Prepared by Tom Mastin PLS

Photogrammtery for Surveyors

Since the domestic introduction of large scale map compilation by photogrammetric methods, the surveyor’s role has continually changed in parallel with the rapid improvement of photogrammetric instrumentation and techniques.

As recently as five to ten years ago, the economic break‑even size of a photogrammetric mapping project was somewhere between thirty and one hundred acres.  Today, a project comprising as few as five to six acres becomes an economic consideration with photogrammetric methods.  It is now generally accepted that photogrammetric mapping from aerial photographs is the best mapping procedure yet developed for both large and small scale projects.  It is faster and less expensive than any other method and provides more complete and accurate detail, supported by evidence which is historically retained in the wealth of detail of the aerial photograph.

As versatile and diversified as photogrammetry has become, so has the surveyor’s role in supplying the quality foundation on which good mapping is based.  It is impossible to make a map from aerial photographs without the underlying data provided by field surveys.

Field surveys are required to:

1)         provide the basic horizontal and vertical control needed to determine the scale, azimuths, and basis of datums for the photogrammetric process,

2)         provide mapping of certain desert or plain areas, sandy beaches, or snow where photographs do not show the ground surface well,

3)         provide mapping of deep canyons or high obstructions that conceal the ground surface in the photographs,

4)         provide mapping of areas covered with dense conifer or tropical rain forests,

5)         provide as‑built and sub‑surface structure detail,

6)         provide mapping of boundary and land net features,

7)         secure information which cannot be obtained solely from the observation of aerial photographs, and

8)         obtain precise supplemental map data such as cross‑sections, field edit and map accuracy verification.

Questions the surveyor is most often confronted with in a photogrammetric project are:

• How to design the most expedient mapping plan for a given project?
• How many photographs (stereo models) will be required to cover a given project area?
• How many horizontal and vertical control points will be required?
• Will the use and accuracy of aerotriangulation satisfy the extension of photo control?
• What must the distribution of control points be?
• What size targets will be required?
• What map accuracy can be expected?

Photogrammetry Definition

Photogrammetry is defined as the art, science, and technology of obtaining reliable information and measurements from aerial photography.

Cameras and Film

By design, the successful execution of any photogrammetric project depends upon good quality photography.  Aerial photographs taken with a frame camera are commonly classified as either vertical or oblique.  Vertical photographs having a 9‑inch by 9‑inch format are the most common type taken for photogrammetric work.

Basically, there are three types of aerial cameras presently used for photogrammetric mapping.  They are:

1)       the 8‑1/4″ focal length, narrow angle lens, designed for mapping, forestry, extreme relief, and other areas with similar high obstructions

2)       the 6″ focal length, wide angle lens, designed for universal mapping

3)       the 3‑l/2″ focal length, super‑wide angle lens, designed to accomplish high altitude, reconnaissance‑type mapping using a single engine aircraft

Of these three available camera types, the 6‑inch focal length lens provides the best compromise between stereo‑photo geometrical strength, scale and ground coverage.  This is the focal length that is in most common use throughout the world in photogrammetry today.  The comparative measure of stereoscopic geometrical strength in stereo photography is generally expressed in terms of base to height ratio (B/H).  Geometrical strength increases with increased B/H ratios; for the larger these ratios become, the greater the angles of intersection of corresponding light rays; thus increasing the stereoplotter pointing accuracy.  (Figure 1)  For this reason, the 3‑1/2″ focal length lens provides better photography from a geometrical strength point of view, and in certain instances is used to increase the vertical accuracy in large scale mapping.

Flight Planning

In order for the photography to satisfactorily serve its intended purposes, the photographic flight mission must be carefully planned and faithfully executed according to the flight plan.  A flight plan generally consists of two items:  1) a flight map which shows where the photos are to be taken, and 2) specifications which give the details on how to take the photos including requirements such as camera and film requirements, scale, flying height and a flight schedule which will coincide with completion of control pre‑marking.  In order to maintain the metric quality required in the photogrammetric process, aerial photography must be taken and processed in accordance with the U.S. Geological Survey publication, “Standard Specifications for Aerial Photography for Photogrammetric Mapping,” revised 1972.

The scale of a vertical photograph is the ratio of photographic distance to the distance it represents on the ground.  Inasmuch as a single frame perspective photograph represents a plane system, it follows that the scale of a vertical photograph will change throughout the photograph with variations in the elevation of the ground.  As flying height above the terrain increases, scale decreases; as ground elevation increases, scale increases.  These are important points to remember when considering photographic scale.

In photogrammetric work it is convenient to use an average photo scale which applies to the average terrain elevation in the project area.  The required flying height above average terrain (AMT) and above mean sea level (ASL) can be readily calculated once the required photo scale has been selected and the camera focal length is known.

Height above mean terrain (in feet) = Camera focal length in inches x scale of Photography ft/in = feet AMT

Height above sea level (in feet) = Camera focal length in inches x scale of Photography ft/in + avg. terrain = feet ASL

Vertical aerial photographic coverage of an area is normally taken as a series of overlapping flight strips.

To assure stereoscopic coverage of an area, a forward overlap (FOL) of 55 to 60 percent, and a sidelap of 30 percent is required.

The neat‑model for any overlapping stereopair is the area between adjacent principal points and extending in the y direction to the middle of the sidelap area.  For 60 percent forward overlap and 30 percent sidelap, the photographic area of the neat‑model is 3.6 inches by 6.3 inches.  (Figure 2)  In photogrammetric mapping, the actual area that is compiled, per stereopair, is generally limited to the neat‑model area.  Therefore, to roughly estimate the number of stereomodels that must be compiled in order to map a given area, the area to be covered must be divided by the area of each neat‑model.  Once the photo scale is selected, the area covered by a 9‑inch square single vertical aerial photo may be readily calculated.

Figure 3 gives the flying heights required to achieve various photo scales with the universal 6‑inch focal length lens camera.  Also tabulated is the corresponding area of coverage for each respective neat‑model.

Stereo Plotters and C Factor

Although the design characteristics of the various stereoplotting instruments may vary significantly, they all operate on the principal of stereovision.  Stereovision occurs when two photographs are made of the same object from different positions in space, and then instrumented so that the right‑hand photograph is seen by the right eye, and the left‑hand photograph is seen by the left eye.  The perspective intersection of light rays of corresponding images in the photographic pair is expressed as a three dimensional model and allows the observer the ability to view depth perception.  The stereoplotting instrument in turn, makes it possible for an observer to place the photographs in their proper geometric relationship with respect to their true position on the ground and to locate and plot planimetric features and contour lines by viewing the photographs in three dimension.

More technically defined, model orientation of two stereo photographs in the plotting instrument is the re‑construction of intersecting light rays of identical images from two separate photo stations in space.  The intersection of light rays from corresponding image points is sometimes referred to as the picture plane of the stereo‑model.  For stereoscopic measurement of corresponding images in the stereoplotter, the intersection of light rays in the instrument’s optical‑train is referenced by a black or white dot known as the floating mark.

This floating mark is also seen stereoscopically in respect to the reconstruction of image rays, and is viewed in direct relationship to the photographic image scale.  The size of the floating mark will vary from .025 mm to over .080 mm, depending on the type and accuracy of the respective stereoplotting instrument.  The nomenclature of the stereoplotter and floating mark accuracy is important to the surveyor, since the size of aerial targets or photo identities must correspond closely in image size.  In other words, the floating mark can “swim” within a target too large, or totally obliterate a target too small.

It is generally found that the greater the precision (C‑Factor) in the plotting instrument, the higher the allowable flying height.  (Figure 3)  This relationship is of great value due to the fact that increasing the flying height in turn increases the ground coverage per photograph, and therefore reduces the necessary ground control.  Due to the fact that vertical accuracy is usually the limiting factor in the photogrammetric process, the flying height is often derived by the contour interval of the finished map.  This relationship is expressed as a precision factor, referred to as the “C”‑Factor of the photogrammetric equipment, and applies to 6‑inch focal length aerial photography only.

C‑Factors as given by the instrument manufacturers should be considered as applicable to the calibration specifications of specific equipment, and not to the inherent errors of the photogrammetric process.

Contour plotting accuracy of course, depends not only on the stereoplotting instrument, but also upon the characteristics of the aerial camera, the quality of the photography, method of photo control, operator pointing ability, and many other errors inherent to the photogrammetric process.  All of these factors combined give what may be referred to as an “effective C‑Factor.”  Accordingly, the instrument rated C‑Factor should be reduced approximately 20 to 25 percent for large scale precision plotting.  For example, an instrument with a C‑Factor of 1500 should probably be considered to have an effective C‑Factor more in the range of about 1200 for plotting large scale design mapping where grading plans or earth quantities are a consideration.  Applying this equation, an instrument having a C‑Factor of 1200, can theoretically plot a 1‑foot contour interval from photography taken at 1200 feet above terrain.  (Figure 3)

While photogrammetry is essentially a tool designed to eliminate a large part of the field work from engineering and mapping projects, it is not possible to accomplish a mapping project without a minimal amount of field survey control on the ground.  Quality ground control, precisely discernible in the aerial photograph, is an extremely important part of any photogrammetric operation.  The accuracy of the finished map can be no better than the ground control upon which it is based.  National map accuracy standards assume that no discernible error exists in the ground control network.  Many maps that have been carefully prepared in the office have failed to meet accuracy standards simply because of poor quality ground control.  Because of its importance, the ground control phase of any photogrammetric project should be carefully planned and executed.

Photogrammetric ground control consists of objects whose positions are known in a ground reference coordinate system and whose images can be positively identified in the photographs.  The ground positions of control points are established by field surveying.  Ground control provides the means for orienting stereomodels in a stereoplotter and for relating aerial photographs to the ground in other photogrammetric processes.  Photogrammetric ground control is generally classified as either horizontal (the position of the object is known with respect to a horizontal datum such as a state plane coordinate system), or vertical (the elevation of the point is known with respect to a vertical datum such as mean sea level).

The accuracy required in any photogrammetric control survey depends primarily upon the accuracy required in the photogrammetric map or computation that it controls.  This is not the only consideration, however, as accuracy may also depend upon whether the control will serve other purposes in addition to controlling the photogrammetric work.

For large scale mapping, national map standards of accuracy required that 90 percent of principal planimetric features be plotted to within 1/30th inch of their true position.  On a map plotted at a scale of 1 inch to 40 feet, this represents an allowable horizontal mapping error of 1.3 feet on the ground.  If national map standards of accuracy are to be met, horizontal photo control must be located to considerably better accuracy than the allowable horizontal mapping error.  In general, all horizontal photo control surveys should be of third order or higher.

National map standards of accuracy further requires that the elevations of 90 percent of all points tested for elevation be correct to within one half the contour interval.  A rule of thumb in topographic mapping states that elevations of vertical photo control points should be correct to within plus or minus one‑tenth of the contour interval to assure that national map standards of accuracy can be met.  According to this rule, a map being plotted with a contour interval of 1 foot requires photo control to be accurate within plus or minus 0.1 foot.

Both the prudent surveyor and photogrammetrist will utilize some amount of redundant control to prevent mistakes from going undetected.  Redundant control also yields better accuracy in the photogrammetric processes.  As a practical minimum, therefore, each stereomodel oriented in a plotter should have three horizontal control points and four vertical control points.  The horizontal points should be fairly widely spaced in the model, and the vertical control points should be near the corners of the model.  Frequently both horizontal and vertical positions are known for the four corner points.  This is always the case if the control has been densified using analytical bridging.  An idealized control configuration for stereoplotter orientation is a minimum of six control points.  One control point shall be located in each corner, a point near the principal picture point and a point near the conjugate picture point.  Accordingly, a substantial amount of photo control is needed for stereo mapping of a block of photos.

Photogrammetric control can be established in one of two ways:  1) all necessary photo control can be established by field surveying, or 2) a sparce network of field‑surveyed photo control can be established followed by aerial triangulation to densify the control network thereby providing the needed photo control.  If a mapping project consists of compiling only a few stereomodels, field methods will likely be most practical.  However, for blocks of photos the amount of photo control needed is so great that the cost of establishing all of the control by field methods would be very high.  In these cases, aerial triangulation methods afford the most economical solution to the photo control problem.

Regardless of the method used to establish a quality control network, it can be no better than the accuracy of its identification in the aerial photograph.  The importance of good photo identification cannot be overemphasized.  Unless the images of control points are correctly identified, the usefulness of the ground control survey is lost.

In general, images of acceptable photo control points must satisfy two basic requirements:  1) they must be sharp, well defined and positively identified on all photographs, and 2) they must be in favorable locations in the photographs.

Two methods of image identification are presently used:  1) photo‑identification requiring location and selection of control points on the aerial photograph, and 2) pre‑set artificial targets which will subsequently appear as imagery in the aerial photograph.

Photo control images are selected after careful study of the photos.  The study should include the use of a stereoscope to insure a clear stereoscopic view of all points selected.  This is important because many of the subsequent photogrammetric measurements will be made stereoscopically.  Once the photos are available, a preliminary selection of control images can be made in the office, prior to going into the field.  Final selection should be made in the field, however.  This enables positive identifications to be made and it also permits an assessment of the accessibility of the points; an important economic consideration in planning the survey.

Images for horizontal control have different requirements than images for vertical control, because their horizontal positions on the photographs must be very sharp and well defined.  Some objects whose images are commonly satisfactory for horizontal control are intersections of sidewalks, paint marks on roads, manhole covers, corners of buildings, fence corners, power poles, points on bridges, etc.

Images for vertical control need not be so sharp and well defined horizontally.  Points selected should, however, be well defined vertically.  Best vertical control points are small flat or slightly crowned areas.  Large open areas such as the tops of grassy hills or open fields should be avoided if possible because these areas present difficulties in stereoscopic depth perception.  Areas having limited image definition such as black asphalt, white concrete, snow, water, etc., should also be avoided.  Intersections of roads and sidewalks, small patches of grass, small bare spots, small paved areas, etc., make good vertical control points.  It is imperative that ground features selected for photo control be discrete photoimages visible on both photos of the stereomodel to assure positive photogrammetric recovery.

One cannot overemphasize the importance of exercising extreme care in marking photo images which correspond to objects located in the field.  Mistakes in point identification are common and costly.  A power pole for example, may be located in the field but an incorrect pole nearby may be marked on the photos.  Mistakes such as this can be avoided by identifying enough other details in the immediate vicinity of each point so that verification is certain.  A small sketch of the immediate area around each control point, compared with the same area as imaged on the photo is also helpful.  A pocket stereoscope taken into the field can be an invaluable aid in point identification.

Artificial targets provide the most ideal photographic images for subsequent measurement, and therefore are used for controlling the most precise photogrammetric work.

There are many different theories and resulting formulas published by the many survey and mapping agencies throughout the world for determining the optimum application of artificial target.  Basically, they all have the same interest; to design a target which will be readily discernible in the photographic image, and provide the most accurate pointing ability in the photogrammetric process.

Positions of targets may be determined by field survey methods or by photogrammetric bridging (aero‑triangulation).  This procedure is commonly called pre‑marking or paneling.  Besides their advantage of excellent image quality, their unique appearance makes misidentification of artificial targets unlikely.  However, there are some disadvantages of artificial targets such as:

1)       extra work and expense are incurred in placing the targets,

2)       the targets could be moved between the time of survey and the time of photography,

3)       targets may become confused with other projects being conducted in the same vicinity,

4)       targets must be maintained until completion of the aerial survey, and

5)       extra expense may be incurred for their removal in areas subject to environmental concern.

The main elements in target design are good color, contrast, and a symmetrical target size that yields a satisfactory image on the resulting photography.  Contrast is best obtained using white targets against a black background.  Experience shows that white usually is the best color for the target material.  If the target must be placed on a light background, as on sand or gravel, a black or dark background material should be used under the target.  This might be tar‑paper, lamp black, black cloth, or the like.  The legs of the targets help in identifying the targets on the photos, and they also help in determining the central point of the target for measurement purposes should the image of the center panel be unclear.

Target sizes must be designed on the basis of intended photo scale so that the target images are the desired size in the photogrammetric instrument.  An image size of about 0.050 mm square for the central panel is generally ideal.  This image size corresponds with the average floating mark size of most instruments.  Target sizes are readily calculated once photo scale and optimum target image size are selected.  (Table 1)

Paneling material with high reflectivity such as pie tins, shiny polyethylene, enamel base paints, etc., should be avoided.  At certain sun angles these materials will have a tendency to reflect false light up the image light ray, and appear to be slightly floating in space when observed in the photogrammetric instrument.

Artificial targets should be set exactly over the station monument where horizontal and vertical stations have been recovered or are to be determined.  Actual target elevations must be measured and furnished in addition to monument elevations where monument elevations are not set flush with the ground.  An attempt should be made to set each monument and its respective photographic target where perspective layover or shadow of height objects will not obscure the target images on any photograph.

As indicated earlier, an extensive amount of photogrammetric ground control is necessary for orienting stereo models in photogrammetric instruments, especially for large mapping projects.  If the needed control were established exclusively by field surveying methods, its cost would be very high or possibly prohibitive to certain projects.

The technique of analytical aerial triangulation of blocks of photographs is defined as a photogrammetric method of extracting any number of x, y, z photo control points from only a skeleton network of field surveyed control.  This process provides a cost effective and significant time savings to the field operations and is conducted within the map compilation facilities, free of inclement weather, hostile or absent land owners, limits of access and other costly and time consuming conditions.

Aerial triangulation (or bridging as it si frequently called) yields satisfactory accuracy for the horizontal and vertical map control, and provides uniform continuity to the control network by the systematic ability to place unlimited numbers of control points in ideal positions of the stereomodel.

In analytical bridging, the amount of field‑surveyed photo control needed will vary depending upon the procedures, instruments, and personnel to be used.  In general, the more dense the ground surveyed network of photo control, the better the resulting accuracy in the supplemental control determined by bridging.  There is an optimum amount of ground surveyed control, however, which affords maximum economic benefit from bridging and at the same time yields a satisfactory standard of accuracy.

Analytical aerial triangulation has a number of advantages over aerial triangulation previously accomplished by the use of universal optical‑mechanical bridging instruments.

Specifically, analytical aerial triangulation accuracy is significantly increased because systematic errors can be accounted for by computational methods.  The corrections that are made are:

1)       compensation for systematic errors of the comparator measurement system,

2)       compensation for film shrinkage or deformation,

3)       correction for radial lens distortions,

4)       corrections for atmospheric distortions, and

5)       corrections for earth curvature.

Supplemental photo control (pass points) are the points whose coordinates are determined through aerial triangulation.  As such, they are the points of extended control that are used in subsequent photogrammetric operations.  The pass points may be natural images but more often are artificial points selected by the photogrammetrist.  Artificial points are preferred because they can be placed in desirable locations in the photographs, they can be quickly made, more accurate readings can be made on them, and misidentification problems are circumvented.

Marking of pass points in the photographic image is done with a precise stereoscopic instrument called a point transfer device.  This device allows overlapping photos to be viewed stereoscopically by means of a binocular viewing system.  A reference floating mark is superimposed within the viewing optics and can be stereoscopically placed coincident with any given point within the photographic plane.  To transfer a point from one diapositive to another, the floating mark is stereoscopically brought into coincidence with the desired point and the point is then precisely and permanently marked into the image emulsion by calibrated 60 to 100 micron diamond drills.

If a block of photos is being aerial triangulated, adjacent sidelapping strips are joined by means of similar pass points which occur in the sidelap area of the flight strips.  Pass points selected in the sidelap area and used to join strips are called tie points.

All control points and pass points identified and drilled in the photographic image are subsequently measured by comparator instrument methods, resulting in x, y plate accuracies of ± 3 microns.

The accuracy with which pass point coordinates and elevations can be determined in the analytical process depends upon several variables such as:

1)       the quality of the photography,

2)       the exactness of the camera calibration,

3)       the extent of film shrinkage or deformation,

4)       the quality and density of the ground control,

5)       whether or not artificial targets were used,

6)       the method of aerial triangulation and type and condition of equipment used,

7)       the ability and experience of the personnel, and

8)       photo scale.

Photo scale is perhaps the most important variable.  Because photo scale is linearly related to flying height above ground, aerial triangulation accuracies are commonly expressed as ratios of the flying height above ground.  For carefully executed fully analytical aerial triangulation, performed with vertical photographs taken on film negatives with 60 percent endlap and 30 percent sidelap, and ground control sufficient in density and quality to satisfy the normal adjustment equations, the accuracy of pass points can be expected to equal about 1/10,000th of the flying height.

Verification of map accuracy and its completeness is normally conducted by the surveyor in the field.  Ground features which were not discernible in the aerial photographs due to dense vegetation, thick underbrush, overhanging objects, etc., must be identified and measured.  This phase of work is often referred to as field edit or field completion.

Specific information pertaining to as‑built features such as inverts, bridge measurements, sub‑surface utility risers, material classification, as‑built measurements, etc., is also measured or described by field methods.

The field edit data must subsequently be computationally or geographically compiled in such a manner that it may be accurately incorporated into the map.

Almost every engineering type map is compiled in accordance with project guidelines or specifications.  Although there are as many different specifications as there are users, the most generally accepted specifications used are U.S. National Map Accuracy Standards published by the Bureau of the Budget, revised June 1947.

A mapping project meeting these standards is a job well done.

What is Photogrammetry

Over the last few years the concept of photogrammetry has changed dramatically within the general public and with many non technical professions, including the GIS Profession. The primary reason is because of inexpensive digital cameras and drones (or UAV’s or UAS’s). It is true that new technology has changed the way that we can look at doing 3D modeling of areas or objects. However for large projects with precise requirements, conventional Photogrammetry is still needed.

Film Format

Commercial photogrammetry uses 9″ x 9″ film and usually with any project you will receive a contact print of that film. A contact print is just as it sounds in that the paper is in contact with the film at the time of exposure. Therefore the contact print will be 9″ x 9″ for the area showing the imagery.

Currently the most common film type used is color. Back in the 19070’s and 80’s it wasn’t uncommon to see black and white film. It was used, not because it was cheaper, but because the quality of the color film wasn’t as high and so information for mapping could be lost. These days that film is very high quality. For standard photogrammetry, color infrared (CIR) is not used.

Cameras

The Cameras are high end metric cameras. A metric camera is one that has been calibrated, which allows for more precise measurements. There are metric digital cameras. However because of the cost of the cameras and the changes in processing between digital and film, most of the photogrammetric work is still done with film cameras.

 Parts of a photo

• 1 and 8 are Fiducial Marks. The 8 fiducial marks are precise marks transferred to the film, that when the opposite fiducial marks are intersected the “Principal Point” (#9) or the geometric center of the film can be determined. Note the lines shown on the image to the left do not show up on actual film images.
• 2. Date the photograph was taken
• 3: Strip and Photo number. This allows you to determine the order of the photos.
• 4. Level Bubble
• 5. Clock
• 6. Altimeter
• 7. Some sort of Project indicator
• 9 Principal Point
• 10. Lens Number
• 11. Exposure counter
• 12. Calibrated focal length in mm.