University of Houston
Department of Computer Science
In partial fulfillment of the Requirements for the Degree of
Master of Science


Manav Goel
will defend his thesis

Segmentation of Images using Machine Learning and Real Time Systems Feedback

Abstract

This thesis reports the results of Segmentation of Images using Machine Learning and Real Time Systems. Overall, in this thesis, we proposed an approach to classify images. The approach is based on the technique whereby we use pixel level classification to achieve object level clustering. This process is image dependent and hence is very specific for the type of objects inside an image. However as we use supervised classification we can apply this technique to any type of image, thus making the system universal in its approach.

The technique is usual in the way that it uses a user feedback system that can customize the output as per the user preference. This thesis talks about an approach whereby image classification is done using six different segmentation routines that help extract six different features of a pixel. This helps in the universal approach as some kind of images might respond well to some segmentation routine whereas another type of iamge might interact better under some different segmentation routine. The results show that our approach works well under different domains of images and under different user preferences the results vary in accuracy


Date: Monday, November 28th, 2005
Time: 3:30 PM
Place: 550-PGH

Faculty, students, and the general public are invited.
Thesis Advisor: Dr. Ricardo Vilalta