Lab 7 : Photogrammetry

Lab 7 : Photogrammetry

Introduction

Photogrammetry is defined as the science of making measurements from photographs.  In this lab, we gained experience of photogrammetric tasks using aerial photography in ERDAS Imagine. The tasks completed in this lab included calculating scale of aerial photographs, calculating relief displacement of aerial photographs, and orthorectification.  The tasks were designed to introduce us to diverse photogrammetric tasks to obtain experience for applications in the future.  

Methods 

Part 1 : Scales, Measurements, and Relief Displacement 

Section 1 : Calculating Scale of Nearly Vertical Aerial Photographs 

We were given an aerial photograph of a portion of interstate 94 in Eau Claire County, WI (Figure 1).  Two points shown on the photograph with a real distance between them of 8822.47 feet.  I measured the distance on the photo and got a result of 2.7 inches.  The real distance was 8822.47 ft. I first converted 8822.47ft to 105,869.64in.  So 105,869.64 in real life is 2.7 inches on the given image.  I then took 105,869in and divided by 2.7in which gave me a value of 39210.977.  Therefore the scale is 1 : 39,210.997 or 1 : 40,000

Figure 1

We were then given an another similar image (Figure 2) and asked to calculate scale, this time using altitude, elevation,  and the focal length lens.    The values given were the altitude (20,000ft), focal length lens (152mm), and elevation of Eau Claire County (796ft).  The formula to calculate the scale is:  Photo length/(Altitude – Elevation). I first converted everything into inches.  152mm = 5.98 in, 20,000ft = 240,000in, and 796ft = 9552in.  The formula was then 5.98in / (240,000in – 9552in).  The following equation ends up resulting in the Scale = 1: 38,536.45 or 1: 40,000

(Figure 2)

Section 2 : Measurement of Areas of Features on Aerial Photographs 

This section of part 1 focused on using the measure and perimeter tools to obtain the real world measurements of the perimeter and area of a lagoon. The measurement and perimeter tools used were very similar to the process of digitizing.  I carefully traced the lagoon with the drawing tools to receive a result for the perimeter and area for the real world measurement in ERDAS Imagine.  The results were as follows: 

The area is 37.74 hectares = 93.26 acres 

The perimeter is 4131.17 meters = 2.57 miles 

Section 3 : Calculating Relief Displacement from Object Height 

We were given a portion of the previous images zoomed (Figure 3) in to calculate the relief displacement using the height of the smoke stack on upper campus.  In order to calculate the relief displacement, I needed the height of the object, the distance from the principal point to the top of the object, and the height the photo was taken. Height of the object turned out to be 1,604.5 inches, the distance from the principal point to the top of the smoke stack was 10.5 inches, and the height the photo was taken was 47,760 inches.  I then plugged these values into the formula:   (1,604.5 * 10.5)/(47,760) = 0.3527 relief displacement.  0.3527 is a positive number, so you have to move the smoke stack 0.3527 inches towards the principal point to make it vertical.  

Figure 3

Part 2 : Stereoscopy 

This portion of the lab focused on using ground control points (GCPs) to create an image that shows a 3-dimensional perspective of elevation using polaroid glasses.  To accomplish this, I used the Anaglyph tool under 'Terrain'.  The Anaglyph Generation window opened up and I set the Input DEM as ec_dem2.img and the Input image as ec_city.img.  I increased the vertical exaggeration to 2 and ran the model.  This created an output anaglyph image that appeared 3-dimensional through polaroid glasses. 

Part 3 : Orthorectification 

The final portion of this lab introduced us to orthorectification, which is the process of producing a planimetrically correct image by removing tilt and terrain effects.  This process is quite time consuming but very accurate.  ERDAS Imagine Lecia Photogrammetric Suite was used for orthorectification.  

Section 1 : Creating a New Project 

The first step in orthorectification was to create a noew project focusing on images of Palm Springs, California.  I first created a personal sub folder, then opened the LPS Project Manager through the Toolbox.  I then created a new block file and proceeded to set up my model.  The geometric model was set as a Plynomial-based Pushbroom.  I had to set a horizontal reference source, which I set the projection type to UTM, the spheroid name to Clarke 1866, the Datum Name to NAD27(CONUS), the UTM zone as 11, and finally set the 'North or South' parameter to North.  

Section 2 : Add Imagery to the Block and Define Sensor Model 

In the LPS Project Manager window, I clicked image and then Add Frame.  I added the Spot_pan.img and hit okay.  I reviewed the fields and parameters in the SPOT Pushbroom Frame Editor and the sensor information window, in turn specifying the sensor, turning the 'Int.' column green for Spot_pan.img in the LPS Project Manager window.  The sensor is now defined. 

Section 3 : Activating Point Measurement Tool and Collecting Ground Control Points (GCP's)

The first step in this portion of part 3 was to click the 'Start Point Measurement' tool in the LPS Project Manager window.   I selected Classic Point Measurement and the point measurement window popped up.  I reset the horizontal reference source to the file xs_ortho.img.  I then began adding GCP's at various road intersections.  After the first two GCP's were added, the Automatic Drive function was activated, which helps with the rapid collection of GCP's. The Automatic Drive Function approximates the position of a GCP in the block image file using the reference GCP. I then collected 9 more GCPs. However, the last two points I collected were done using a different horizontal reference source, this time NAPP_2m-ortho.img. The next step is to collect elevation information from a vertical reference source.  The vertical reference source used was palm_springs_dem.img.  I clicked 'Update Z Values on Selected Points' to obtain the z values.  

Section 4 : Set Type and Usage, Add 2nd Image to Block, Collecting those GCPs

 The first step in this section was to set the Type column to 'Full' and set the Usage column to 'Control' in the cell array of reference GCPs.  After updating the columns, I have now finished the collection of reference coordinates for the first block file, spot_pan.imag.  The second block file was spot_panb. I used the Add Frame button to add this file and specified its sensor. I now had to collect GCPs for spot_panb.img based on the ones already collected in spot_pan. 

Section 5 : Automatic Tie Point Collection, Triangulation, and Ortho Resample

       

The process in this section involved measuring the image coordinate positions of round points in the overlapping area of the two SPOT images (spot_pan.img and spot_panb.img).  In the automatic tie point generation properties, I made sure that the parameters were correct and ran the model. The automatic tie point generation process completes and I was presented with an Auto Tie Summary, which can be used to check for the tie point accuracy.  Now that I have obtained all the control and tie points, I now had to conduct triangulation.  I opened the triangulation dialog and set the parameters to the desired option.  Iterations with relaxation value was set to 3, Image coordinate units for report was set to pixels, Ground Point Type and Standard Deviations type was set at Same Weighted Values, all of the coordinate values (x, y, z) were set to 15, and I verified that the Simple Gross Error Check Using box was checked. I ran the model and a triangulation summary appeared.  I then had to use the tool Start Ortho Resampling Process.  In the Ortho Resampling dialog, I selected the palm_springs_dem.img for the DEM, set the output cell sizes to 10, and set the output image name and location.  In the advanced tab, I set the resampling method to Bilinear Interpolation, and added the two block files(spot_pan.img and spot_panb.img) to the image inputs. I then ran the model and the Ortho Resampling Process completed. 

Section 6 : Viewing the Orthorectified Images

I brought both outputs (orthospot_pan.img and orthospot_panb.img) into a single viewer and it displayed both Orthorectified images (Figure 4).     The overlap seems to be extremely accurate down to the pixel.  You can also see the z values with vertical overlaps over the mountains and other tough terrain. 

      

Results 

Figure 4

Figure 4 shows our final Orthorectified Images.  Note the spatial accuracy of the overlap. The overlap seems to be extremely accurate down to the pixel.  You can also see the z values (elevation) adjustments with the overlaps over the mountains and other rough terrain. 

Sources 


 National Agriculture Imagery Program (NAIP) images are from United States Department of Agriculture, 2005.

Digital Elevation Model (DEM) for Eau Claire, WI is from United States Department of Agriculture Natural Resources Conservation Service, 2010.

Spot satellite images are from Erdas Imagine, 2009.

Digital elevation model (DEM) for Palm Spring, CA is from Erdas Imagine, 2009.

National Aerial Photography Program (NAPP) 2 meter images are from Erdas Imagine, 2009.