Orthomax is an ERDAS Image add-on program that allows one to perform quantitative photogrametry on stereo imagery. The software package was originally designed by a small company called Autometric; however, ERDAS has purchased the right to distribute the software as an add-in to Imagine. Consequently, Orthomax and Imagine do not always work seamlessly together but transfer between the two programs is usually problem free.
This document will serve to develop some of the important concepts in photogrametry, the basics of how the Orthomax software works, and step-by-step instructions explaining how to use stereo imagery to extract Digital Elevation Models (DEM) and then orthorectify the imagery. In addition, several flowcharts and checklists are provided to aide in performing the photogrametric analyses.
Concepts in Photogrametry
Several important concepts and terms will be important to understand the analyses described below:
* Air Photos: Air photos are taken with special long-focal length cameras. Air photos have an inherent scale to them which is related to the focal length of the camera and the height at which the photo is taken. While it is a common practice to change the scale of the air photo through photographic enlargement or reduction, it is important for the software that you use a photo that has not been enlarged or reduced for the photogrametric analyses.
* Stereo Pairs: Air photos which are taken at intervals such that there is a common area of ground that can be seen on two or more photos contains information about the elevation in the common areas. These areas are termed the overlap areas of the stereo pairs. Commonly, a 40%/60% overlap is common for air photo lines taken. If common points on each photo can be identified and the three dimensional position and orientation at which each photo was taken are known, the elevation of that point can be determined.
* Cameras: Each type of air photo camera takes photos differently than the other. Depending on the lens in the camera and the focal length, the camera will distort the image. This distortion generally increases away from the center of the photo. This distortion can be characterized by a polynomial function. In order to find the points common to each stereo photo which is necessary to calculate the elevation of the point, the lens distortion must be known. There are three parameters which are commonly used to define the lens distortion of the camera: the Ko, K1, and K2 parameters. These parameters, as well as the camera focal length, must be known to perform photogrametry using Orthomax.
* Fiducial Tics: There are markings on the air photos which orient the photo relative to the inside of the camera. These ticks are generally cross-hairs in the corners and on the sides of the air photos. These marks must be present on the scan of the air photos in order to allow Orthomax to correctly orient the photo relative to the principle point (point of no distortion in the photo).
Checklist of necessary photo attributes:
The air photo datasets adequate for photogrametry must poses the following attributes:
__ Focal length of the camera must be known.
__ Camera distortion parameters must be known.
__ Camera is contained within Orthomax database.
__ You can figure out which camera was used by trial and error, assuming
camera is contained within Orthomax database.
__ Scans must contain fiducial ticks.
__ Common Ground Control Points are known.
General Procedure for Preparing Imagery for Photogrametric Analysis:
1) Open up Orthomax: first, open ERDAS Imagine by either selecting it from the applications menu on Alai or by typing "imagine" at the UNIX prompt.
2) You will see an option for "Orthomax" at the right extreme of the Imagine toolbar after Imagine starts up. Click on this button.
3) Wait a couple of seconds. The Orthomax main toolbar should appear. At this point, you have encountered your first glitch in Orthomax. The "Loading Orthomax... Please Wait" window will remain on the screen even after Orthomax is fully loaded. Get rid of this annoying window by clicking "Cancel".
Create a Project:
Before performing photogrametric analyses, you must set up a project directory in which your images, DEMs, and orthophotos will be saved. To do this:
1) Select "New Project Directory..." from the Orthomax file menu.
2) Under "New Project Name:" type the complete path of the directory that you want your files to be stored in. This should not be a preexisting directory.
3) Click OK.
Also, if you wish to resume work on a project that already exists, use "Set Project Directory..." from the File menu and follow the same procedure.
Creating a Block: Importing images and specifying camera properties
Blocks contain the information such as the frames (photos) you wish to work on, the camera parameters that is associated with a set of photos, the positions and orientations of the camera associated with each frame, extracted DEMs, and extracted Orthophotos.
1) Click on the icon "Block Tool" in the Orthomax main toolbar.
2) Create a new block or open an existing block: Select either "New..." or "Open..." from the block File menu.
3) If you select new, you will have to enter information about the block that you wish to create. First, you must specify what type of imagery you are working with. The two "Sensor" options are "StdFrame" and "SPOT", use "SPOT" when dealing with SPOT satellite imagery, use "StdFrame" for everything else. Then, give the block a name under the "New Block Name." Specify the units of your reference frame under the "Linear Unit" pulldown menu. Finally, you must specify the reference frame you will reference the imagery to. This is a coordinate system and projection. Click on the "Set..." button to set the projection. A dialogue box displaying some of your options for coordinate systems will pop up. Choose the coordinate system to which your ground control is referenced by clicking on the "Map Projections" button. If your projection is not in the list, click "Cancel" and choose the "Projection Editor" under the "Utilities" menu of the main Orthomax toolbar. The Edit menu provides you with several tools to create projections. Once you have added the correct projection to the list of projections available, you may return to the "Reference Frame, Set..." button and select the proper coordinate system. Once this is completed, click "OK" at the bottom of the box. Click "OK" at the dialogue box that warns you that multisensor capabilities will be disabled.
4) Select the camera that you wish to use by opening the "Camera Editor" under the Edit menu in the Block Tool window. Click on the "Add Camera" button. A black line will be displayed. Now, click "Select Camera" and select the camera that was used to take the air photos and click "OK". Click "OK" in the Camera Editor window once the camera has been added.
5) Add the imagery that you wish to use to the block: In the Block Tool window under the Edit menu, select Frame Editor. Click the button that says "Add Frame." A blank line will be added to the list of frames. Now, add an image by importing it using one of the import options. Usually "IMAGINE Import..." and "TIFF Import..." work well. When done adding all of the images you wish to analyze for an area, click "OK." Click "OK" when it asks you if you would like to build the RRDS now.
6) Orient the photos relative to the fiducial tics: Under the Activities menu of the Block Tool, select "Interior Orientation..." It will ask you which of the frames you wish to orient in a dialogue box. Select a frame that you have not oriented and click "OK." A locator window will appear. You can adjust the image brightness and contrast with the "Brightness" and "Contrast" sliders. Depending on your camera, you will have to locate a number of fiducials on the image. Use the 1 button to place or remove a fiducial and the 3 button to zoom in on an area for a closer look. Once you place the fiducial corresponding to the fiducial number in the lower list of fiducial ticks, click "Accept." (++ is lower left) After you measure two fiducials, the computer will jump to the next fiducial automatically. After measuring several fiducials on the image, the RMSE at the bottom of the window will be updated. Generally, you should keep your RMSE below 1 pixel. Once you have completed measuring the fiducials for a frame, select "Apply" under the File menu in the window to save changes to the fiducial locations. Continue this process for the remainder of your photos, opening each new photo with the File-> Open command.
7) Measure Ground Control: Now you must tell Orthomax where several points on each image are actually located in real life. I recommend at least 4 ground control points per frame to perform an accurate triangulation. From this data and the tie points (see below), Orthomax will figure out the position from which the photo was taken and the orientation of the camera at the time the photo was taken. First, enter ground control points by selecting the "Ground Point Measurement..." option from the Activities menu from the Block Tool window. Next, a window asking for 3 frames appears. Select 3 frames that you wish to work on in each column (one frame per column). Then click "OK." This will throw up a dialogue window. First, select the "Auto-Create Point ID" option under the Options menu in the window. Next, click the "Add Tiepoint" button at the bottom of the screen. Now, locate the ground control point that you have X, Y, Z coordinates for in as many of the three images as possible. Then, zoom in on the point by placing the arrow above the point and using the third button (right) on the UNIX mouse. Use the 1 button to place the Ground Control Point (GCP) onto each image. Once the GCP has been identified on each frame, click the "Take Measurement" button. Then, click the "Ground Editor..." button to bring up the Ground Control Point editor. Select the point that you have just entered and change the "N" under the "Active" column to a "Y". Then, change the first letter of the ID from a "T" to a "C". Then, change the "T" under the "Type" column to "C." Finally, enter the 3D coordinates of the point in the "Longitude a priori", "Latitude a priori", and "Elevation a priori" columns. When finished, click "OK" and repeat all of #7 for each ground control point.
8) Measure Tie Points: Tie points are points that are common to each frame; however, ground control is not known. I suggest specifying about 45 ground control per set. To specify Tie Points, repeat procedure for Measure Ground Control except do not use the Ground Editor at any time. Simply take the measurement and do not enter any lat/long/elevation information. **Note: after entering several tie points, you can click on a tie point in one frame (the Reference Panel frame) and then click the "Auto Place" button. The autocorrelation algorithm usually finds the corresponding points on the other frames***
9) Triangulate the block: After you have entered all of the ground control points and tie points for all of the frames you are working on, click the "Triangulation..." button on the Ground Point Measurement panel. The "Triangulation" dialogue will appear. Click the "Run" button. After several moments, the computer will solve for the position and orientation of the camera for each frame. After the triangulation is complete, it is a good idea to view the results by pushing the "View Results..." button. I usually look to make sure that 1) the model converged (finished in less than 10 iterations) and 2) the residuals are low. Once you have checked this, you can print the results (they are usually lengthy) or click "Cancel." Next, click the "Accept" button in the Triangulation dialogue and then the "Update" button. Finally, once you return to the Block Tool window, make sure to save the block.
Extract a DEM:
1) Start the DEM tool by clicking on the "DEM Tool" button on the main Orthomax toolbar.
2) Create a new DEM for each pair of overlapping frames by selecting "New..." under the File menu of the DEM Tool window.
3) Make sure the source is a Block. Next, set the Image Block you created previously and select the two overlapping frames from which you wish to extract the Digital Elevation Model. Name the DEM by filling in the "New DEM Name:" field. Make sure that the reference frame is set correctly (as you did previously when creating a block. You do not need to set the reference frame if the field under the reference frame is already filled in.) Next, specify the ground spacing desired. Orthomax will suggest a DEM resolution based on the triangulation results and the definition of the photo. You can set this number to whatever you want and it usually does a pretty good job; however, I would not set it less than 3 meters unless you wish to wait for several days to get a DEM. Click "OK" at the bottom of the window.
4) Create the DEM by selecting "Collect..." under the Activities menu. Click "OK" at the bottom of the window and let the DEM collection run until completion.
5) Once the DEM collection is completed, you can view the results by selecting either "Raster.." or "Relief Shaded.." under the View menu.
6) The DEM is saved under your project directory in a directory called "dtm" under the filename that you specified for the DEM name. In addition, a "elev.img" will be added to the end of the name of the file.
** The edges of the DEM will be unreliable; however, you can splice the good sections of both the DEMs and the Orthophotos (see below) together by using the Imagine "Subset Image" command and specifying an AOI for the good area. Then you splice the DEMs and Orthophotos back together using the "Mosaic Images" command in Imagine.
Create an Orthophoto:
1) Start the Ortho Tool by clicking the "Ortho Tool" button on the main Orthomax toolbar.
2) Create a new orthophoto for each overlapping pair of frames by selecting "New" from the file menu in the Ortho Tool window.
3) The Define Ortho Image window will pop up.
4) Define the Orthophoto parameters similar to defining parameters for DEM. Click "OK".
5) Under the Activities menu, select the "Orthorectify" option. A dialogue box will appear asking you what you would like the density of the orthophoto calculations to be. Enter 1. Click "OK".
6) Your orthophoto will be stored under the project folder in the "ortho" directory with the name you assigned it in the New Orthophoto definition dialogue.