Image indexing and retrieval techniques and applications

• Main Author: Lam, Xiao Hui.
• Other Authors: Yap Kim Hui
• Format: Final Year Project
• Language: English
• Published: 2012
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Computer hardware, software and systems; DRNTU::Engineering::Electrical and electronic engineering::Electronic systems::Signal processing
• Online Access: http://hdl.handle.net/10356/47708
• Institution: Nanyang Technological University

The main goal of this project is to recognize and categorize landmarks captured in images, automatically with the help of computer systems. Computer systems are generally inept in performing this task, unless effective algorithms are implemented to do so. This project was tackled through a multi-facet approach. Images are collected and separated into two sets: training and testing. Both of the training and testing images will go through the same processes under the Bag of words (BoW) model. After that, the training images will have to be learnt and classified by the system (computer) while the testing or query images skip the learning stage and proceed to classification. The ultimate objective is doing the comparison between training and testing images. In the Content Analysis, the selected methods: Scale Invariant Feature Transform (SIFT), K-means clustering and Support Vector Machine (SVM) are discussed. Mobile devices with built in features: Global Positioning System (GPS) and direction are introduced in the Context information. Finally, the concept of the view cone is presented in integration. The algorithms will be evaluated with three tests and they are namely: Content Analysis, Context Analysis and Integration of both analyses. The results were found to vary in terms of processing time and recognition rate.