1. CE497 Urban Remote Sensing, Jie Shan1 Geometric rectification Homework 5

2. CE497 Urban Remote Sensing, Jie Shan2 Description Objective: geometrically rectify a portion (2km*2km) of SPOT image over West Lafayette area Reference (GCP): to be measured from the USGS digital ortho quad (DOQ). DOQ is used as the reference image Methodology: see the example Requirements: at least 10 GCP and 6 check points need to be measured 10 meter spatial resolution for the rectified image Use and compare affine, polynomial (2 nd order), projective transform (2 nd order) methods Data: ftp://ftp.ecn.purdue.edu/jshan/ce497URS/ Hand in: a report including (max. 3 pages, 12 font, 1.5 line spacing) A brief description of the principles of the used methods Compare the results from different methods The rectified image with coordinate grid and scale info in Letter format Show and discuss your selected GCP and their residuals Show and discuss your check points and their differences Discuss and summarize your findings

3. CE497 Urban Remote Sensing, Jie Shan3 Where to get help ? – pdf’s FieldGuide: photogrammetry principles TourGuide: step-by-step tutorial by modules OrthBASE: orientation, matching, triangulation, DEM generation StereoAnalyst: feature collection, stereo map compilation

4. CE497 Urban Remote Sensing, Jie Shan4 For the “geometric rect.” project, Read: FieldGuide: Vol. 1 One-line manual : TourGuide: step by step

5. CE497 Urban Remote Sensing, Jie Shan5 A worked example – tutorial Landsat image is to be rectified based on SPOT image, Atlanta

6. 6 Choose one desired band to display Image to be rectified – Landsat TM To display only one band, use Raster > Band Combinations dialogue box as below CE497 Urban Remote Sensing, Jie Shan

7. 7 Reference image – SPOT Pan Load the SPOT Pan image in a separate viewer. This image will be used as a reference image, i.e., any location on this image can be used as (2-D) ground control. D:\IMAGINE\examples\ panAtlanta.img

8. CE497 Urban Remote Sensing, Jie Shan8 Steps for geometric rectification In the first view (the image to be rectified), run Raster > Geometric Correction command Choose Polynomial in the Select Geometric Model dialogue box, select the polynomial order then click OK. Click Close for now as below. (will make selection later)

9. CE497 Urban Remote Sensing, Jie Shan9 Steps for geometric rectification Choose OK in the right window (Existing Viewer is checked) At the following prompt, Click the second viewer (the one to be used as reference). This step let you choose the reference image.

10. CE497 Urban Remote Sensing, Jie Shan10 Steps for geometric rectification Measure GCP. Note measure the 1 st image (image to be rectified) first, and then the 2 nd image (the reference image) 1 st and 2 nd polynomials can be tested. Rarely go beyond 2 nd order. However, you can test what happens if so is chosen. Measure a number of check points to evaluate the quality of rectification (set the point type from the “GCP Tool” menu: edit > set point type > check) Analyze the error distribution and decide if more GCPs are needed and where. Produce the image map using the resample tool. Check the on-line TourGuide for details.

11. CE497 Urban Remote Sensing, Jie Shan11 Options for Projective Transform Model At least 9 GCPs are required to solve the 2D model (using the same denomination for x and y – default option-) Choose “Projective Transform” in the “Set Geometric Model” dialogue box, select the RPC order, check “2D Dimension Projective Model” option and make sure that the “Use same denominator for x and y” option is checked. Click Close