• home
• current students
• graduate
• events

Presenter: Rizwan Pyar Ali Hemani
Supervisor: Dr. T. Aaron Gulliver

Date: Tue, May 27, 2014
Time: 10:00:00 - 11:00:00
Place: EOW 430

ABSTRACT

ABSTRACT:

Robust and fast feature extraction is crucial for many computer vision applications such as image matching. Speeded Up Robust Features (SURF), a novel scale and rotation invariant detector and descriptor, outperforms previously released schemes such as Scale-Invariant Feature Transform (SIFT) in terms of repeatability, distinctiveness and robustness. SURF is a better scheme than SIFT for image recognition based on its computation speed and accuracy. In this work, a new algorithm for logo recognition based on SURF and feature-based image matching algorithms such as hierarchical K-means clustering, the distance ratio test, and Random Sample Consensus (RANSAC) is proposed. This algorithm is assessed on a set of 100 company logos images. A series of experiments are conducted, splitting these 100 images into a training set and a testing set respectively. The experimental results show that the enhanced version of the SURF algorithm performs better than its counterpart. The proposed algorithm is noise-resilient, scale-invariant and rotation invariant up to a certain degree.