Photogrammetric

Procedure
Unit 7
DEM (Digital Elevation model)
• The surface of the earth can be modelled effectively in digital
environment.
• The digital representation of the elevation of the earth surface is called as
digital elevation model.
• Change of elevation of the ground surface with position.
• DEM distinctively used to indicate how the surface is modelled, elevation
is some times used for height of another surface than the ground surface
terrain. Sometimes distinctively used to indicate that a surface is modelled
by elevation values. So it is a general term used to represent a elevation
model which doesn’t tell specifically, elevation of what is modelled.
• DTM (Digital Terrain Model)is the origin term for terrain relief i.e the
shape of the bare ground surface.
• DHM (Digital Height
Model) mostly used as
synonym for DEM.
• DGM (Digital Ground
Model) Mostly used as
synonym for DTM.
• DSM (Digital Surface
Model) Refers a surface
model which includes the
natural/artificial objects
on the surface. Sometimes
used as synonym for DEM.
Techniques of DTM generation
• Digitizing existing maps
• Digitizing contour maps
• Less accurate
• Field Survey
• GPS, Tacheometry, Levelling
• Highly accurate
• Laser Scanning
• Airborne laser Scanning
• Highly accurate
• Photogrammetry
• Digitizing, image matching
• High- medium level of accuracy
Photogrammetric method of DTM
generation
Photographs

Scanning

Orientation (Interior, Relative, Absolute, AT)

3D restitution (stereo model)

Point cloud Generation

DTM

Editing

Final DTM
Scanning: - (in case for hardcopy photos)
• Scan with photogrammetric scanner
• Preserve perspective geometry
• Define scan resolution
• Large scale = near 2400 dpi
• Small scale = near 1600 dpi
• For more dpi, large file size takes time requires high operating system.
Interior orientation: -
• Load cam file for rotation parameters and calibration files
• Read fiducials
• RMSE check of fiducials
• 1/3rd of pixel size
• Large scale input
• Small pixel size – minimum tolerance and vice versa
Exterior orientation: -
• Input GCPS/RPC files
• Generate Tie points
• Check RMSE, if both within tolerance go ahead
• If not repeat
3D restitution: -
• Creating 3D mode as the surface was during photograph.
• Working with stereo pairs, stereo pair orientation is necessary for
DTM/ DSM generation
Point cloud Generation: -
• After stereo restitution point surfacing is done which is automatic
surface induced process in which the surface is created by point
clouds.
• For this: -
• Algorithms need to be assigned (this also determines DTM accuracy)
• These algorithm depends on users.
• During this process , software creates point at corner of different
features like house roofs.
• Such super elevated features are corrected during editing.
• It uses “search window” either of 3 x 3 or 5x5 which is also defined by
user themselves.
• Then, it places points based on pixel correlation.
Editing
• Most complex and tedious part.
• The generated points may float on roof top areas, tree without
remaining on ground.
• Such points need to be moved, deleted or new points are to be
generated.
• Every points are visited for editing.
• Corner of depression and elevation may be generated.
• In forest area like upper chure and mahabharat range 90% points
needs to be edited.
• Hence it depends on land use, land cover and vegetation as well.
 Image Matching
• It refers to the automatic identification and measurement of
corresponding image points that are located on the
overlapping areas of multiple images. The various Image
matching methods can be divided into three categories.
• Area based Matching
• Feature based Matching
• Relation based Matching
• Area based matching:- It is also called signal-based matching. A target/template
window is moved across a larger window pixel by pixel to find the position of
maximum similarity. Cross correlation and Least Square correlation techniques
are well known methods for area based matching. Cross correlation only
considers the gray value intensity for comparing the pixels in template and
search window. Least Square correlation template looks for gray value change
and translation as well as the rotational parameters and scale change between
the image patch.
• Feature based matching: - Determines the correspondence between two image
features. Distinct features like points, edges and patches are independently
extracted in all Images. They are matched with the corresponding feature
entities of the other image. Feature attributes and consistency in location of
corresponding feature is compared
• Relation based matching: -It is also called as structural matching. This technique
uses the image features and the relationship between the features. With this
matching, the corresponding image structures can be recognized automatically
without any prior information. The process is much time consuming.
 Orthophoto
• A digital orthophoto is a digital image, produced from an aerial or
satellite photograph/image in which image distortions caused by
camera tilt and terrain relief were removed.
• Correction for tilt displacement can be achieved by rectification where
as the correction for relief displacement is achieved by differential
rectification to the orthogonal projection using a DTM.
• An orthophoto map is an orthophoto + cartographic features
“annotated image map”
• A mosaic is an assembly of two or more orthophotos.
• An orthophoto differs from a standard rectified photograph in that only
tilt has been removed from a rectified photo, whereas topographic
displacement is also removed in an orthophoto.
• Ordinary rectification is accomplished by copying the non rectified print
or negative with a camera that is tilted relative to the original negative
in such a way that the new photo is free of tilt.
• Differential rectification as used to produce ortho photos also requires the re-
photographing of the original imagery.
• However, instead of re-photographing the entire image all at once, it is done in
very small segments (in thousands for a 9 in. × 9 in. photo) or by a series of
narrow strips. These segments or strips are all brought to a common scale by
keeping the scanning slit continually in contact with a three-dimensional terrain
model.
• The resulting orthophoto is planimetrically correct, which allows one to make
accurate measurements of distance, area, a direction directly on the orthophoto.
A theoretically perfect orthophoto can be achieved only if each infinitely small
• point in the stereoscopic model is exposed separately. Ortho photoscopes using
scanning slits of about 5 mm in size can therefore only approximate a perfect
orthophoto. However, for all practical purposes, this approximation is entirely
satisfactory. The size of the slit and the scanning speed can be altered to achieve
the results desired.
• Smaller slits and slower scanning rates should be used for rugged topography
than for flatter terrain.
Index Mosaics
• An index mosaic is usually prepared for the purpose of providing an
index to individual photographs. The overlap of the individual
photographs is purposely not trimmed so that the indexing numbers
(flight line or roll number and photo numbers) are clearly visible.
From such a mosaic one can quickly determine which photographs
cover a particular piece of ground—thus, the primary use is for
indexing. Index mosaics are usually produced at additional cost for
most photo missions but are the least expensive type of mosaic
because they are uncontrolled and not permanently mounted on a
backing. Figure 7.6 is a partial index mosaic of a section of forested
land.
Photo Mosaic
A photo mosaics is an assembly of vertical aerial photographs in
which the edge have been cut and matched and fitted together
systematically to form a continuous photographic representation of a
portion of the Earth’s surface.
• Mosaics are usually reproduced at a much smaller scale than the
original photography.
• There are three principal types: index, controlled, and uncontrolled.
Types of Mosaic
1. Controlled Mosaic: Control Points and Rectified Photos are used
2. Uncontrolled Mosaic: By matching features
3. Semi Controlled Mosaic: Control Points are used but not rectified
photos
 Controlled Mosaics
• controlled mosaic is the most accurate of the three classes.
• Sometimes it is convenient to view several photographs covering a large area in a single
view. To do this, two or more photographs are trimmed so that the overlap duplication is
eliminated and are permanently mounted on a hard backing so that the photo detail is
matched as well as possible.
• In a controlled mosaic, several points are located on the ground and precisely located on
the photographs. Distances and bearings among the ground control points are measured
and drawn to scale on the mosaic backing.
• The photos are then glued to this backing so that the control points on the photographs
are directly over the corresponding ground control points on the backing. Ratioed and
rectified prints are usually used.
• Mosaics constructed with limited ground control are semi controlled mosaics. Additional
control points can be established using the photos alone by an older process called
radial line triangulation or by the use of more sophisticated mapping instruments.
Steps of Making Controlled
Mosaics
• Rectify all the photographs into a desire scale
• Mark control points on each of the photographs
• Mark trim lines
• Peel off part of the area to be matched
• Try to fit the photographs using identical control points of the
adjacent photographs
• Check the matching of the details
• If they are matched, paste together the adjacent photographs using
mosaics tape.
Uncontrolled Mosaics
• Because controlled mosaics are expensive, most mosaics are uncontrolled.
• Uncontrolled mosaics are put together by matching the detail from one photo to
the next as best as possible.
• There is no ground control, and aerial photographs that have not been rectified or
ratioed are used.
• Because of tilt and topographic displacement, measurements on these mosaics are
not reliable.
• Uncontrolled mosaics are more easily and quickly prepared than controlled mosaics.
• They are not as accurate as controlled mosaics, but for many qualitative uses they
are completely satisfactory.
• The errors involved from one photo to the next are usually additive and not
compensating.
Semi Controlled Mosaic
• Semi-controlled mosaics are assembled by utilizing some combinations
of the specifications for controlled and uncontrolled mosaics.
• A semi-controlled mosaic may be prepared, for example, by using
ground control but employing photos that have not been rectified or
ratioed.
• The other combination would be to use rectified and ratioed photos
but no ground control.
• Semi-controlled mosaics are a compromise between economy and
accuracy.
Advantages
• A mosaic shows as a single picture of all details which cannot be
depicted on a map.
• Features are represented in their natural appearance and dimension
• A mosaic can be easily understood by users who cannot read a map.
• It is a rapid and cheap method of producing map information
Disadvantages
• Excessive details
• Lack of topographical information like, names elevation and contours
• There could be errors due to relief and scale variations.
Factors Affecting Quality of mosaic
• Degree of Height Relief: In a flat terrain, the scale variation will be
small and the mosaic is comparable to a map whereas terrain having
height relief causes scale variation and leads to mismatches.
• Quality of Photographs: good camera and high quality photographic
materials are necessary for getting good prints.
• Quality of Flying: Excessive tilts and variations in height of aircraft
cause scale variation.
• Skill of the Compiler: Trimming, detail matching and pasting of the
photograph need good skill to get a reasonably good mosaic.
Aerial vs Orthophoto
Aerial photo
Perspective/central projection
No direct measurement
Relief and tilt displacement
No uniform scale throughout the photo
No coordinate system is assigned
Original radiometric property in the photo
Orthophoto
Orthogonal projection
Map like geometry, coordinate system
Direct measurements can be made
Uniform scale
Distortion due to tilt and relief are removed
Radiometric property on the photo is slightly distorted due to the resampling
Feature Extraction
• To collect or extract the required topographic or plannimentric
features from photograph.
• To obtain meaningful descriptions of selected parts of the world.
• Mainly topographic features i.e. natural as well as artificial features.
Feature Extraction
• Main steps in feature extraction are
• Image interpretation
• Detail classification
• Digitization (2D or 3D)
Feature Extraction
• Image interpretation and identification
• Spontaneous recognition
• familiar detail type which can be identified easily e.g. building, river
• Recognise the details without any difficulty because they are familiar
• Logical inference
• detail which are not easily identified are interpreted by applying logical inference e.g.
colour, shape, size, pattern, texture, orientatin/location, association etc.
• Needs interpretation elements or clues and logic for recognising the detail.
Interpretation elements
• Shape
– Shape of the detail on the photo, e.g. Linear, rectangular
• Size
• Size of the object, e.g. Big , small
• Colour (hue, value)
– Colour, brightness, intensity values etc of the detail, e.g. Red, dark blue, light green
• Texture
– Surface roughness or smoothness of the detail
• Pattern
– Homogeneous structures or repeating patterns, e.g. Agricultural fields, regularly spaced building
structures
• Association
– Attachment or association of one detail with other, together help to identify the detail, e.g. Big building
and open ground with some recreation and sports structure could be a school/college
• Shadow
– Height of a detail and shadow casted by it can be helpful to recognise a detail
 Feature Extraction
• Classification
• There may be different types of details on the photographs
• They should be classified for our simplicity
• Features on the photo should be classified according as their
common characteristics
• e.g. Differenrent types of buildings like residential, commercial,
industrial, school, hospitals can be put together in a class
”Building”
• Similarly, highway, motorable, cycle track, foot trails can be
catogarised as ” Road”
• As these extracted information can be used as GIS layers, so
this type of clasification plays important role for further data
management.
Feature Extraction
• Digitizing
• Some specification should be developed, what to extract, how to extract
• Digitizing rules should be followed; point, line area
• The features should be digitized under the same feature classes
• Appropriate attributes should be given,
Methods of feature extraction
• Manual
• Digitizing in photograph
• Digitizing over orthophoto
• Mono plotting
• Using a single photograph or orthophoto, known as monoplotting
• Not much accurate
• Accuracy depend upon DEM used
• Easy and faster
• Stereo digitizing
• Using stereo pair of photographs or stereo resstitution
• Good positional accuracy
• 3D view
• Complex
• (Semi) Automatic feature extraction
• Supervised
• Unsupervised
• Object oriented Image Analysis
Feature Extraction
• Application
• Extracted features can be used
• For topographic mapping,
• Image mapping
• GIS vector layer for different purposes