IndexBasic Photogrammetry Session IntroductionBasic Photogrammetry Session Technical Session Introduction Technical Section - Vectors Files Technical Section - Terrain Files Technical Section - Raster Files Technical Section - Raster Processing Technical Section - Process and Product Examples Technical Section - Planning |
This is most of the text from the notes of a seminar called MAN Home |
| The basics of the aerial camera haven’t changed in decades, in general we are looking for a very large format (9 x 9” typical) 6” focal length camera which can expose photography in a series at varying amounts of overlap (60-80% typical), the overlap allows us to look at the same area from two different views and see stereo or 3-D. While every aerial camera has the same distortions that any other camera does they are measurable and can be corrected which is what the term Metric implies. What has changed is the quality of the lenses and some bells and whistles. Explain; FMC, resolving power (AWAR - Area Weighted Average Resolution), auto leveling, on board GPS for navigation and control with its affect on control requirements and flight planning. We are now starting to see the advent of cameras that expose the image digitally . The concept is similar to the digital cameras found everywhere but the requirements of photogrammetry caused problems that mean most ditigal aerial cameras a just beginning to come to market. They will have unique advantages and disadvantages that will become more evident as time goes on. Most recent model cameras are well within reasonable expectations, but it is worth knowing what is reasonable and what camera is being used on a project, especially if it will result in digital imagery. |
| Allows for 3-D binocular viewing of photo area at enlarged scale and measuring in a ground coordinate system as provided by the surveyor. Purpose is to remove relief displacement (explain for benefit of mapping and orthos). This has also come to include digital photogrammetric workstations, the concept is the same except the viewing is done on a computer screen in 3-D using high resolution scans of the aerial negative. While the measuring and mapping parts are very similar the advantage to a digital workstation is that the mapping can be superimposed right on the 3-D photographic view allowing you to see both the ground being measured and the map as it was measured at the same time. This also aids in updating if you are looking at an old map and new photography, differences are very apparent. They have both their place and their drawbacks and we have seen proposals that specifically say they can not be used. . |
|
| Typical CAD systems though many mapping firms use mapping specific systems then translate to common formats. Most are using engineering type packages to generate contours from terrain models, these can also be used for profiles, volumes, and sections although that is primarily done by the client now. We are also using two levels of image manipulation software either for simple referencing and plotting in a CAD environment, or for specialized ortho processing both of which will be explained later. |
| 60-80% Stereo lines at a scale to match project design specs. Usually leaf off and with no snow. It is important to remember that the photo requirements for a digital imagery job are going to be different than those for a straight mapping job. Mapping has a far less stringent set of requirements so if there is even a possibility that imagery will be needed it needs to be mentioned as early in the process as possible. That way if possible it can be flown when light and ground conditions will maximize the visual appearance and detail of the photography. It comes down to this with mapping photography you want the operator to be able to make measurements that meet the specs and you don't really care how he does it, but with the imagery you want not only an accurate product but also a pretty picture. |
| In the coordinate system you plan on using for the project, or tied to any past or potentially useful existing system. |
| It is important to realize that every phase takes time. Times are usually defined as From date of Photography, or From date of control. The clock can't start until all materials are gathered in one place. Then there is still the collection time whether it is 50 acres an hour for 200 with 5's or 5 acres an hour for 50 with 1's . |
| Design and plan have the ultimate importance and are detailed below. Special care needs to be taken to decide on the proper level of accuracy. Very accurate, totally comprehensive, completely flexible, up to date information is always the best, but each of those adjectives cost a certain amount. There may be a time and place for sloppy, old, rigidly, limited data and indeed there are times when we find stuff like that invaluable for it's use. There is an infinite number of combinations that can be manipulated and balanced to form a product that meets the majority of needs within the budgeted amount but this will all be explained in much greater detail later including how and when to mix and match the various factors and their influence on cost. |
| GPS, conventional, aerotriangulation. Control placement is important. Control plans are laid out to give the maximum accuracy with the minimum amount of work in the field. Target placements are laid out on the quads in a precise fashion based on the flight lines, depending on the photo scale a shift of 200 feet in any direction could be important and require someone to go back out and replace the point with something that is almost certainly inferior. Of course points can be moved but should always be done with prior consultation and plenty of documentation. |
|
Point by point collection in 3-D. Level of detail and accuracy depends on
project design.
|
|
Final clean up with particular attention paid to validation of terrain surface,
this is especially the case if the job is going to be used for orthophotography.
|
|
Hard copy at full or reduced scales, disks in a variety of formats.
|
|
Hard copy from ink-jet, film writers, or photo plotters. Files in variety of
formats and probably in multiple resolutions.
|
|
Yes it is titled Photogrammetry, GIS, and the Surveyor but we are not GIS
experts or even GIS consultants in the strictest sense of the term. I am not
going to try to give the last word in designing creating and implementing a GIS
project. What we do on a regular basis is provide large quantities of data
balancing accuracy needs and cost constraints to give the most usefulness for a
reasonable dollar amount. It may better be called A photogrammetrist’s look at
the surveyor and emerging sources of data and new technologies . We will also
discuss some of the questions we get on a weekly basis regarding hardware,
software, and techniques. Finally we will try to better explain or at least
define some of the terms you will see in GIS proposals.
Is this a valid use of your time? yes if it gets you to start thinking more about the way things are changing around you and how you can take part either as a data or service provider or as an intelligent user of existing data. There are vast stores of data out there which represents work that has already been done, all that needs be done is an evaluation of the expected accuracies and proper uses. There is also a need to collect new data to either expand the area or usefulness of what is already there, as well as verification of what has been or is being collected, and finally the need to and keep it all up to date. We also have the tendency to hear the term GIS and think of the vast, complex, city or county wide projects that are in the news and on the web. These are fantastic tools and possibly opportunities, but today I would like to expand our thinking to the purest components, those being that in our offices we can use any System of Information that is stored in a Geographic manner to help us every day. It could be done with tools, and information that is available already, free, or very reasonably priced when weighed against the benefits. Some good example of why spending some time on education is valuable come from a four month survey of the magazine Professional Surveyor . Articles from the period July/August, September, and October 1999 show the following uses of, or need for Geographic Information knowledge by our professions. |
| These are the files everyone is familiar with creating using and exchanging. They could be Autocad DWG., Intergraph DGN., or DXF from almost any source. The U.S.G.S. also provides quad data in a couple of different formats. In Ohio the most important one may be the DLG format because every quad in the state of Ohio is available on the OGRIP web site. DLG is simply an ASCII format that is a collection of XY coordinates with attributes. There is at least one very good free translator from DLG to DXF which can then be read into most CAD programs. The main purpose here though is to draw out the differences between vector in whatever format and raster type files. |
| Huge amounts of data can be collected in very timely fashion with very good accuracy, and at a density that is impractical in the field however this data will always need to be field verified to some extent. |
| Data collected by a surveyor on the ground either by conventional or GPS methods. This has the advantage of someone actually looking at the details of the object being identified so it is possible that minute attribute data could be picked up at the same time (e.g.: pole number, manhole type, number of outlets, hydrant address). This might also include TRANSMAP type data where large amounts of data are collected in a dynamic drive by method. Field procedures would also include verifying or identifying things that have been place by photogrammetric methods or heads up digitizing, and could go as far as tagging items with attributes in the field. |
| This could be an existing county wide GIS, or digital quads from U.S.G.S., or any old existing data. You may have done a large job for a good size plant or small community and now might be the time to go back and review big jobs to see if something like digital imagery would enhance the usefulness or possibly aid as a temporary update in changed areas, or possibly what has already been done could serve as a base for adding information or intelligence to what is already there. Especially since much of the most time consuming part of the work may be done. |
|
Layering is only one example of an attribute that can be manipulated in a
simple CAD program to provide the basics of an Information System. Think of
the number of variables in a simple CAD data file that can add intelligence to
the information being displayed in a visual manner, linetype; both names and
graphic appearance, Symbol; likewise names and graphics, along with sizes,
there will also be other software specific tags. Click Here for a table showing the layering scheme used as a standard by M.A.N. Mapping. |
| Every line, arc, and symbol has a coordinate attached as part of the entity. It can be snapped to, listed, and extracted. |
| Regardless what the specific package is, you probably have years of experience with some type of vector manipulation. |
| It is easy to go in to small selected areas and update them from new source data, these cut-outs can then be transmitted either independently or with the remaining original data as a whole. |
| Because of wide acceptance of certain standards (file formats), and the proliferation of the most popular software. Moving data between firms and systems has become much easier. This would include the increased understanding of CAD terms. |
| This might be helped to a small extent by planning. For example monumented permanent control, sheet or file design that can be expanded or updated with little problem. A planned approach for building the full data set over time. |
| What do you do if the data collection didn't include fences, field lines, or treelines. It could also be true that you need to see or count certain things like drives, walks, or paint stripes, but can't justify the cost of collecting the actual entities. |
| The definition itself is a problem because the three terms are used interchangeably. While there is no official designation, the definitions provided are the ones most often used in each instance. |
| Usually describes the combination of vector data as break lines and a random or regular array of grid points. This is typically the data collected either in the field by photogrammetric methods, and delivered either in a CAD or an ASCII file suited to the design system. The raw data. |
| This term most often describes the elevations in a raster format. Usually this is used internally by digital ortho systems because it gives a very even data set for the ortho to be corrected to. |
| Named this because it is the network of odd shaped triangles formed by connecting every point in the file to it's nearest neighbors. Triangles are formed and refined until no plane contains a point inside the triangle. Break lines are held absolutely as triangle edges, and because of this their position and configuration is crucial to the accuracy of the surface described. Most typical analysis we think of is based on T.I.N.’s, including volumes, contours, slope maps, and cross sections. |
|
Collecting of terrain data by photogrammetric methods is very efficient, for
example a 100 acre site with moderate terrain and a 20’ grid with breaklines
could be collected in 6-10 hours. Note that the point density for the mapper
will be much higher than for the surveyor but the point by point accuracy will
be lower, this is due not only to the inherent inaccuracies of photogrammetry
but also site specific factors such as ground cover and surface appearance.
A photogrammetric firm can be a tremendous source of information on the collection of terrain data because of the huge amount of data gathered every day and the varying situations and circumstances met. The mapper must form a ground surface that will create accurate representative contours whether it is corn field or a multilevel interstate interchange with vertical walls and stacked bridges. |
| For mapping this can be an important supplement to the photogrammetry where visibility is poor. It is also a crucial check on the actual accuracy of the mapping data. While it probably wouldn't work to collect the terrain for a full ortho project, There could be instances where you have existing field data that could well be used for some sort of photographic rectification in limited areas. |
| Here again an existing GIS, digital quads, U.S.G.S. one degree D.E.M.s, or any previously compiled mapping might have terrain data that could be used at varying degrees of accuracy for ortho generation. |
| This is the process where the computer compares similar pixels in adjacent photographs to determine elevation. In instances where photography is exposed very high and distractions like trees and buildings are very small variations relative to the flying height it can work every where with very little problem. For normal large scale ortho photos care must be taken in and near these anomalies with some photogrammetric checking and fill in where necessary. It can work for contours if used under ideal circumstances and when supplemented with breaklines. |
| The actual accuracy of the terrain data can vary based on the intended use. Speed and affordability are balanced with cost, accuracy, and usefulness. Note the affordability on one side and cost on the other, they are two very different things and are usually both pretty well defined. The cost doesn't matter no matter how inexpensive it is per unit if the project is not affordable. Usefulness is defined here by the amount of flexibility and potential uses for the data (e.g.: a raster dem is useless for anything but creating an ortho but might be free and currently available). |
| Here the requirements are much lower, only enough data to properly position the pixels is needed. For example in many types of terrain a much looser grid with only the most significant breaklines such as rivers and large streams may be necessary. |
| Here it is assumed that the contours displayed will be used to show the general shape of the ground and are not for final design work or quantities. A tighter grid will be used, probably no more than 30 feet and many more breaklines will be collected, such as ditches, swales, and banks depending on final scale and contour interval. |
| This terrain file is the same as any that would be provided with topographic mapping. It would be as accurate as you could expect for the scale and contour interval. Here the grid drops to around 20 feet and any linear features that could affect the shape of the contours are collected. Here one might expect breaklines that approximate the backs of curbs, centerlines of roads, or backs of walks where they determine a break in the terrain. |
|
We need some understanding of what we are talking about when we start speaking
of digital imagery , a lot of the questions we get could be answered fairly
easily with a little basic background into just what exactly it is.
To begin with digital imagery is delivered (usually on CD) as a raster file. By definition a raster file is simply a file organized in rows and columns of values. For a digital image each row/column address has either a greyscale or color value. The term pixel depth comes in here, you may have heard the terms 8 bit, 24 bit, pallated, and true color. These terms just refer to what information is stored at each row/column or pixel address. In an 8 bit greyscale image, each pixel contains 8 bits or 1 byte of information which gives it a range from 0 - 255 so the appearance of the pixel ranges from 0 or black to 255 or white with 253 steps of gray in between. If the file is an 8 bit, 256 color image there is a pallet stored with the pixel addresses which describes what color each number between 0 and 255 are displayed as.
Finally if an image is 24 bit or true color then each of the pixels has 24 bits or 3 bytes of information to store the color value. This is accomplished by using 8 bits or the 0-255 range for red, green, and blue, which is obviously where the term RGB image comes from.
The concept of the red, green, blue number triplets isn't really that complex and it applies to everything from mixing colors to make the pen values in your ink-jet plotter, to designing web pages but is a little more than we want to get into here. The moral of the story is that the viewing system simply reads the values in the file and assigns them to the correct screen pixel to reproduce the image. There are hundreds of ways to organize, compress, and attach other information such as headers to these files. The most commonly used file types are TIFF, TARGA, JPEG, and RAW though many viewing programs also have a native format. These vary the way vector files differ in organization and composition. In CAD or GIS applications they are usually expressed in pixel sizes for example, in a one foot pixel digital images each dot of information (row and column address or pixel) occupies one foot in the geographic coordinate system. A service provider also needs to be concerned with scan rates. This is the resolution at which the original photography was scanned, it should be finer than the final output pixel size. See Handout. |
| The key here are the quality of the image captured by the aerial camera. High quality scanners are capable of up to 3500 dpi but if the photography being scanned is of poor quality or flown too high then you only get a very high resolution image of a fuzzy photo. |
| Actual USGS quads scanned at a moderate resolution and georeferenced. |
| As the name says, one fourth of a U.S.G.S. quadrangle as a digital image at a 1 Meter resolution. Generally out of date unless you happen to catch your area in the midst of the update cycle. |
| Resolutions at this point are measured in multiple meters, with price and availability being prime considerations. However all of these factors are getting better at a reasonable pace and should be considered for very large, low resolution, preliminary planning projects. |
| This could be useful for inserting some significant existing detail right into the plans or as additional field in the database. |
| Especially spatial relationships because you see things you recognize in the context of everything that is in the area. Plus you have the wow factor in which even a marginal photo looks great if someone sees their own house or anything they are familiar with in it. |
| Whether it means learning a totally new GIS or image viewing system, or adding capabilities or function to an existing software (Colorview example), or simply further exploring the capabilities of an existing system. |
| May not be able to just read into a viewing or editing program and start digitizing, in the first place your viewer may not have the capability to digitize, measure, or draw, and if it does it may not be able to read the geographic coordinate system attached to the raster file. |
|
As the project size increases the problems add up as well. For example:
Add a second image and now you have to logically name them both. [expand on that thought] Add a large enough area and new you need to consider multiple resolution data sets and overviews. Add enough to require a second CD and there must be some kind of logical division and cataloguing. |
| Because of the structure of raster files it is difficult to update small portions of a large image with new data. Matching existing colors and tones can add to the problems. |
This is simply the process of adding a coordinate system to the digital image.
As noted above the photography has no coordinate system other than the rows and
columns. It is the software that establishes the real world coordinates based
on the information in a header or supplemental file containing the scale,
rotation, and insertion point necessary to match any given pixel to it's ground
position. The accuracy of geo-coding depends on how the photograph is tied to
the ground. Geo-coding is automatically a part of each of the processes since
each of them tie the image to the base data. Geo-coding can be accomplished in
one of several ways.
|
| Very similar to what would be done in a photo lab when photo plan profile sheets or photo mylars. This would be a simple two point measurement to establish the approximate relationship between the image and the coordinate system. The two measured points would be very close, but everything else would just be place based on their relationship. Errors would be significant in the case of photography with a good deal of tip, tilt, or relief . The primary goal here could be to just get an image into a digital form quickly for either display or very rough measurements. |
| Rectification would be the process of making known visible points in the image match the corresponding points on a vector base map. The number of points depends on how many are available and how long someone wants to spend matching points. once known points are matched everything between them is prorated accordingly. The position of all measured points will be very close. Data between measured points can also be quite reasonable if the displacement corrected is fairly uniform. Significant errors should only occur where the data is sparse or there is much relief between the measured points. Think of it as a multi- point scaling which can be especially useful in matching multiple photos. |
| In this process every single pixel is processed according to it's location in the photograph, the exterior orientation of the camera platform (tip, tilt, rotation), and the elevation of the ground at the point represented by that pixel. This is a measurable product with the position of everything in the photograph being nearly as accurate as a map at the same scale. Since orthorectification takes place at ground level the only differences would occur for things that are not on the ground like roof lines. |
| A simple two point scale, or rectification of a few points along the way may be sufficient on a simple project. In many cases the scale and position over the entire route is important especially on a large project area, though the actual accuracy needs are many and varied. It might also be worthwhile to add in the D.T.M. if that is a factor. |
| Here is the ability to record the state of large areas with tremendous amounts of detail with a limited amount of terrain data. A better D.T.M. could also be added to show contours good enough for drainage studies or slope analysis. An easy add on to mapping that is being done for other purposes. [Reference article 3-D Subdivision Design in Oct 1999 Professional Surveyor]. |
| Can be used for digitizing new data, inventory of what exists, or simply for display purposes. Accuracy and type of processing dependent on use. |
| For preliminary "what if" or "go / no go" decisions there could be low cost, low accuracy solutions using existing data or limited control. |
| Imagine the potential market for large plant management. Everything from a simple CAD file with layering and image background to a full blown GIS with complex intelligent data scanned plans, and digital views of existing parts or situations. |
|
| Now that we have some idea of the types of data we are going to be either acquiring, generating, or using, it is time to start thinking about how it will all come together. The primary purpose of planning is to match the requirements of the final product with the procedures used which will determine the cost and usefulness of that product. (As stated in introduction) Special care needs to be taken to decide on the proper level of accuracy. We said that Very accurate, totally comprehensive, completely flexible, up to date information is always the best, and that each of those adjectives cost a certain amount. There is an infinite number of combinations that can be manipulated and balanced to form a product that meets as many needs as possible within the budgeted amount. |
Whether you are creating a simple cad file with raster quad background to
display jobs you have done in a given area, or a community wide full blown
graphic and attribute laden information database you need to consider who will
have access to it or desire to use it. Will it be used for design, display,
inventory or information management, or for planning. What software systems
are involved (among potential users) and what is the easiest, most generic way
to transfer data, the native format may not be the best way to receive raw
data. Think ahead to other systems that may be used. It is at this point that
consideration should be given to how the files will be structured:
|
| Are there other departments that could use the data if small changes were made, and possibly help defray some of the costs. Are there other functions that the data could fill if it was modified or enhanced slightly. |
| Very accurate | vs. | preliminary, quick, what if, display. |
| totally comprehensive. | vs. | starting point, add-on, limited scope. |
| completely flexible. | vs. | one time, limited use, no other needs. |
| up to date information. | vs. | available, rough base. |
| This would run the full range from high order field control and full large scale topo mapping --- through some checking and digital quads --- to no control or base data whatsoever for an uncontrolled mosaic. The important point is that every process has it’s own absolute practical limits to it’s accuracy which may or may not balance out with other benefits. |
| Since this will determine how accurate individual measurements on the final product will be it is important to consider whether the data going to be used for display purposes only, route checking, preliminary planning, or final design. These factors will influence the decision to digitize or convert existing data, map it as new or original data, collect in the field by conventional, GPS, or a roving system, or a combination of all using different methods depending on the importance of each type of data. |
| Usually discussed in terms of microns or dpi, most photogrammetric scanners have resolutions up to 2000 dpi true optical available with a geometric accuracy of 2-4 microns ( that is to say a scanned object will be within 2-4 microns of where it was on the physical film). |
|
Higher = Less photos, more coverage, less
relief
displacement, less expense per area unit.
Lower = More accuracy, greater detail, flexibility (can go to lower resolution). There is no hard and fast rule, every situation needs to be evaluated based on all factors listed here. Generally the magnification factor from film to final planned output is in the 5-8 times range. There are many reasons that exceeding this would be a valid option but for the detail most people expect to see and the accuracy levels generally stated this range fits most cases. |
|
Not only how high, but the physical location of the flight lines must be
considered along with whether or not mosaicing will be allowed (it can either
be a great advantage or a big problem; don’t get locked into a set idea).
As a rule: Closer flights = more neat area per photo = more flexibility = higher cost.. |
| File formats, naming convention, inventory system, this is a great place to put a little HTML learning into practice. |
| Determine needs of self and other interested parties. |
| Control, supplementary data. |
| Input into final project design. |
| Expectations, time constraints. |
