.

If you would prefer to have the darker image of thehigh school from SE_DOQ appear in the mosaic, youcan use the icon buttons above the Input list to changethe overlap order of the input images. Youcould either raise SE_DOQ or lower SW_DOQ.

STEPS� change the Output Area

setting on the Outputtabbed panel to TotalExtents

� select the MAP object inthe Input list andpress theRemove iconbutton

� select the SE_DOQ objectin the input listand press theRaise icon button


Change Input Object Order

Lower part of the output mosaic incorporat-ing the high school image from SE_DOQ.

Object SE_DOQ raised above SW_DOQ inthe input list and displayed overlapping itin the View window.

� press the Remove All icon button on the Input panel when you have completedthis exercise (choose No when asked if you want to save the layout)

ToTop

ToBottom Raise Lower

With the Auto-Popup Layer display optionturned on (on the Parameters panel), youcan view any input raster in full, regardlessof its position in the stacking order, bymaking it the active raster. Simply click onthe Input list entry for the desired raster.The View window redraws with the newactive raster temporarily on top of anyoverlapping images. (The order of imagesin the input list and in the output mosaicdoes not change.)


Choose the Output ProjectionSTEPS� click the Add icon

button on the Inputpanel

� select objects SEC_1 andSEC_2 from the BENNET

Project File� on the Output tabbed

panel, change the OutputProjection menu toLambert ConformalConic and note thechange in orientation ofthe input objects in theView window

� change the OutputProjection menu back toUniversal TransverseMercator

Objects SEC_1 and SEC_2 are extracted portions ofscanned airphotos that have been georeferenced todifferent map coordinate systems. SEC_1 isgeoreferenced to the Universal Transverse Mercator(UTM) coordinate system and SEC_2 is georeferencedto a User-Defined coordinate system in the LambertConformal Conic projection. Neither photo is ori-ented to its projection (raster lines and columns arenot parallel to coordinate system grid lines).

You can orient the mosaic to either of these inputcoordinate systems by making the appropriate selec-tion on the Output Projection menu on the Outputtabbed panel. All input object coordinate systemsare automatically shown on this menu, with that of

the first-added object selectedby default. All input objectsare automatically reprojectedto the selected coordinate sys-tem (if necessary) in themosaic raster. The View win-dow automatically shows theinput objects in the selectedcoordinate system as well,providing a preview of the ori-entation of the mosaic.

View of input objects oriented to LambertConformal Conic projection.

View of input objects oriented toUniversal Transverse Mercatorcoordinate system.

Keep the current settings andcontinue on to the next page.


Create a Processing Area for MaskingSTEPS� choose Processing Area

from the Image Areaoption button on the Inputpanel

� click theProcessing Areaicon button on themosaic View window

� change the Mode settingof the Line / Polygon EditControls window toStretch

� use the Add Endoperation to add verticesto outline a ProcessingArea polygon for theSEC_2 photo as illustrated

� use the Drag Vertexoperation to adjust thepolygon shape asneeded

� press the right mousebutton to accept thepolygon

Keep the current settings andcontinue on to the next page.

The scanned photos used in this exercise exhibit se-vere vignetting (darkening toward several corners).You probably wouldn’t want these dark areas in-cluded in a mosaic. To eliminate them, you can definea processing area for each input raster (or RGB ras-ter set) and set the Image Area option to ProcessingArea. Only the portion of the image inside the pro-cessing area is then included in the final mosaicimage. Unwanted parts of the image outside the pro-cessing area are masked out (set to the null value) asthe mosaic is assembled. (As you will see later, theImage Area menu is necessary because processingareas can serve several functions in the mosaic pro-cess.)

Clicking on the Processing Area icon button opensthe Line / Polygon Edit Controls window. Use theseedit controls to draw and edit an appropriate poly-gon to define the processing area. When you clickthe right mouse button (or press [Apply]) to accept

the polygon, theprocessing area isoutlined andshaded in color inthe View window.

The Line / Polygon EditControls are described indetail in Getting Started:Editing Vector Geodata.

Use the Line / Polygon EditControls to create aprocessing area for theSEC_2 photo.

Add End Drag Vertex Stretch


STEPS� select the BENNET / SEC_1

object from the input list� use the Line / Polygon

Edit Controls to create anew processing area forthe SEC_1 photo

Create a Second Processing Area

� choose Set from the NullValue option button onthe Output tabbed panel


Create a new processing area for the SEC_1 photo (shownbefore accepting the selected area). The current area isshown in red, while any others are in yellow.

Mosaic of the designatedportions of the two photosoriented to the UTMcoordinate system. Vignettedportions of the photos havebeen trimmed off. Blankareas around the image wereset to the null value and aretransparent in the MosaicResult window.

The Clear All icon buttondeletes all currentprocessing areas.

The Clear Area icon buttondeletes the processing areafor the currently selectedraster object.

A processing area is specific to a particular inputraster (or RGB raster set), so you can create a uniqueprocessing area for each input object, if needed.Before creating another processing area, select theappropriate object from the list on the Input panel; aprocessing area you draw is applied only to the cur-rently selected object.

You can use processing areas to design custom cutlines between overlapping objects to make seams inthe mosaic less obvious. If possible, draw the pro-cessing area boundary for the top object within areasof uniform tone or color, rather than following tonalboundaries in the image. Cross linear features suchas roads or railroads at low angles (rather than a rightangle) to minimize visual mismatch in position.Avoid long straight seam lines.


Causes of Spatial Mismatch in MosaicsAs you examined the mosaic created in the preced-ing exercise, you may have noticed some slightmisalignment of roads and other features along theseam between the input images. This is probably agood point to stop and consider the potential causesof such problems.

When you mosaic scanned planimetric or topographicmaps, all map features are presumed to be shown intheir correct horizontal positions, as if each objectwere viewed from a unique vantage point directlyoverhead. If the maps have the same projection andcoordinate system, we might expect features at theseam to match exactly. However, minor mismatchesmight occur because of georeferencing errors or car-tographic errors in the original maps.

There are additional sources of image mismatch whenyou mosaic remotely-sensed images, because all ob-jects in a particular image were viewed from a singlevantage point and that view point varied from imageto image. The perspective view of a single imagecan cause the apparent positions of ground featuresto be displaced from their correct horizontal posi-tions. Relief displacement and tilt displacement(illustrated at left) are the main causes of this spatialdistortion, which contributes to errors ingeoreferencing component images. These effects aremost obvious in low-altitude images, such as air videoand aerial photographs (including those used in thelast exercise). An object pictured in adjacent photosmay be displaced from its true position by differentamounts and in different directions in each image.When the photos are mosaicked, the two images ofthe same ground object are not placed at the samelocation in the mosaic.

To produce the best mosaic from georeferenced images showing tilt displacement,first use automatic raster resampling to reduce the tilt effects (see Getting Started:Rectifying Images). Removing relief displacement requires full stereoscopicmodeling of image pairs to produce an orthophoto (see Getting Started: MakingDEMs and Orthophotos for more information).

Ground positions projectalong sight lines (blackarrows) to the image plane,producing several types ofhorizontal displacements.Red arrows show projectiondirections needed tomaintain correct relativepositions.

Tilt Displacement:horizontal object shiftsresulting from a tilted imageplane (sensor not pointedstraight down).

Relief Displacement:horizontal position shiftsarising from the differingelevations of ground objects.

Common Distortions inRemotely Sensed Images

Ground

Tilted image plane

Sensor

Sensor

GroundImage plane


Apply Contrast TablesSTEPS� press the

Remove All iconbutton on the Input paneland choose No whenasked if you want tosave the layout

� click the Add icon buttonand select objectsSPOT_RED and TM_RED

from the FRANKLIN ProjectFile

� on the Contrast panel,make sure that the ApplyContrast Tables togglebutton is turned on


TNTmips allows you to enhance the brightness andcontrast of images for display by creating and sav-ing contrast tables. A contrast table maps each rawraster value to a corresponding screen brightnessvalue, enhancing the display of the image while pre-serving the original numerical values in the raster.The Mosaic process automatically uses the saved dis-play parameters for each input raster to display it inthe View window. If you last viewed the raster in

the Display process using a saved contrasttable, that contrast table will be used au-tomatically to display the raster in theMosaic View window.

If the Apply Contrast Tables toggle but-ton on the Contrast panel is turned on (the defaultstate), then contrast-enhanced values are transferredto the output mosaic rather than the raw input rastervalues. Using this option ensures that the mosaicincorporates the results of your previous contrast-enhancement efforts with the individual input rasters.

Both input rasters in this exerciseare displayed automatically withsaved Exponential contrast tables.Contrast-enhanced values weretransferred to the output mosaic.

Input rasterTM_RED dis-played withno contraste n h a n c e -ment forcomparison.


Set Contrast All LayersSTEPS� on the Input panel,

click the Contrasticon button

� close the resultingRaster ContrastEnhancement window bychoosing Close from itsFile menu

� click the SetContrast AllLayers icon button andselect AutoNormalize


You can also adjust the contrast of any grayscale in-put raster in the Mosaic process. Pressing theContrast icon button on the Input panel opens thestandard Raster Contrast Enhancement window forthe currently-selected input raster. You can changethe contrast method, modify the ranges, or make anyother adjustments just as you would in the SpatialData Display process. (For more information oncontrast enhancement, see the booklet GettingStarted: Getting Good Color).

The Set Contrast All Layers icon button allows youto apply a single automatic contrast enhancementmethod to all of the input rasters. The dropdownmenu that opens when you press this icon buttonincludes the standard automatic contrast en-hancement methods provided in the Displayprocess. If the Apply Contrast Tables togglebutton is turned on, contrast-enhanced values

computed using the se-lected automatic methodare transferred to the mo-saic. The Default optionon the menu returns con-trast enhancement foreach input raster to the de-fault state.

Mosaic created with Auto Normalizeenhancement applied to each input raster.

When you apply contrast in creating amosaic, either using contrast tables orsetting automatic enhancement for allinput rasters, a Linear contrast table iscreated for the resulting mosaic raster.This table ensures that the contrast-enhanced mosaic is displayed asintended, without further automaticenhancement by the Display process.


STEPS� click the Remove

All icon button onthe input panel and donot save the layout

� make sure the AutoNormalize display optionis turned off on theParameters panel

� click the Add iconbutton and selectobject TM_NIR fromthe BRK_TM Project Fileand object SPOT_NIR fromthe BRK_SPOT Project File

� select TM_NIR from theAuto-Update menu in theCell Size portion of theOutput panel

� select Reference Rasterfrom the MatchingMethod option button onthe Contrast panel

� select TM_NIR from theReference Raster optionbutton

� turn on the SmoothHistogram toggle button


Contrast Match Grayscale Rasters

Mosaic with contrast matching.Input grayscale rasters with differing contrast.

Grayscale rasters that you mosaic commonly willshow differing brightness ranges and differing con-trast. The Contrast panel provides several contrastmatching options. When you choose the ReferenceRaster option, you then must select one input objectas the Reference Raster for contrast matching. TheMosaic process matches the brightness histogram ofeach input object as closely as possible to the histo-gram of the Reference Raster.

The histogram for the selected ReferenceRaster is displayed for your inspection inthe Target Histogram panel.


Mosaic RGB Color Raster SetsSTEPS� click the Remove

All icon button onthe input panel

� choose Add Project File(RGB) from the Filemenu, and select theBRK_TM and BRK_SPOT

Project Files� select TM_NIR from the

Auto-Update menu in theCell Size portion of theOutput panel

� on the Contrast panel setthe Contrast MatchingMethod to ReferenceRaster and choose theTM raster set as thereference

� run the Mosaicprocess andname the Red, Green,and Blue output rastercomponents

When you apply contrast matching to RGB rastersets, histogram matching is applied separately to thered, green, and blue color components. The bright-ness and contrast of each color component areindependently adjusted to match the correspondingcomponent of the reference set. This procedure ad-justs the overall color balance of each input rasterset to match the color of the reference set. In thisexample, a darker SPOT image is matched to abrighter Landsat Thematic Mapper raster set. Theselected bands are Near Infrared (displayed as red),Red (green), and Green (blue). This combinationyields an image similar to a color-infrared photo-graph, with vegetated areas appearing in red.

Instead of choosing a reference raster and adjustingthe contrast of the other input objects to match it,you can choose to match all input images to a modelbrightness distribution. The two available modelsare Equalize (an equal number of cells at each bright-ness level) and Normalize (a normal or Gaussiandistribution of brightness). Both of these modelsapply the greatest contrast enhancement to the mostpopulated range of brightness values in each image.

Input RGB raster sets with different color balance. Mosaic with contrast matching.

The Add Project File (RGB)option adds the first threerasters in the selectedproject file as an RGB rasterset. Color assignments areby object order: 1 = R, 2 =G, 3 = B.




� click the Add iconbutton on theInput panel

� select object TM_24COMP

from the BRK_TM ProjectFile and objectSPOT_24COMP from theBRK_SPOT Project File

� select TM_24COMP fromthe Auto-Update menu inthe Cell Size portion ofthe Output panel

� on the Contrast panelchoose TM_24COMP asthe Reference Raster


You can also mosaic color composite rasters such as24-bit or 16-bit composites (with separate red, green,and blue values stored for each raster cell) or 8-bitcomposites with color maps. When you performcontrast matching with color composites, the Mo-saic process automatically creates red, green, and bluehistograms for each input object. Contrast match-ing is then done exactly as it would be with RGBraster sets. The best color matching results areachieved with 24-bit composites such as those usedin this exercise (or when matching RGB and 24-bitcomposites). Composite 16-bit or 8-bit rasters willusually not yield as close a color match.

Input color composites with different color balance. Mosaic with contrast matching.

Use this option button to choose which oneof the three histograms (Red, Green, orBlue) from the reference raster (or rasterset) is displayed for inspection in theTarget Histogram pane. Each of these histograms is used as the target for contrastmatching for its respective color, regardless of which one is currently displayed. If you areusing the Selected Area matching option for the reference image, the histogram of thematching area is shown.

Mosaic Color Composites


The Matching Area is the portion of an input objectused to build a histogram for contrast matching. Thedefault selection is Whole Raster. By drawing pro-cessing areas for one or more input objects, andchoosing Selected Area as the Matching Area op-tion, you can designate which portions of the inputimages control the matching process.

In this example, we want to match the darker Landsatimage to the brighter SPOT image. But the latterimage includes a large area of turbid, sunlit waterand some very bright structures, neither of whichare present in the Landsat image. In order to pro-duce the best match for the more typical land areasin the two images, we draw a processing area for thereference SPOT image that excludes the water andanomalous structures. Histograms computed fromthis area (one for each color) are used as the targethistograms for matching. The histogram-buildingprocess reverts to the whole-raster mode for theLandsat image, for which we didn’t define a pro-cessing area.



� choose Add RGBRasters from the Filemenu, and select objectsNIR_TM, RED_TM, andGRN_TM from the PINOLE

Project File� repeat the last step, this

time selecting objectsNIR_SPOT, RED_SPOT, andGRN_SPOT

� set the Image Areaoption to Whole Raster

� set the ContrastMatching Method toReference Raster andchoose Selected Areafrom the Matching Areamenu

� select the SPOT rasterset as the contrastreference

� draw a processing areafor the SPOT image asshown below left(outlined in black)


Match Contrast Using Processing Areas

Input raster images with processing areafor reference Landsat image for matching.

Portion of mosaic with contrast matchedto selected area.


Save and Open Mosaic LayoutsSTEPS� click the Remove


� click the OpenLayout iconbutton

� select object MOS_LAYOUT

from the BENNET ProjectFile

If a mosaic layout object thatyou save refers to inputobjects in different ProjectFiles, do not move thesefiles to different directoriesor drives after saving thelayout. If you do, the Mosaicprocess may not be able tofind them.

Input list after opening theMOS_LAYOUT object.

View window with Bennetairphotos and processingareas restored from theMOS_LAYOUT object.

The Mosaic process allows you to use almost anynumber of input objects and to define processingareas, reference objects, overlap processing, contrastmatching, and other processing parameters. Whenyou are setting up a complex mosaic process, it is agood idea to save the mosaic layout using the SaveLayout option on the File menu. The input objectlist and all process settings are saved in a layout ob-

ject which you name and place in aProject File of your choice.

If you need to modify or add to themosaic at a later time, you can load

the saved layout using the Open Layout icon buttonor the corresponding option on the File menu. Allinput objects are added in the correct order, and allMosaic option selections and parameter values areset as they were when you saved the layout. You canthen add other input objects or modify settings asneeded. The layout you open here includes process-ing areas and other settings that you will add to inthe next exercise.

If you repeatedly make mosaics using subsets of alarge group of airphoto or airvideo images, you can

save a mosaic layout whichincludes the entire set ofimages, then use an extentsbox to define the input ob-jects needed for the currentmosaic. Saving the layoutin Wire Frame DisplayMode speeds loading of thelayout.

Keep the current settings andproceed to the next page.


Trend RemovalSTEPS� on the Trend panel

choose Selected Areafrom the Remove Trendoption button

� select 2 from the TrendOrder option button

� change the value in theSample Lines by: field to2, and repeat for theSample Columns by:field


Spatial brightness variations related to illuminationand lens effects are common in aerial photographsand video images. Shadowing due to oblique illu-mination can cause one side of an image to appearsignificantly darker than the other. This is a linear(first-order) brightness trend. Lens effects can causea radial darkening of the image (second-order trend).These trends can cause brightness mismatches acrossthe image seams in a mosaic, since the same areacan look brighter than average in one image butdarker than average in the adjacent one.

The controls on the Trend panel allow you to auto-matically process input images to reduce or removebrightness trends prior to contrast matching and as-sembly of the mosaic. You can choose the trend orderand base the trend removal on the whole raster or ona portion of the image selected by a process-ing area. In order to speed processing forlarger images, trend analysis can utilize asample of the image cells, with the size ofthe sample determined by the sampling in-terval.

The Bennet airphotos used in this exercise show awestward brightening due to sun angle, and a radialdarkening related to lens effects. A combination ofsecond-order trend removal and contrast matchingprovides a much closer match between the two im-ages than contrast matching alone.

Mosaic produced usingcontrast matching (settingssaved in the layout) andsecond order trendremoval. The linear andradial brightness variationsin the original images arenoticeably reduced,providing a better matchalong the seam. Comparewith the raw image mosaicon page 12.

Choose Selected Area touse a Processing Area todetermine the portion of theimage used for trendremoval.

The sampling interval fortrend removal is set by thevalues in these fields (thedefault value is 4).


The Overlap Method option menu on the Parame-ters panel allows you to choose different ways toprocess the overlapping portions of input rasters. Thedefault method (Last Raster) uses the topmost rasterin each overlap area for the output image. The othermethods attempt to create a gradual transition byassigning overlap cell values on the basis of a com-parison or mathematical combination ofcorresponding cell values from the input images.

The Average method uses the mean of the corre-sponding input cells. The Maximum method choosesthe maximum value, while the Minimum methoddoes the opposite. In the Chessboard method, inputcell values from the overlapping rasters are regularlyalternated in two dimensions to form a chessboardpattern. The Random Mixing method uses aweighted average of corresponding input cells, withrelative weights assigned randomly. In the Feather-ing option, the weighted average varies with distancefrom the image boundaries. The weighting coeffi-cients for each image are 0 at the boundary, andincrease inward to 1.0 at the feathering distance fromthe edge. Larger feathering distances provide a moregradual transition.

Try repeating this exerciseusing different overlapoptions so you can judgetheir effects for yourself.

The Deviation Filter allows you to selectivelyreject extreme differences in input cell valueswhen performing overlap operations. Ifcorresponding input cell values differ by morethan the Max. Deviation value you have set,the filter overrides the selected overlapoperation and uses the Last Raster value.

All of the Overlap methodswork best when there is verygood cell-by-cell registrationbetween the overlappingrasters.

With the Feathering method you can chooseeither a Linear or Nonlinear variation inweighting relative to image edges.

Raster Overlap OptionsSTEPS� click the Open

Layout icon buttonand select BRK_LAYOUT

from the BRK_TM ProjectFile

� on the Parameterspanel, chooseFeathering from theOverlap Method optionbutton

� turn on the DeviationFilter toggle button andset the Max.Deviation to 40.00


Mosaic using the Linear Feathering option.


Fill GapsSTEPS� click the Open

Layout iconbutton and selectGAP_LAYOUT from the GAP

Project File� choose Set from the Null

Value option button onthe Output tabbed paneland accept the defaultnull value

� select Average 4 fromthe Gap Filling Typemenu


With some mosaic projects, the problem is not deal-ing with overlap between component objects, butcompensating for lack of overlap between them. Thisproblem occurs most commonly with governmentdigital elevation data distributed by map quadrangle.Although the boundaries between adjacent elevationrasters should be coincident, in practice there maybe gaps between them that are one or two cells wide.

The Mosaic process enables you to automatically“fill” the gaps between adjacent mosaic components.These gaps must be represented as narrow strips ofnull values in the preliminary mosaic, so you mustchoose Set from the Null Value menu to create nullvalues in the mosaic and activate the Gap Filling Typemenu. The gap-filling filter interpolatesnew values for null cells in the gaps us-ing the average of either the foursurrounding or eight surrounding cells.

Portion of mosaicrun with gap-fillingturned off. A gap orseam of null cellsremains along thequadrangleboundary. Full mosaic with boundary gap filled.

The Maximum Width parameterspecifies an upper limit to the width ofthe gap that is allowed to be filled. Nullcells along the edges of the mosaic andthose in gaps wider than this thresholdare left unchanged.


page 24

STEPS� click the Open

Layout iconbutton and selectMAN_LAYOUT from theRAYMOND Project File

� note that the Positioningmenu option haschanged to Manual

� select SEC1 in the Inputlist

� click on theManualPositioning icon buttonon the View window

� drag the lower rightcorner of the extents boxfor the middle photo(SEC1) to enlarge it

� click the right mousebutton to redraw theimage

� repeat until features inSEC1 appear at about thesame scale as in theflanking images

� drag SEC1 to the right ifnecessary to uncoverthe right edge of SEC2

Manual Mosaicking

For nongeoreferencedimages, drag an edge orcorner of the extents boxto resize the selectedimage. Drag within theimage to move it.

Change to Manual positioning mode when you wantto mosaic a set of nongeoreferenced objects or a mix-ture of georeferenced and nongeoreferenced images.

Nongeoreferenced ob-jects can be selected onlyin Manual mode. Whenyou add the input objects,

georeferenced images are automatically placed intheir correct relative positions within the View win-dow. (If you are adding a mixed set of objects, add ageoreferenced object first in order to establish thecorrect window coordinate system.) Non-georeferenced images are tiled horizontally to theright of the first-placed image, in the order in whichthey are added. (The number of image tiles in a rowis set by the Autoplacement Tiles setting on the Pa-rameters panel. Additional images are tiled beneaththe previous row.)

The left and right images in this layout aregeoreferenced, but the middle one has nogeoreferencing or cell size. The middle image there-fore is not initially displayed at the correct scalerelative to its neighbors. After selecting thenongeoreferenced object in the input list, use theManual Positioning tool to move or resize it to makeit easier to identify and place tie points for pairs ofadjacent images.

Keep the current settingsand proceed to the nextpage.


STEPS� select Define Tie Points

from the Manual menu� with SEC1 still selected in

the Input list, in the TiePoints window click theReplace iconbutton for theRight image

� select SEC2 in the Inputlist and click the Replaceicon button for the Leftimage

� click the DefineTie Points iconbutton in the Viewwindow

� drag a line connectingthe road intersection inthe lower right corner ofSEC2 with its counterpartin SEC1

Place Tie Points for the First Image Pair

� in the Tie Points window,use the arrow buttonsbeneath the magnifierwindows to refine the tiepoint positions, then click[Add]

� add three more tie pointsas shown below

In Manual mode you must place tie points to estab-lish the position of a nongeoreferenced image relativeto its neighbors. Tie points identify correspondinglocations in a pair of overlapping images. At leastthree tie points are required for each image pair, andthey should be distributed to cover as much of theoverlap area as possible. Members of a pair are des-ignated Left and Right to identify their magnifierimage in the Tie Points window. To designate anobject as Left or Right, select it in the Input list, thenclick the corresponding Replace icon button at thetop of the Tie Points window.

After designating the image pair, use the Define TiePoints tool to establish a preliminary tie point loca-tion in the View window. Then use the controls inthe Tie Points window torefine the position beforeadding the tie point.

You can change thezoom level of themagnifier windowsas needed.

The Auto Correlate optionoperates when at least threetie points have been added.After you place anotherpreliminary tie point, clickingin one magnifier framelaunches a spatial correlationprocess that automaticallylocates the correspondingpoint in the other image.

The Positioning Interval setting defines the increment ofposition change (in pixels) triggered by the arrow buttons.

Keep the current settings andproceed to the next page.

Use the arrowbuttons to make fineadjustments in the tiepoint position.

Click on a point in amagnifier frame tosnap the tie point to it.


STEPS� select SEC1 in the Input

list� click the Manual

Positioning iconbutton

� move the SEC1 photo tothe left until you uncoverthe left edge of SEC6

� designate SEC1 as theLeft image and SEC6 asthe Right image

� click the DefineTie Points iconbutton and place tiepoints for this image pairas suggested below

Complete Tie Points and Run Mosaic

� click [Adjust] to previewthe final imageplacement and asummary of positionerrors

� runtheMosaicprocess

Final mosaicimage resultingfrom the tiepoint locationsshown.

Manual mosaic mode computes a least-squares bestfit for all tie points and any available ground controlpoints and uses the resulting locations to positioncomponent images in the mosaic. The Adjust but-ton on the Tie Points window lets you preview theresults of this bundle adjustment before you createthe mosaic. You can update the model after placingtie points for each image pair, or wait until all tiepoints are placed. When you adjust the model, theupdated positions are used to place the images in theview window, position errors are calculated for thecurrent image pair, and an Adjustment Report win-dow appears, listing the RMS (root-mean-square)

position errors for each image (0 for fullygeoreferenced images). You can return to specifictie points and adjust their position if necessary toimprove the model results.

Click [Adjust] to preview the mosaic results.

Click on a list entry to select the tie point for editing.


The Mosaic process accepts a full range of input raster types including binary (1-bit), grayscale, RGB color, and color composites. The range of grayscale rastersincludes 2-bit and 4-bit integer, signed or unsigned 8-, 16-, or 32-bit integer, andfloating point (32-bit or 64-bit). Color composite types include 24-bit and 16-bitRGB and BGR composites and 8-bit composites with a color map.

You will usually get the best results from the Mosaic process if all input rasterobjects contain the same data type. However, the process does allow you to in-clude different raster data types in a mosaic. The default output raster type dependson the specific input raster types and is designed to preserve the maximum datafidelity. The sections below summarize these relationships.

Grayscale Input Grayscale OutputAll same bit-depth and sign Same as inputAll same bit-depth, signed and unsigned Signed integerDifferent bit-depths Highest bit-depth32-bit unsigned and 32-bit floating point 32-bit floating point32-bit signed and 32-bit floating point 32-bit signed integer

If you mix signed and unsigned integer rasters, be aware that no scaling of rastervalues is performed by the Mosaic process, so some loss of data may occur ifinput values extend beyond the range of the output raster type.

Color Input Color OutputAll RGB RGBAll 24-bit composite 24-bit compositeAll 16-bit composite 16-bit compositeSame bit-depth RGB- and BGR composite RGB-compositeAll 8-bit composite with color map RGBRGB and any composite RGBComposites with different bit-depth Maximum composite bit-depth

There is no option to produce an 8-bit color-mapped mosaic. You can use theColor Conversion process (Prepare/Raster/Convert Color) to convert a completedcolor mosaic to an 8-bit composite raster if desired.

Mixed color and grayscaleYou can also mix color and grayscale rasters in a mosaic if you lack completecolor image coverage. The grayscale input raster will appear in grayscale in thecolor mosaic raster or RGB raster set.

Input and Output Raster Types

Mosaicking Raster GeodataAdvanced Software for Geospatial AnalysisMOSAICKING

M icro Im a g es, In c .11th Floor - Sharp Tower206 South 13th StreetLincoln, Nebraska 68508-2010 USA

Voice: (402) 477-9554 email: [email protected]: (402) 477-9559 internet: www.microimages.com

MicroImages, Inc. publishes a complete line of professional software for advanced geospatialdata visualization, analysis, and publishing. Contact us or visit our web site for detailed prod-uct information.

TNTmips TNTmips is a professional system for fully integrated GIS, image analysis, CAD,TIN, desktop cartography, and geospatial database management.

TNTedit TNTedit provides interactive tools to create, georeference, and edit vector,image, CAD, TIN, and relational database project materials in a wide variety of formats.

TNTview TNTview has the same powerful display features as TNTmips and is perfect forthose who do not need the technical processing and preparation features of TNTmips.

TNTatlas TNTatlas lets you publish and distribute your spatial project materials on CD-ROM at low cost. TNTatlas CDs can be used on any popular computing platform.

TNTserver TNTserver lets you publish TNTatlases on the Internet or on your intranet.Navigate through geodata atlases with your web browser and the TNTclient Java applet.

TNTlite TNTlite is a free version of TNTmips for students and professionals with smallprojects. You can download TNTlite from MicroImages’ web site, or you can orderTNTlite on CD-ROM.

Indexautomatic positioning mode......................3,4bundle adjustment...........................3,26cell size...............................................6contrast, applying

contrast tables.....................................14setting for all input..............................15

contrast matching...............................3,16-19grayscale rasters..................................16color................................................17-18matching area......................................19

display options..........................................5,9gaps, filling..................................................23georeferencing..........................................3,4histogram................................................16-19image area options......................................11layouts........................................................20manual positioning mode.........................3,24masking unwanted areas.......................11,12null value..............................................6,12,23

object order..................................................9output area................................................7,8Output panel..............................................6,7output projection, choosing........................10overlap areas............................................3,22Parameters panel..........................................5reference object

matching extents to................................8matching contrast to.......................16-19

relief displacement.....................................13processing areas..............................3,11,12,19rasters

composite..................................3,18,27grayscale....................................3,16,27RGB sets..................................3,17,27

tie points...............................................24-26tilt displacement.........................................13trend removal.............................................21

mosaicking raster geodata

Documents