Image rectification in Erdas.
We’ll start by grabbing a historic aerial photo from Geography’s website. Pick any you like, but make sure there are a bunch of features that existed both then and now (roads, etc).
Head here Pick and download any (original size, uncompressed) photo. Of course, make it one around here so that it’s in the same area as your landsat.
Fire up ERDAS. Open the airphoto in one view window; in a second, open up one of the earlier landsat images. I’ve chosen to use the 15m panchromatic image from 14 July, 2014
As always, make sure everything is on your flash drive (or D).
Zoom in on the landsat so it’s about the same spot as the photo.
Notice how when you mouse over the airphoto, it gives pixel location values (row/column), while it gives UTM for the landsat image.
Dang.. things have changed since 1954.
OK. Make the photo the active layer. On to the panchromatic tab, then control points. Select Polynomial as the geometric model.
We’ll be collecting reference points from an image layer. Looks like there are a bunch of cool options, including Bing imagery. But… we want our data to overlay as best as possible, so registering one image to the other is the best option.
Choose a 3rd order polynomial. Hit apply then close.
We’ve now got a split screen window to work from.
Zoom in on a spot you can locate in both images. Make sure GCP1 is highlighted at the bottom. Highlight the photo Click on the circle with a + inside it at the top. Click on the photo. Go back and click on the button, then on the pan image. You should now see, at the bottom, for the first GCP, x and y input and x and y reference.
Repeat for, say, 16 GCPs. Be sure to spread them out as much as you can.
After 10 points, you will start seeing the residuals and RMS error starting to pop up. Carry on.
Now. Looks like the residuals are given in the units of the photo – which isinches (check the metadata for the airphoto if unsure). You can also see which points contribute the most to the overall RMS error – and whether x or y. Go back and edit the worst points. Clean that puppy up. If you have to, delete points and digitize new ones.
Once you’re happy, time to actually do the rectification.
But first, grab me a screen shot or two showing your GCPs, images, etc. Provide me enough information that I can see what you’re doing in both images and the GCPs. Paste these images into Word.
This process is resampling: Resampling is the process of calculating the file values for the rectified image
and creating the new file. All of the raster data layers in the source file are resampled. The
output image has as many layers as the input image. ERDAS IMAGINE provides these
widely-known resampling algorithms: Nearest Neighbor, Bilinear Interpolation, and Cubic
Convolution Resampling requires an input file and a transformation matrix by which to
create the new pixel grid.
Click on the 4 little colored boxes next to the ruler at the top. This brings up the resample dialog box.
Give it an output file name. Select nearest neighbor. And OK.
Now, go back to the main erdas window and let’s see how things look.
Close the viewer with just the airphoto in it.
Add the new, referenced photo to the view with the landsat image in it.
Visually decide how good it is! Also, bring in the 30m (color) landsat image and look at that.
Give me a paragraph or two, and a screenshot or two, telling me how well this worked. And why. Did the image match up on all 4 sides? About how far off are you in meters? And if it wasn’t consistently off (in the same direction), tell me all about it!