Please use this identifier to cite or link to this item:
|Title:||Artwork visual recognition||Authors:||Ho, Si Hao||Keywords:||DRNTU::Engineering::Electrical and electronic engineering::Electronic systems::Signal processing||Issue Date:||2014||Abstract:||In the recent years, advancement in mobile technology has produced an evolutionary device: known as the Smartphone. It had gained high popularity among all ages. This led to an increase in consumerism and major manufacturers have been aggressively releasing new models annually. A smartphone allows the user to retrieve information anywhere as long it is connected to a 3G/4G network. This encourages applications developer to create web based apps for users to retrieve information immediately from internet or cloud based server instantly. This project aims to help mobile users to retrieve information of famous artwork through identifying the images captured on their smartphone. Development and research in image recognition software has been ongoing while there are none in the context of oil painting. In this project, we will study and discuss the main visual recognition technique and methods. We will also evaluate and compare how each method will contribute to the overall accuracy of the system. As the majority of the oil painting images will contain undesirable features such as the wooden frame or human occlusions. These features do not “faithfully” represent the image and the extracted vectors will introduce “noise” into the feature matching process. This will cause the matching accuracy to drop drastically. Hence, we will use Geometric Verification (GV) to reduce the impact of these issues. The GV method is quite effective and it’s able to increase matching accuracy in the range of 5-10%. While the downside is, it required longer processing time to perform feature matching. We will then evaluate the results of the experiments conducted and recommend a suitable method for the characteristics of project. Lastly, we will explore on ways to improve the matching efficiency and optimizing the image database for the future developments.||URI:||http://hdl.handle.net/10356/61455||Rights:||Nanyang Technological University||Fulltext Permission:||restricted||Fulltext Availability:||With Fulltext|
|Appears in Collections:||EEE Student Reports (FYP/IA/PA/PI)|
Page view(s) 50111
checked on Oct 26, 2020
checked on Oct 26, 2020
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.