9.2 Aerial Photogrammetry: Relief Displacement, Parallax, Orthorectification & Ground Control

9.2 Aerial Photogrammetry: Relief Displacement, Parallax, Orthorectification & Ground Control

Core Principle: A raw aerial photograph is a perspective projection, not a map. Scale varies across the image because of variations in terrain elevation and inadvertent aircraft tilt. Objects extending above the reference surface are displaced radially outward from the photo nadir. Transforming perspective imagery into an orthographic map requires orthorectification—a differential geometric correction using interior orientation, exterior orientation, and an elevation model to ensure uniform scale everywhere.

---

1. Photogrammetric Principles and Camera Geometry

Photogrammetry is the science, technology, and art of obtaining reliable three-dimensional geometric measurements and spatial information about physical objects and the environment through the process of recording, measuring, and interpreting photographic images and electromagnetic patterns.

Perspective Projection vs. Orthographic Projection

The fundamental geometric distinction governing photogrammetry is the difference between a perspective projection and an orthographic projection:

1. Perspective Projection (Raw Photograph): All light rays reflected from the three-dimensional terrain pass through a single point in space—the front nodal point (perspective center) of the camera lens—before exposing the sensor focal plane. As a result, scale is non-uniform across the photograph. Features closer to the camera (e.g., hilltops, tall buildings) appear at a larger scale than features farther away (e.g., valley floors), and elevated structures lean radially away from the center.
2. Orthographic Projection (True Map): All projection rays are mathematically parallel and perpendicular to the horizontal map projection plane. Scale is completely uniform throughout the entire extent of the map, and all features appear in their true planimetric positions $(X, Y)$ regardless of elevation.

          PERSPECTIVE PROJECTION (PHOTO)          ORTHOGRAPHIC PROJECTION (MAP)
          
                   [ Camera Lens ]
                       /  |  \                               |    |    |
                      /   |   \                              |    |    |
                     /    |    \                             |    |    |
                    /     |     \                            |    |    |
                   v      v      v                           v    v    v
                 Tower   Flat   Valley                     Tower Flat Valley
                (Leans) Ground (Smaller)                 (True) (True) (True)
                * Scale varies with elevation             * Scale uniform everywhere

Photographic Axis and Camera Tilt Classifications

Aerial photographs are classified based on the orientation of the camera's optical axis relative to the gravity plumb line:

• Vertical Aerial Photograph: The optical axis of the camera is oriented as nearly vertical as possible. In practice, minor aircraft turbulence causes slight tilt; an aerial photograph is technically classified as vertical if the tilt angle is less than $3^\circ$ from the vertical plumb line.
• Low Oblique Aerial Photograph: The camera optical axis is intentionally or unintentionally tilted significantly ($> 3^\circ$), but the tilt angle is shallow enough that the apparent horizon is not visible in the frame.
• High Oblique Aerial Photograph: The camera optical axis is tilted to such an extent that the apparent horizon is visible within the background of the image.

Key Geometric Reference Points on an Aerial Photograph

Every aerial photograph contains three fundamental geometric points that dictate how distortions propagate across the image plane:

                      GEOMETRIC REFERENCE POINTS (TILTED PHOTO)
   
   Fiducial Mark -------------------------------+------------------------------- Fiducial Mark
   |                                            |                                            |
   |                                     [PP] Principal Point                                |
   |                                     (Optical Axis / Intersection of Fiducials)          |
   |                                            |                                            |
   |                                            |                                            |
   |                                     [ I ] Isocenter                                     |
   |                                     (Origin of Tilt Displacement)                       |
   |                                            |                                            |
   |                                            |                                            |
   |                                     [ N ] Nadir Point                                   |
   |                                     (Plumb Point / Origin of Relief Displacement)       |
   |                                            |                                            |
   Fiducial Mark -------------------------------+------------------------------- Fiducial Mark
   
   * Note: In a truly vertical, untilted photograph (Tilt = 0°), PP, I, and N coincide at the exact same pixel.

1. Principal Point ($PP$): The geometric center of the photographic frame. On traditional film cameras, it is defined by the intersection of lines connecting opposite fiducial marks etched into the camera focal plane frame. In modern calibrated digital mapping cameras, it corresponds to the calibrated principal point coordinates $(x_0, y_0)$ where the optical axis intersects the sensor plane.
2. Nadir Point ($N$): The point where a vertical plumb line dropped from the camera's front nodal point intersects the photograph (photo nadir) and the ground surface (ground nadir). The nadir point is the center of radial relief displacement.
3. Isocenter ($I$): The point on the photograph that lies midway between the principal point and the nadir point along the line of tilt (the principal line). The isocenter is the origin and axis of radial tilt displacement.

Critical Photogrammetric Rule: In an idealized, perfectly vertical photograph (tilt $\theta = 0^\circ$), the optical axis is aligned with the plumb line. Therefore, the Principal Point, the Nadir Point, and the Isocenter coincide at the exact same physical point on the image.

---

2. Relief Displacement: Mechanics and Height Determination

Relief displacement (topographic displacement) is the radial shift in the photographic position of an image feature caused by that feature's elevation above or below the chosen reference datum.

Because an aerial camera captures an image via central perspective projection, light rays originating from elevated objects enter the lens at oblique angles. Consequently:

• Objects that extend above the reference datum (such as mountaintops, buildings, trees, and transmission towers) are displaced radially outward away from the nadir point.
• Depressions located below the reference datum (such as river canyons and open pit mines) are displaced radially inward toward the nadir point.

                          RELIEF DISPLACEMENT GEOMETRY
   
                                 [ Camera Lens ] (Flying Height H above Datum)
                                      /|\
                                     / | \
                                    /  |  \
                 Focal Length (f)  /   |   \   [ Negative / Sensor Plane ]
                                  /----+----
                                 /   r | d  \
                                /      |     \
                               /       |      \
                              /        |       \
                             /         |        \
                            /          |         \
                           /           |          \
                          /            |           \
                         * [Top]       |            \
                        /|             |             \
             Height (h)/ |             |              \
                      /  |             |               \
                     *---+-------------+----------------* [Ground Datum]
                   [Base]            [Nadir]
                               <------- R ------->

The Mathematical Relief Displacement Equation

In a vertical aerial photograph, relief displacement is mathematically formulated as:

$d = \frac{r \cdot h}{H}$

Where:

• $d$ = magnitude of relief displacement measured on the photograph (in millimeters or pixels) from the image of the object's base to the image of its top.
• $r$ = radial distance on the photograph measured from the principal point / nadir to the displaced image point (the top of the object).
• $h$ = height of the object above the reference ground datum (in meters or feet).
• $H$ = flying height of the aircraft above the same reference ground datum (in meters or feet).

Proportionality Rules of Relief Displacement

1. Directly Proportional to Object Height ($h$): A 100-meter building exhibits twice the relief displacement of a 50-meter building located at the same radial distance from nadir.
2. Directly Proportional to Radial Distance ($r$): An object situated near the edge of the photographic frame experiences significantly greater displacement than an identical object located near the photo center. At the exact nadir point ($r = 0$), relief displacement is zero ($d = 0$).
3. Inversely Proportional to Flying Height ($H$): Increasing the flying altitude of the aircraft above the ground diminishes relief displacement across the entire frame. Imagery acquired from high altitudes or satellite orbits exhibits substantially less relief displacement than low-altitude aerial photography.

Calculating Object Height from Relief Displacement

Rearranging the relief displacement formula enables photogrammetrists to determine the physical height ($h$) of tall vertical structures (such as skyscrapers, cooling towers, and antennas) directly from a single vertical photograph, provided both the top and base of the structure are visible:

$h = \frac{d \cdot H}{r}$

Worked Example: Calculating Structure Height

An aerial survey aircraft flies at an altitude of $2{,}400\text{ meters}$ above mean terrain datum ($H = 2{,}400\text{ m}$). On a vertical photograph, an analyst measures a communications broadcast tower. The radial distance from the principal point to the top of the tower is measured as $r = 75.0\text{ mm}$. The displacement between the base of the tower and the top of the tower is measured as $d = 3.75\text{ mm}$. What is the physical height of the tower?

$h = \frac{d \cdot H}{r} = \frac{3.75\text{ mm} \cdot 2{,}400\text{ m}}{75.0\text{ mm}} = \frac{9{,}000}{75.0} = 120.0\text{ meters}$

---

3. Photographic Scale Calculations

Unlike an engineering map which possesses a singular, invariant scale, an aerial photograph exhibits a variable scale that fluctuates across the frame in direct response to changes in surface terrain elevation.

Scale Formulas

1. Scale over Flat Terrain at Datum:

$S = \frac{f}{H}$

1. Scale at a Specific Elevation ($h_g$ above datum):

$S = \frac{f}{H - h_g}$

Where:

• $S$ = photographic scale (dimensionless representative fraction, e.g., $1:20{,}000$).
• $f$ = calibrated focal length of the camera lens (e.g., $152.4\text{ mm}$ or $6\text{ inches}$). Units must match $H$.
• $H$ = flying height of the aircraft above the datum.
• $h_g$ = ground elevation of the specific feature above datum.

Worked Example: Scale Variations Over Mountainous Terrain

A metric camera with a calibrated focal length $f = 150\text{ mm}$ ($0.150\text{ m}$) flies at an absolute altitude of $4{,}500\text{ m}$ above sea level ($H_{\text{MSL}} = 4{,}500\text{ m}$). The photographic flight crosses a mountain ridge with a peak elevation of $1{,}500\text{ m}$ and an adjacent river valley at an elevation of $500\text{ m}$. Calculate the photo scale at both locations:

• Scale on Mountain Peak ($h_g = 1{,}500\text{ m}$):

$S_{\text{peak}} = \frac{0.150\text{ m}}{4{,}500\text{ m} - 1{,}500\text{ m}} = \frac{0.150}{3{,}000} = \frac{1}{20{,}000} \quad (1:20{,}000)$

• Scale in River Valley ($h_g = 500\text{ m}$):

$S_{\text{valley}} = \frac{0.150\text{ m}}{4{,}500\text{ m} - 500\text{ m}} = \frac{0.150}{4{,}000} = \frac{1}{26{,}667} \quad (1:26{,}667)$

The scale on the mountain peak is significantly larger ($1:20{,}000$) than in the valley floor ($1:26{,}667$), meaning an acre of land on the mountain occupies more physical pixels on the sensor than an acre in the valley.

---

4. Stereoscopic Parallax and 3D Elevation Extraction

Human binocular vision perceives depth because our two eyes view objects from slightly different vantage points. In photogrammetry, stereoscopic parallax is the apparent displacement of the position of an object relative to a reference frame, caused by a change in the observation point (the camera position moving along a flight line).

Stereoscopic Flight Planning Overlap Requirements

To view aerial photographs stereoscopically and extract three-dimensional terrain models, flight plans must incorporate substantial photographic overlap:

• Forward Overlap (End-lap): Successive exposures along the same flight line must overlap by $60%$ to $70%$ (standard specification is $60%$, often boosted to $75-80%$ for dense urban or steep terrain). The overlapping area between two consecutive photos forms a stereopair, ensuring that every point on the ground is imaged from at least two distinct camera stations.
• Sidelap (Lateral Overlap): Adjacent parallel flight lines must overlap by $20%$ to $30%$ to prevent coverage gaps between flight lines caused by aircraft drift or navigation errors.

                       STEREOSCOPIC OVERLAP GEOMETRY
   
   Flight Line Direction ---------------------------------------------------->
   
   [ Photo 1 Station ]                       [ Photo 2 Station ]
           | <-------------- Air Base (B) -------------> |
           v                                             v
   +---------------+                             +---------------+
   |               |                             |               |
   |         +-----+-----------------------+-----+               |
   |         |     |                       |     |               |
   | Photo 1 |     |  Stereopair Overlap   |     | Photo 2       |
   |         |     |       (60% - 70%)     |     |               |
   |         +-----+-----------------------+-----+               |
   |               |                             |               |
   +---------------+                             +---------------+

The Air Base and Base-to-Height ($B/H$) Ratio

• Air Base ($B$): The ground distance separating two consecutive exposure stations along the flight path.
• Base-to-Height Ratio ($B/H$): The ratio of the air base ($B$) to the flying height ($H$) above ground. Typical aerial photogrammetric surveys exhibit a $B/H$ ratio of approximately $0.55$ to $0.65$.
• Vertical Exaggeration: Because the human eye base ($\approx 65\text{ mm}$) to viewing distance ($\approx 250\text{ mm}$) ratio is approximately $0.26$, viewing an aerial stereopair with a $B/H$ ratio of $0.6$ through a stereoscope exaggerates vertical relief by a factor of 2 to 4, making subtle terrain undulations clearly visible.

Mathematical Parallax Formulation

In a coordinate system where the positive $x$-axis is oriented along the flight line between exposure stations:

$p = x - x'$

Where:

• $p$ = absolute stereoscopic parallax of a ground point.
• $x$ = coordinate of the point on the left photograph measured from the principal point along the flight axis.
• $x'$ = coordinate of the identical ground point on the right photograph measured from its principal point along the flight axis (usually a negative value, so $x - x'$ becomes an addition of absolute distances).

Points at higher elevations exhibit larger stereoscopic parallax than points at lower elevations. The differential parallax ($\Delta p = p_{\text{top}} - p_{\text{bottom}}$) between the top and bottom of an object enables calculation of the object's height ($\Delta h$):

$\Delta h = \frac{H' \cdot \Delta p}{p_{\text{top}}}$

Where $H'$ is the flying height above the base of the object.

---

5. The Orthorectification Process

Orthorectification is the rigorous computational process of removing the geometric distortions introduced by terrain relief displacement, camera tilt, and optical lens aberrations from perspective imagery, converting it into a planimetrically accurate orthophoto.

An orthophoto combines the image characteristics of a photograph with the geometric integrity of a map. Distances, angles, and areas can be measured directly on an orthophoto without correcting for scale distortion.

                       THE ORTHORECTIFICATION PIPELINE
   
   [ Raw Perspective Images ]
              |
              v
   +-------------------------------------------------------------------------+
   | Mathematical Differential Rectification (Collinearity Equations)       |
   |                                                                         |
   | Inputs Required:                                                        |
   | 1. Interior Orientation (IO): Camera Calibration Report                 |
   |    - Calibrated Focal Length (f)                                        |
   |    - Principal Point Coordinates (x0, y0)                               |
   |    - Radial & Tangential Lens Distortion Parameters (K1, K2, P1, P2)   |
   |                                                                         |
   | 2. Exterior Orientation (EO): Direct Georeferencing & BBA               |
   |    - 3D Position of Camera Center: (XL, YL, ZL) [GNSS]                  |
   |    - Angular Attitude: Roll (ω), Pitch (ϕ), Yaw/Kappa (κ) [IMU]         |
   |                                                                         |
   | 3. Ground Control Points (GCPs): Surveyed targets (X, Y, Z)             |
   |                                                                         |
   | 4. Digital Elevation Model (DEM / DTM): Terrain heights (Z)             |
   +-------------------------------------------------------------------------+
              |
              v
   [ Resampling Algorithms: Nearest Neighbor, Bilinear, or Cubic Convolution ]
              |
              v
   [ Orthorectified Image (True Map Projection & Uniform Planimetric Scale) ]

The Collinearity Equations

The mathematical foundation of modern digital photogrammetry is the Collinearity Condition, which states that the perspective center of the camera lens ($L$), any image point ($a$), and its corresponding ground point ($A$) must all lie along a single, unbroken straight line in three-dimensional space.

x - x_0 &= -f \left[ \frac{m_{11}(X - X_L) + m_{12}(Y - Y_L) + m_{13}(Z - Z_L)}{m_{31}(X - X_L) + m_{32}(Y - Y_L) + m_{33}(Z - Z_L)} \right] \\[6pt] y - y_0 &= -f \left[ \frac{m_{21}(X - X_L) + m_{22}(Y - Y_L) + m_{23}(Z - Z_L)}{m_{31}(X - X_L) + m_{32}(Y - Y_L) + m_{33}(Z - Z_L)} \right] \end{aligned}$$ Where: - $(x, y)$ are measured photographic coordinates. - $(x_0, y_0)$ are calibrated principal point coordinates. - $f$ is calibrated focal length. - $(X, Y, Z)$ are real-world ground coordinates. - $(X_L, Y_L, Z_L)$ are ground coordinates of the camera perspective center. - $m_{ij}$ are elements of the $3 \times 3$ rotation matrix ($M$) computed from the three rotational orientation angles: **Roll ($\omega$)**, **Pitch ($\phi$)**, and **Yaw/Kappa ($\kappa$)**. ### Essential Parameters for Orthorectification 1. **Interior Orientation (IO):** Reconstructs the internal geometry of the camera at the instant of exposure, obtained from a rigorous lab calibration certificate: - Calibrated focal length ($f$). - Principal point offset $(x_0, y_0)$. - Radial lens distortion coefficients ($K_1, K_2, K_3$). - Decentering/tangential lens distortion coefficients ($P_1, P_2$). 2. **Exterior Orientation (EO):** Defines the spatial position and angular orientation of the camera in the ground coordinate system: - **Position (3 Parameters):** $X_L, Y_L, Z_L$ determined via airborne differential kinematic GNSS. - **Attitude / Rotation (3 Parameters):** Omega ($\omega$, roll around flight axis), Phi ($\phi$, pitch around cross-track axis), and Kappa ($\kappa$, yaw or rotation around optical axis) measured by an onboard Inertial Measurement Unit (IMU). 3. **Ground Control Points (GCPs):** Highly conspicuous, surveyed ground features with precisely known coordinates $(X, Y, Z)$ established via geodetic GNSS receivers. GCPs are used in a simultaneous mathematical **Bundle Block Adjustment (BBA)** to refine exterior orientation parameters, resolve systematic camera drift, and tie the photo block to the terrestrial datum. 4. **Check Points:** Independent surveyed ground targets held out of the bundle adjustment. Check points are used strictly to audit and report the absolute horizontal and vertical accuracy (Root Mean Square Error - RMSE) of the final orthophoto product according to ASPRS or NSSDA accuracy standards. 5. **Digital Elevation Model (DEM/DTM):** Provides the elevation value ($Z$) for every ground coordinate $(X, Y)$, enabling the orthorectification algorithm to solve the collinearity equations backward and project each orthophoto pixel onto its correct location in the raw image. --- ## 6. Traditional Orthophotos vs. True Orthophotos A critical technical development tested on the GISP exam is the distinction between **traditional orthophotos** and **true orthophotos**, particularly in dense urban environments containing tall man-made structures. ``` TRADITIONAL ORTHO vs. TRUE ORTHO TRADITIONAL ORTHOPHOTO (Uses Bare-Earth DTM): - Terrain relief displacement is removed. - Tall buildings still experience radial displacement (Building Lean). - Rooftops are displaced away from nadir, obscuring streetscapes behind them. - Street geometry behind tall structures is hidden in "Blind Spots" (Occlusion). TRUE ORTHOPHOTO (Uses High-Density DSM + Multi-Ray Photography): - Terrain AND above-ground structures are rectified. - Building displacement is corrected so rooftops align more closely with footprints. - Occluded areas behind buildings are filled using view rays from overlapping photos. - True planimetric view of all streets, sidewalks, and parcel boundaries. ``` ### Traditional Orthophotos Traditional orthophotos use a **Digital Terrain Model (DTM)**—a bare-earth surface model that reflects ground elevations while removing all above-ground features (buildings, bridges, vegetation). - **Consequences:** Because the elevation model only accounts for the ground surface, any feature extending above the ground remains subject to perspective relief displacement. - **Building Lean:** Skyscrapers and bridges lean radially outward from the nadir point of the photograph. The taller the building and the farther it sits from the center of the frame, the worse the lean. - **Occlusion (Blind Spots):** As a building leans outward, its displaced facade and roof obscure the terrain, streets, sidewalks, underground utilities, and property parcel boundaries situated directly behind it. Orthorectification cannot display obscured features because they were blocked from the camera's single line of sight. ### True Orthophotos A **true orthophoto** eliminates building lean and perspective occlusions entirely, depicting all features—both terrain and structures—in their exact planimetric locations. - **Requirements for True Ortho Production:** 1. **High-Resolution Digital Surface Model (DSM):** A surface model (typically derived from dense airborne LiDAR or dense image matching) that captures the exact three-dimensional geometry of every building roofline, parapet, overpass, and tree crown. 2. **High Forward and Lateral Image Overlap:** Flight plans require **$80\%$ forward overlap and $70-80\%$ sidelap** so that every patch of urban ground is imaged from multiple divergent angles across multiple flight lines. 3. **Multi-Ray Visibility and Occlusion Analysis:** Sophisticated ray-tracing algorithms identify areas occluded by tall structures in one image and patch those blind spots using nadir or near-nadir lines of sight extracted from overlapping adjacent photographs. - **Result:** In a true orthophoto, building rooftops are projected directly over their physical foundation footprints; street-level infrastructure is visible across the entire urban corridor; and vector parcel boundary lines align with building foundations. | Dimension | Traditional Orthophoto | True Orthophoto | | :--- | :--- | :--- | | **Elevation Model Utilized** | Bare-Earth Digital Terrain Model (DTM) | Digital Surface Model (DSM) including all structures | | **Handling of Buildings / Bridges** | Uncorrected; structures lean radially away from nadir | Fully corrected; structures rendered in true vertical planimetry | | **Urban Occlusion / Blind Spots** | Significant; streets behind tall buildings are obscured | Eliminated; occluded ground filled from adjacent overlapping photos | | **Parcel Line / Footprint Alignment** | Displaced; roof does not match cadastral parcel line | Exact; building footprint coincides with property parcel boundary | | **Flight Overlap Requirements** | Standard ($60\%$ forward overlap, $20-30\%$ sidelap) | Dense ($75-80\%+$ forward overlap, $70-80\%$ sidelap) | | **Computational Complexity** | Moderate; standard single-pass differential rectification | High; requires 3D mesh modeling, ray tracing, and ghosting removal | | **Primary Use Cases** | Regional rural mapping, forestry, agriculture, county base maps | Dense urban core GIS, cadastral mapping, asset management, utilities | --- ## 7. Common GISP Exam Traps & Pitfalls > [!CAUTION] > **Exam Trap 9.2.1: Confusing the Principal Point with the Nadir Point on Tilted Photos.** > Candidates frequently confuse the principal point and nadir point. Remember: the **Principal Point ($PP$)** is the geometric optical center of the photograph defined by fiducial marks. The **Nadir Point ($N$)** is the point directly vertically beneath the camera lens along the gravity vector. They coincide **only on a perfectly untilted vertical photograph**. Relief displacement is strictly radial from the **nadir point**, whereas tilt displacement is radial from the **isocenter**. > [!CAUTION] > **Exam Trap 9.2.2: Misapplying the Relief Displacement Formula.** > When calculating object height ($h = \frac{d \cdot H}{r}$), ensure that radial distance ($r$) and displacement ($d$) are measured in the **exact same units** (e.g., both in millimeters or both in inches) so that they cancel out, leaving the resulting object height in the units of flying height ($H$). Additionally, remember that $r$ is measured to the **top** of the displaced object, not its base. > [!CAUTION] > **Exam Trap 9.2.3: Assuming Traditional Orthophotos Have No Distortion.** > Many GIS users assume that an orthophoto has zero distortion everywhere. While a traditional orthophoto has uniform scale across the ground terrain, it still suffers from **radial relief displacement of tall man-made features (building lean)** because it was rectified using a bare-earth DTM. If an exam question mentions buildings leaning over streets and obscuring parcel lines in an orthophoto, that is a traditional orthophoto, not a true orthophoto. > [!CAUTION] > **Exam Trap 9.2.4: Conflating Interior and Exterior Orientation Parameters.** > Interior Orientation (IO) refers to internal camera geometry (focal length, principal point, lens distortion) established during laboratory camera calibration. Exterior Orientation (EO) refers to the camera's spatial position ($X, Y, Z$) and angular orientation (roll $\omega$, pitch $\phi$, yaw $\kappa$) in the real world at the moment of exposure, measured by airborne GNSS and IMU sensors.