Satellite Remote Sensing of Urban Areas
Fall 2013 Term Project (sample)
Detection of Urban Expansion using Multi-Temporal satellite Images
Objective:
You are provided by two (or more) Multi-Spectral images and you are requested to apply your knowledge on image registration, image enhancement, and image classification in detecting expansion and changes of urban areas.
Proposed Data:
? One Multi-Spectral Landsat ETM image (30 m resolution –Landsat_ETM_2000.img) to be used as a reference image for image registration process and in the image classification process.
? One Multi-Spectral unregistered Landsat TM image (30 m resolution landsat_tm1984_Unregistered.img) to be registered and used later in classification and change detection.
Methodology:
Part I. Image Registration:
In this part of the project, you will perform few experiments that will demonstrate some principles of image-to-image registration. The registration will involve the two images used in the project:
Use image-to-image registration technique to register the “landsat_tm1984_Unregistered.img” image to the reference image “Landsat_ETM_2000.img”. Digitize at least 20 points as GCPs and checkpoints. Use different orders of the polynomial transformation mathematical model to register the image. Examine the effect of using different transformation functions and different numbers of GCPs on the accuracy of the registration process.
Deliverables:
Prepare a report including the following information and analysis:
1. Print out the distribution of the GCPs and the checkpoints on the image. Provide a table of the coordinates of the GCPs/checkpoints on the input and the reference image.
2. Describe the image registration process and its procedure.
3. Use different numbers of GCPs and different polynomial orders in the image registration process. List for each trial the residuals, the min/max error, and the total RMS errors of the checkpoints in both X and Y…