Near duplicate image identification

• Main Author: Sadykov, Zhomart.
• Other Authors: Xu Dong
• Format: Final Year Project
• Language: English
• Published: 2012
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Electronic systems::Signal processing; DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
• Online Access: http://hdl.handle.net/10356/48837
• Institution: Nanyang Technological University

This paper suggests solution for Near Duplicate Image Identification problem. The term “Near Duplicate” refers to an images which are perceptually identical, in which the differences comprise of noise, editing operations like scaling or rotation, also it could be scene illumination changes or minor viewpoint 3D rotations. Proposed NDII algorithm uses SIFT-features introduced by David Lowe (2004) as a basic unit on which all operations are performed. This paper describes algorithm that provides capabilities of pre-building image database which can be efficiently loaded and queried for “Near Duplicate” images. Exhaustive performance testing and analysis of the NDII algorithm is done after development, allowing to state efficiency of the proposed solution. Possible application of the NDII algorithm to its sub-problem “Object detection/tracking” demonstrated as well. Lastly, pros and cons of the algorithm are summarized.