Corner Detction.

Date of Submission

December 2010

Date of Award

Winter 12-12-2011

Institute Name (Publisher)

Indian Statistical Institute

Document Type

Master's Dissertation

Degree Name

Master of Technology

Subject Name

Computer Science


Machine Intelligence Unit (MIU-Kolkata)


Kundu, Malay Kumar (MIU-Kolkata; ISI)

Abstract (Summary of the Work)

A comer is defined as the junction point of two or more straight line edges. Corners are special features in a image. They are of great use in computing the optical flow and structure from motion. The earliest corner detection methods involved first segmenting the image into regions and representing the object boundary as a chain code. Corners were identified where the direction changed rapidly. Later attempts were directed at coming up with a comer detector which operated directly on gray level images. These include the one developed by Zuniga and Haralick (SI, Kitchen and Rosenfeld and the one by Dreschler and Nagel. In these approaches cormers are considered as the points where the rate of change of gradient direction is maximum.Harris-Stephens is shown as the most successful detector. Additionally, another commonly cited comer detector, SUSAN, is compared with other detectors. Image gradient based comer detectors are very popular despite the fact that they are not robust to changes in corner orientation, corner angle and/or contrast. Moravec developed an interest point detector based on the auto correlation function, He measured the differences between the feature window and its shifted versions. The main idea is that for a defined neighborhood around a comer point, movement in any direction should yield a considerable intensity change. To determine the corner points, Forstner, in 1986, utilized the so called auto-correlation matrix which is in essence the outer product of image derivatives along x and y axes. Different functions of the trace and determinant of the same matrix are used by Harris, Stephens and Noble in 1988 and 1989, respectively.In this work, we describe a Neural Network based comer detection technique that will use techniques changes in corner orientation, angle and contrast.


ProQuest Collection ID:

Control Number


Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.


This document is currently not available here.