Near duplicate image/keyframe retrieval from large multimedia databases

• Main Author: Lim, Yun Fong.
• Other Authors: Hoi Chu Hong
• Format: Final Year Project
• Language: English
• Published: 2009
• Subjects: DRNTU::Engineering::Computer science and engineering::Information systems::Information storage and retrieval
• Online Access: http://hdl.handle.net/10356/18916
• Institution: Nanyang Technological University

This project relates to retrieving near duplicate images from a database of images on portal webpage. First of all, this report begins with introduction to the topic of this project. It elaborates the objectives and scope of the project to readers. Next, a review to this topic is elaborated together with some relevant details in Chapter 2 Literature Review. The project sparked off by using global features extraction on grid color moments, edge direction histogram, local binary pattern, GIST and gabor features for implementation. Following that, it touched on local features extraction particularly with Scale-invariant Feature Transform (SIFT) and finally combining both global and local features which is under Chapter 3 Design Implementation. The result was generated by comparing the nearest Euclidean distance of each feature of image database against given image (Chapter 4 Results and Discussion). From the results, the author would draw some conclusion (Chapter 6 Conclusion) from her research throughout the entire duration of project after sharing some problems faced during the duration of this project (Chapter 5 Difficulties). The project’s scope can be further explored and improved should any party is interested in detection of near duplicate images. Future research can be done to implement an application embedded into any image search engine to minimize browsing images. Efficiency of image search engine always focuses on how related an image to another and thus, the project can be led into this direction.