Photogrammetry & Remote Sensing

Photogrammetry

Photogrammetry is the science and technology of obtaining reliable measurements, maps and 3D information of objects and terrain from photographs (and digital images).

Types

Aerial photographs

Terms

• Aerial camera — high-quality lens (e.g. focal length of about 150 mm for wide-angle mapping cameras), traditional film format 23 cm × 23 cm, now digital aerial cameras.
• Fiducial marks — marks on the edges or corners of the photograph whose lines join at the principal point.
• Principal point (p) — foot of the perpendicular from the lens centre to the photo plane.
• Nadir point (plumb point, n) — point vertically below the lens centre on the photograph.
• Isocentre (i) — point on the photograph bisecting the angle between the principal point and nadir; relevant for tilt displacement.
• In a truly vertical photograph, principal point, nadir and isocentre coincide.
• Flying height $H$ — altitude of the camera above datum (MSL).

Scale of a vertical photograph

Because ground elevations vary, the scale of a vertical photograph of hilly terrain varies from point to point (larger scale for higher ground).

Relief displacement

On a vertical photograph, objects above or below the datum are displaced radially from the nadir (principal) point — tops of tall objects appear displaced outward relative to their bases.

• Displacement is zero at the principal point and increases towards the edges.
• It increases with object height and decreases with flying height.
• Relief displacement enables height measurement from single photographs but distorts planimetric positions, which must be corrected (orthophotos).

Tilt displacement — occurs in tilted photographs, radial from the isocentre; points on the upper side of the photo are displaced inward and on the lower side outward.

Stereoscopy

Overlap

• Forward overlap (end lap) — overlap between successive photographs along a flight strip — commonly about 60% — enables stereoscopic viewing (each ground point appears in at least two photos).
• Side lap — overlap between adjacent strips — commonly about 25–30% — ensures no gaps and allows strip connection.
• Increased overlaps are used in mountainous terrain and for digital photogrammetry.

Stereoscopic vision and stereoscopes

A pair of overlapping photographs (stereo pair) viewed so that each eye sees one photograph produces a 3D (stereoscopic) model.

• Lens (pocket) stereoscope — simple, portable, small photo separation.
• Mirror stereoscope — mirrors allow full photographs to be viewed with magnification.
• Digital photogrammetric workstations use polarised or shutter glasses.

Parallax

Absolute stereoscopic parallax of a point = the difference in x-coordinates (along the flight line) of its images on the two photos: $p=x-x^{\prime }$. Parallax increases with ground elevation (points closer to the camera).

Remote sensing

Remote sensing is the science of obtaining information about objects or areas on the earth's surface without physical contact, by detecting and analysing electromagnetic radiation reflected or emitted by them — usually from aircraft or satellites.

The remote sensing process

1. Energy source (sun or the sensor itself).
2. Radiation and the atmosphere — scattering and absorption on the way down and up.
3. Interaction with the target — absorption, transmission, reflection, emission.
4. Recording by the sensor.
5. Transmission, reception and processing at ground stations.
6. Interpretation and analysis.
7. Application — maps, decisions.

Electromagnetic spectrum

Atmospheric windows — wavelength ranges where the atmosphere is relatively transparent (visible, parts of IR, thermal window, microwaves); sensors operate in these windows.

Spectral signatures

Different materials reflect differently across wavelengths:

• Healthy vegetation — low reflectance in blue and red (absorbed by chlorophyll), a peak in green, and very high reflectance in NIR (leaf structure).
• Water — low reflectance, especially absorbs NIR strongly (appears dark in NIR images); turbid water reflects more in visible.
• Soil — reflectance generally increases with wavelength; decreases with moisture and organic matter.

False colour composite (FCC) — standard FCC displays NIR as red, red as green and green as blue, so vegetation appears red.

Passive and active sensors

• Passive sensors — record natural energy (reflected sunlight or emitted thermal radiation): multispectral scanners, cameras, thermal sensors. Depend on sunlight (for reflected bands) and clear skies.
• Active sensors — emit their own energy and record the return:
  • RADAR / SAR (synthetic aperture radar) — microwave; penetrates clouds; works day and night; used for flood mapping, soil moisture, deformation (InSAR).
  • LiDAR (light detection and ranging) — laser pulses; very accurate elevation data, vegetation structure, DTMs.

Platforms and orbits

• Ground-based, airborne (aircraft, drones) and spaceborne (satellites) platforms.
• Geostationary orbit — about 35 786 km above the equator; satellite appears fixed relative to earth; continuous coverage of a large area — weather and communication satellites.
• Sun-synchronous polar orbit — low altitude (typically about 600–900 km), near-polar; passes over a place at the same local solar time — earth resources satellites.

Resolutions

Indian remote sensing programme

• IRS-1A (1988) was India's first operational remote sensing satellite; the Indian Remote Sensing (IRS) series has grown into one of the largest civilian constellations.
• Missions include Resourcesat (natural resources), Cartosat series (high-resolution stereo imagery for cartography), Oceansat (ocean studies), RISAT (radar imaging), and others.
• The National Remote Sensing Centre (NRSC), Hyderabad, acquires, processes and distributes data; the Bhuvan geoportal provides Indian satellite imagery and thematic maps.

Image interpretation and processing

Elements of visual interpretation

Tone/colour, size, shape, texture, pattern, shadow, and site/association (location relative to other features).

Digital image processing

1. Pre-processing:
   • Radiometric correction — sensor errors, atmospheric effects, illumination.
   • Geometric correction (georeferencing) — removing distortions and registering to map coordinates using ground control points; resampling (nearest neighbour, bilinear, cubic convolution).
2. Image enhancement — contrast stretching, filtering (smoothing, edge enhancement), band ratios, principal components, colour composites.
3. Image classification:
   • Supervised — analyst defines training sites; algorithms such as maximum likelihood, minimum distance, parallelepiped, and machine-learning classifiers.
   • Unsupervised — computer groups pixels into clusters (e.g. ISODATA, k-means), which the analyst labels.
   • Object-based classification — segments images into objects.
4. Accuracy assessment — error (confusion) matrix, overall accuracy, producer's and user's accuracy, kappa coefficient.
5. Change detection — comparing images of different dates.

Geographic Information System (GIS)

A GIS is a computer-based system to capture, store, query, analyse and display geographically referenced (spatial) data together with their attributes.

Components

Hardware, software, data (spatial and attribute), people and methods/procedures.

Data models

• Attribute data — stored in tables linked to features (database management systems); queried with SQL-like commands.
• TIN (triangulated irregular network) — vector representation of surfaces.

Data input

Digitising paper maps, scanning and vectorisation, GNSS/total station field data, remote sensing imagery, existing digital data (CAD, census tables), LiDAR point clouds. Metadata describes source, accuracy, projection and date.

Coordinate systems and projections

All data must share a coordinate reference system: geographic (latitude/longitude on a datum such as WGS 84) or projected (e.g. UTM zones — eastings and northings in metres). Georeferencing registers scanned maps and images to real-world coordinates.

Spatial analysis

Digital elevation models

DEM (bare-earth elevations, also DTM) and DSM (surface including buildings and trees). Sources: contours, stereo photogrammetry, LiDAR, radar interferometry — e.g. SRTM (about 30 m global), CartoDEM (from Cartosat stereo data).