Automatic change detection

Project 1: Change detection from Multispectral remote sensing images

Why change detection from RS imagery is important?

With the increase availability of RS data, there is substantial interest in locating areas that may have changed in order to monitor the difference over time for:

• 1. Land-cover use analysis

  2. Urban expansion planing

  3. Deforestation

  4. Damage assessment

  5. Disaster monitoring

  6. etc.

Exampl 1- Change in ice cover detection - knee lake, Manitoba, Canada.

April 2015 - Source: Landsat-8 OLI

Example 2 - Agricultural change detection, Riyadh, KSA

Feb. 2006 Oct. 2006 (Source IKONOS) Change

Example 3: New construction detection, Mina, KSA

June 2007 December 2007 Difference image

Methods:

• • Classifiers combination

  • Bayes theory

  • Information fusion

Project 2: Change detection from different source images (ongoing)

Montreal 1966 (aerial photo) Montreal 2016 (Satellite photo)

Methods to investigate:

• • Deep learning (CNN, Auto-encoder)