Photogrammetry is the "art, science and technology of obtaining reliable information about physical objects and the environment through the process of recording, measuring and interpreting photographic images and patterns of electromagnetic radiant imagery and other phenomena" (American Society of Photogrammetry, 1980).
Photogrammetry was invented in 1851 by Laussedat, and has continued to develop. Over time, the development of photogrammetry has passed through the phases of plane table photogrammetry, analog photogrammetry, analytical photogrammetry, and into the phase of digital photogrammetry (Konecny, 1994).
The traditional, and largest, application of photogrammetry is to extract topographic information (for example, topographic maps) from aerial images. However, photogrammetric techniques have also been applied to process satellite images and close range images in order to acquire topographic or nontopographic information of photographed objects.
Prior to the invention of the airplane, photographs taken on the ground were used to extract the relationships between objects using geometric principles. This was during the phase of plane table photogrammetry.
In analog photogrammetry, starting with stereomeasurement in 1901, optical or mechanical instruments were used to reconstruct three-dimensional geometry from two overlapping photographs. The main product during this phase was topographic maps.
In analytical photogrammetry, the computer replaces some expensive optical and mechanical components. The resulting devices were analog/digital hybrids. Analytical aerotriangulation, analytical plotters, and orthophoto projectors were the main developments during this phase. Outputs of analytical photogrammetry can be topographic maps, but can also be digital products, such as digital maps and DEMs.
Digital photogrammetry is photogrammetry as applied to digital images that are stored and processed on a computer. Digital images can be scanned from photographs or can be directly captured by digital cameras. Many photogrammetric tasks can be highly automated in digital photogrammetry (for example, automatic DEM extraction and digital orthophoto generation). Digital photogrammetry is sometimes called soft-copy photogrammetry. The output products are in digital form, such as digital maps, DEMs, and digital orthophotos saved on computer storage media. Therefore, they can be easily stored, managed, and applied by you. With the development of digital photogrammetry, photogrammetric techniques are more closely integrated into remote sensing and GIS.
Digital photogrammetric systems employ sophisticated software to automate the tasks associated with conventional photogrammetry, thereby minimizing the extent of manual interaction required to perform photogrammetric operations. IMAGINE Photogrammetry Project Manager is such a photogrammetric system.
Photogrammetry can be used to measure and interpret information from hard-copy photographs or images. Sometimes the process of measuring information from photography and satellite imagery is considered metric photogrammetry, such as creating DEMs. Interpreting information from photography and imagery is considered interpretative photogrammetry, such as identifying and discriminating between various tree types as represented on a photograph or image (Wolf, 1983).
Types of Photographs and Images
The types of photographs and images that can be processed within IMAGINE Photogrammetry Project Manager include aerial, terrestrial, close range, and oblique. Aerial or vertical (near vertical) photographs and images are taken from a high vantage point above the Earth’s surface. The camera axis of aerial or vertical photography is commonly directed vertically (or near vertically) down. Aerial photographs and images are commonly used for topographic and planimetric mapping projects. Aerial photographs and images are commonly captured from an aircraft or satellite.
Terrestrial or ground-based photographs and images are taken with the camera stationed on or close to the Earth’s surface. Terrestrial and close range photographs and images are commonly used for applications involved with archeology, geomorphology, civil engineering, architecture, industry, and so forth.
Oblique photographs and images are similar to aerial photographs and images, except the camera axis is intentionally inclined at an angle with the vertical. Oblique photographs and images are commonly used for reconnaissance and corridor mapping applications.
Digital photogrammetric systems use digitized photographs or digital images as the primary source of input. Digital imagery can be obtained from various sources. These include:
- Digitizing existing hard-copy photographs
- Using digital cameras to record imagery
- Using sensors on board satellites such as Landsat and SPOT to record imagery
This document uses the term imagery in reference to photography and imagery obtained from various sources. This includes aerial and terrestrial photography, digital and video camera imagery, 35 mm photography, medium to large format photography, scanned photography, and satellite imagery.
Why use Photogrammetry?
As stated in the previous section, raw aerial photography and satellite imagery have large geometric distortion that is caused by various systematic and nonsystematic factors. The photogrammetric modeling based on collinearity equations eliminates these errors most efficiently, and creates the most reliable orthoimages from the raw imagery. It is unique in terms of considering the image-forming geometry, utilizing information between overlapping images, and explicitly dealing with the third dimension: elevation.
In addition to orthoimages, photogrammetry can also provide other geographic information such as a DEM, topographic features, and line maps reliably and efficiently. In essence, photogrammetry produces accurate and precise geographic information from a wide range of photographs and images. Any measurement taken on a photogrammetrically processed photograph or image reflects a measurement taken on the ground. Rather than constantly go to the field to measure distances, areas, angles, and point positions on the Earth’s surface, photogrammetric tools allow for the accurate collection of information from imagery. Photogrammetric approaches for collecting geographic information save time and money, and maintain the highest accuracies.
Photogrammetry/ Conventional Geometric Correction
Conventional techniques of geometric correction such as polynomial transformation are based on general functions not directly related to the specific distortion or error sources. They have been successful in the field of remote sensing and GIS applications, especially when dealing with low resolution and narrow field of view satellite imagery such as Landsat and SPOT data (Yang, 1997). General functions have the advantage of simplicity. They can provide a reasonable geometric modeling alternative when little is known about the geometric nature of the image data.
However, conventional techniques generally process the images one at a time. They cannot provide an integrated solution for multiple images or photographs simultaneously and efficiently. It is very difficult, if not impossible, for conventional techniques to achieve a reasonable accuracy without a great number of GCPs when dealing with large-scale imagery, images having severe systematic and/or nonsystematic errors, and images covering rough terrain. Misalignment is more likely to occur when mosaicking separately rectified images. This misalignment could result in inaccurate geographic information being collected from the rectified images. Furthermore, it is impossible for a conventional technique to create a three-dimensional stereo model or to extract the elevation information from two overlapping images. There is no way for conventional techniques to accurately derive geometric information about the sensor that captured the imagery.
Photogrammetric techniques overcome all the problems mentioned above by using least squares bundle block adjustment. This solution is integrated and accurate.
IMAGINE Photogrammetry Project Manager can process hundreds of images or photographs with very few GCPs, while at the same time eliminating the misalignment problem associated with creating image mosaics. In short, less time, less money, less manual effort, but more geographic fidelity can be realized using the photogrammetric solution.
Single Frame Orthorectification/Block Triangulation
Single frame orthorectification techniques orthorectify one image at a time using a technique known as space resection. In this respect, a minimum of three GCPs is required for each image. For example, in order to orthorectify 50 aerial photographs, a minimum of 150 GCPs is required. This includes manually identifying and measuring each GCP for each image individually. Once the GCPs are measured, space resection techniques compute the camera/sensor position and orientation as it existed at the time of data capture. This information, along with a DEM, is used to account for the negative impacts associated with geometric errors. Additional variables associated with systematic error are not considered.
Single frame orthorectification techniques do not utilize the internal relationship between adjacent images in a block to minimize and distribute the errors commonly associated with GCPs, image measurements, DEMs, and camera/sensor information. Therefore, during the mosaic procedure, misalignment between adjacent images is common since error has not been minimized and distributed throughout the block.
Aerial or block triangulation is the process of establishing a mathematical relationship between the images contained in a project, the camera or sensor model, and the ground. The information resulting from aerial triangulation is required as input for the orthorectification, DEM, and stereopair creation processes. The term aerial triangulation is commonly used when processing aerial photography and imagery. The term block triangulation, or simply triangulation, is used when processing satellite imagery. The techniques differ slightly as a function of the type of imagery being processed.
Classic aerial triangulation using optical-mechanical analog and analytical stereo plotters is primarily used for the collection of GCPs using a technique known as control point extension. Since the cost of collecting GCPs is very large, photogrammetric techniques are accepted as the ideal approach for collecting GCPs over large areas using photography rather than conventional ground surveying techniques. Control point extension involves the manual photo measurement of ground points appearing on overlapping images. These ground points are commonly referred to as tie points. Once the points are measured, the ground coordinates associated with the tie points can be determined using photogrammetric techniques employed by analog or analytical stereo plotters. These points are then referred to as ground control points (GCPs).
With the advent of digital photogrammetry, classic aerial triangulation has been extended to provide greater functionality. IMAGINE Photogrammetry Project Manager uses a mathematical technique known as bundle block adjustment for aerial triangulation. Bundle block adjustment provides three primary functions:
- To determine the position and orientation for each image in a project as they existed at the time of photographic or image exposure. The resulting parameters are referred to as exterior orientation parameters. In order to estimate the exterior orientation parameters, a minimum of three GCPs is required for the entire block, regardless of how many images are contained within the project.
- To determine the ground coordinates of any tie points manually or automatically measured on the overlap areas of multiple images. The highly precise ground point determination of tie points is useful for generating control points from imagery in lieu of ground surveying techniques. Additionally, if a large number of ground points is generated, then a DEM can be interpolated using Create Surface tool in ERDAS IMAGINE.
- To minimize and distribute the errors associated with the imagery, image measurements, GCPs, and so forth. The bundle block adjustment processes information from an entire block of imagery in one simultaneous solution (that is, a bundle) using statistical techniques (that is, adjustment component) to automatically identify, distribute, and remove error.
Because the images are processed in one step, the misalignment issues associated with creating mosaics are resolved.
See Ground Control Points in Rectification for more information.