Math and Computer Science Honors Theses

A Computational Model of Heading & Object Detection Using Real World Scenes

Date of Creation

6-13-2018

Document Type

Departmental Honors Thesis - Restricted Access

Department

Mathematics

First Advisor

Constance Royden

Abstract

A seemingly simple task, observing the world visually while moving through it, presents several perceptual problems. How humans compute heading, or the direction in which they are moving, and detect moving objects are examples of such problems. To study these, biology, psychophysics, and computer science are combined to construct computational models— computer implementations of theories of how the human visual system performs these tasks. The Royden model of human heading computation and moving object detection attempts to directly model the receptive field layouts of primate middle-temporal area visual cells and proposes a motion-subtraction approach to heading calculation. In this project, we develop an interface that uses camera images to provide input to the Royden model to test its performance on real-world visual scenarios and the types of noise that may occur. We show that the model maintains a high level of accuracy even in the presence of noise and visual ambiguities introduced when computing motion from real scenes.

Comments

Reader: David B. Damiano

This document is currently not available here.

Share

COinS