Mengjie Zhang, Satellite Image Recognition

Satellite Image Recognition using Intelligent Methods

• Overview of the Satellite Image Recognition Problem:

  Pattern recognition in satellite images is an interesting tasks and has many applications, such as weather forecast, land planning, and military uses. Typically, this problem involves classifying different portions of the images into a small number of categories.

  This image shows an example of this problem. It is an image of the south island of New Zealand with surrounding area/sea. The task is to find all of the cloud, snow, land, and water areas.

• Tasks and Goals of this Project

  The goal of this project is to investigate intelligent methods such as neural networks, decision trees and genetic programming for satellite image recognition problem. Specifically, this project will focus on:

  • What features can be used as inputs to decision trees, neural networks and genetic programming?
  • What transfer functions can be used in neural networks?
  • What functions and fitness measures can be used in genetic programming?
  • Can these methods do a good enough job for this problem?
  • Which of these methods outperform the traditional statistical method?

  The methods developed will be examined on satellite images.

• More information can be seen from Meng's research projects . Related Publications can be viewed from here: Meng's recent publications. Contact me if you want to get a copy of those papers, or if you want to know more detail about this project.