Lab 2: Digital Photogrammetry: Image Registration & DEM Extraction

 

Introduction

 

This lab has three separate tasks that will let you do "conventional" image registration, orthorectification, and DEM extraction using Exelis' ENVI software. You will learn various ways of correcting the distortions on remotely sensed data. In addition, you will learn how to extract DEM from remotely sensed data. In Task 1, you will perform image registration/georeferencing. Task 2 involves the use of a DEM to orthorectify a high-resolution satellite image. In Task 3, you will extract DEM from a pair of stereo images (i.e., a stereo-pair).

 

You will use ENVI and ENVI Classic for the lab. You can find the program from the Windows Start menu (e.g., under All Programs -> ENVI 5.4 -> 64-bit -> ENVI 5.4 (64-bit) for ENVI or All Programs -> ENVI 5.4 -> Tools -> ENVI Classic 5.4 (32-bit) (or 64-bit) for ENVI Classic. This is your opportunity to explore the ENVI software and to get acquainted with the interface. Learning new software takes extra time. However, learning how to use software is a required skill if you plan to work in the GIS industry - given that new software packages (or new versions of existing software) are released frequently. Use the software help system and Exelis' Documentation Center to learn about the different software options and functionality. Type the answers to all questions, attach the requested maps (or images), and submit them to the instructor by the due date.

 

The lab files used in Lab 2 are in the I:\Students\Instructors\Geoffrey_Duh\GEOG4593\Lab2 folder. Please copy the lab data to your lab2 folder in your local workspace. The total file size for the lab data is around 300MB. Make sure the target disk has sufficient space. Spend a few seconds browsing the lab data folders so that you know what are in the folders. The lab documents you need for this lab are in I:\Students\Instructors\Geoffrey_Duh\GEOG4593\Lab2\PDFs. For Task 1, you will use Georeferencing_Registration.pdf (ENVI Classic) and ImageRegTutorial.pdf (ENVI). You will use an online tutorial document for Task 2 (see Task 2 instructions for the link) and DEM_Extraction.pdf (ENVI Classic) for Task 3.

  

Task 1. Image Registration

 

Instructions & Questions

Read the introduction and background of the ENVI Classic Tutorial: Image Georeferencing and Registration (Georeferencing_Registration.pdf). Complete the Image to Image Registration exercises (pages 4 - 9). Prepare the screenshots asked in Question 1 and answer Questions 1 and 2.

 

Continue to complete the Image Registration Tutorial exercises in the ImageRegTutorial.pdf document. Please note that you will use ENVI (instead of ENVI Classic) for this part of the lab. Answer Questions 3 and 4.

 

  1. Write a short paragraph and use screenshot images of a small (blow-up) area of bldr_tm1.wrp, bldr_tm2.wrp, and bldr_tm3.wrp to illustrate the visual difference between the nearest neighbor, bilinear interpolation, and cubic convolution image resampling methods. Make sure you label the images clearly so that the readers know which resampling method was used to create which image.

 

  1. From the Ground Control Points Selection dialog, change the Degree value and see how the value affects RMS Error. Record the RMS Errors for 1st and 2nd degree polynomial method. Warp the bldr_tm.img file with a 2nd degree polynomial method and cubic convolution resampling method. Name the output file bldr_tm4_2.wrp. Use ENVI's dynamic overlay and geographic link functions to visually compare the alignments of the images warped with 1st and 2nd degree polynomial methods (bldr_tm4 and bldr_tm4_2). Write a short paragraph describing which polynomial degree (1st or 2nd degree) is more effective in the registration of the Boulder TM image.

 

  1. In both ENVI Classic and ENVI's registration exercises, you used different warping methods when warping images. These methods are RST, polynomial, and triangulation. Use ENVI's online help to search for these keywords and write a short paragraph describing what these methods are.

 

  1. The built-up area on the lower left corner of the overlapped Ikonos and Qquickbird images is the campus of the National Center for Atmospheric Research (NCAR). Use Google Earth to explore the geographic characteristics of this area. Given the geographic characteristics of this area and your understanding of the limitations of the warping methods, please speculate why some of the warping methods did not work effectively on the registration of the lower left corner of the images.

 

 

Task 2. RPC Orthorectification

 

Instructions & Questions

Complete the ENVI RPC Orthorectification Tutorial (click on the hyperlink to open the document) and export the orthorectified output file in TIFF format, name it "orthoimage.TIF". Then, repeat the same process using the same source TIF and the DEM. This time do not load any GCPs. Export the orthorectified output in TIFF format, name it "orthoimage_RPCOnly.TIF". Open the two orthorectified TIFF files in ArcMap. Add the DEM and the source TIF file (i.e., the unorthorectified image) to ArcMap.

 

1.    Use ArcMap's layer properties dialog windows of these layers to complete the table below.

 

Image

Spatial Reference

Cell size (unit)

Unrectified Source TIF

 

 

Orthorectified TIFF files

 

 

DEM

 

 

 

2.    Add the mroads layer from I:\Students\Data\GIS\ESRI_Data10\streetmap_na\data\mroads.sdc and use it as a reference for assessing the positional accuracy of the two orthorectified images, one rectified with RPC only and the other with GCPs. Use ArcMap's measure tool to check the displacements of some major road intersections on both images. Prepare two blowup screenshot images, one for each rectified image, and write a short paragraph describing how these images were created and why one has a better positional accuracy than the other. Use ENVI's online help to learn about RPC and GCP.

 

  

Task 3. DEM Extraction from Stereo Image Pair

 

Instructions & Questions

Complete the ENVI Classic Tutorial: The DEM Extraction Module (DEM_Extraction.pdf). The tutorial was developed for ENVI Classic. Please use ENVI Classic for the exercises. Alternatively, you can use ENVI's "DEM Extraction Wizard: New" workflow to perform the tasks. However, there are some minor differences in the graphic user interface (GUI). The DEM Extraction Wizard can be found in ENVI's Terrain/DEM Extraction toolbox.

 

 Questions:

 

1.      Review the tie points generated by the Wizard and make necessary changes of the tie points to lower the Maximum Y Parallax value. Insert a screenshot of the "DEM Extraction Wizard: Step 5 of 9" (see an example on page 8 of the tutorial) showing your Maximum Y Parallax value.

 

2.      Export the DEM and the georeferenced ASTER rasters that you created in this task to ArcGIS compatible raster format (e.g., ERDAS Imagine). To save the ENVI rasters to ERDAS Imagine .img file format, select "Save File As.../ERDAS Imagine" from "File" on the ENVI Classic's main menu. After the .img files are generated, open these rasters in ArcScene and create a 3D perspective view of the ASTER image (using the DEM as the base height). You will need to adjust the symbology of the ASTER image so that it appears as CIR (i.e., false color infrared composite). Insert a screenshot of the 3D ASTER image and write a short paragraph describing how the DEM used in the image was generated.

 

3.      Where is the area depicted on the ASTER image?