Affine Invariant-Based Classification of Inliers and Outliers for Image Matching

12:30pm, April 21, Tuesday, 2009, Engineering Building Room 4201

Speaker

Dan Fleck
Department of Computer Science
GMU

Abstract

This presentation describes a new approach to classify tentative feature matches as inliers or outliers during wide baseline image matching. Wide-baseline matching is the process of matching one image to another. After typical feature matching algorithms are run and tentative matches are created, our approach is used to classify matches as inliers or outliers to a transformation model. The approach uses the affine invariant property that ratios of areas of shapes are constant under an affine transformation. Thus, by randomly sampling corresponding shapes in the image pair we can generate a histogram of ratios of areas. The matches that contribute to the maximum histogram value are then candidate inliers. The candidate inliers are then filtered to remove any with a frequency below the noise level in the histogram. The resulting set of inliers are used to generate a very accurate transformation model between the images. In our experiments we show similar accuracy to the standard RANSAC approach and an order of magnitude efficiency increase using this affine invariant-based approach.

Short Bio

Dan Fleck is currently an instructor of Computer Science at George Mason University (GMU). He earned his B.S. in Electrical Engineering from the University of Texas at Austin before moving to Northern Virginia. While working full-time Dan completed his M.S. in Software Engineering from GMU.

Currently he is pursuing a doctorate degree in Computer Science. Dan's doctoral research is in Computer Vision under Dr. Zoran Duric. Specifcially researching novel approaches to matching images taken at different viewing angles, locations and scales.

Previously, Dan was a technical lead and project manager at SRA International. At SRA he led projects ranging from 5 to 50 people for a variety of government clients. Dan worked within SRA's Health Systems group, Data Mining Center and most recently as technical lead within the Advanced Technology Group. Dan continues to serve in an advisory role at SRA.